Evidence Paper & Study Protocols
Transcript of Evidence Paper & Study Protocols
EU JANPA WP4 Deliverable 4.1
Evidence Paper & Study Protocols
Deliverable 4.1
Work package WP 4: Evidence (the economic rationale for action on childhood obesity) Responsible Partner: IPH IRL
Contributing partners: HZJZ, HZZO, ATEITH, AHEPA, UCC-CHDR, ISS, MS, IOMC, NIJZ
JANPA – Joint Action on Nutrition and Physical Activity (Grant agreement n° 677063) has received funding from the European Union’s Health Programme (2014-202
EU JANPA WP4 Deliverable 4.1
The content of this Deliverable represents the views of the author only and is his/her sole
responsibility; it cannot be considered to reflect the views of the European Commission
and/or the Consumers, Health, Agriculture and Food Executive Agency or any other body of
the European Union.
The European Commission and the Agency do not accept any responsibility for use that may
be made of the information it contains.
EU JANPA WP4 Deliverable 4.1
GENERAL INFORMATION Joint Action full title Joint Action on Nutrition and Physical Activity
Joint Action acronym JANPA
Funding This Joint Action has received funding from the European Union’s Health Programme (2014-2020)
Grant Agreement Grant agreement n°677063
Starting Date 01/09/2015
Duration 27 Months
DOCUMENT MANAGEMENT Deliverable D4.1 “Evidence Paper & Study Protocols”
WP and Task WP4 and Task 4.2
Leader IPH IRL
Other contributors HZJZ, HZZO, ATEITH, AHEPA, UCC-CHDR, ISS, MS, IOMC, NIJZ
Due month of the deliverable May 2016
Actual submission month October 2016
Type
R: Document, report DEC: Websites, patent fillings, videos, etc. OTHER
R
Dissemination level PU: Public
PU
CONTRIBUTORS
Lead Team IPH IRL
Kevin Balanda, Director of Research and Information, Institute of Public Health in Ireland (Lead)
Jude Cosgrove, Senior Statistical Researcher, Institute of Public Health in Ireland
Lorraine Fahy, Information Analyst, Institute of Public Health in Ireland
Lindi Gatchell, Projects Officer, Institute of Public Health in Ireland
Suzanne Kirk, PA to Director of Research and Information, Institute of Public Health in Ireland
Sinéad Ward, Finance Officer, Institute of Public Health in Ireland
UK Health Forum (sub-contractor)
Laura Webber, UK Health Forum, London
Abbygail Jaccard, UK Health Forum, London
André Knuchel-Takano, UK Health Forum, London
Laura Pimpin, UK Health Forum, London
National Teams CROATIA (HZZO & HZJZ)
Zlatko Boni, Croatian Health Insurance Fund (HZZO)
Sanja Music, Croatian Institute of Public Health (HZJZ)
Jasmina Pavlic, Croatian Institute of Public Health (HZJZ)
GREECE (ATEITH & EPHEA)
Maria Hassapidou, Department of Nutrition and Dietetics, Alexander Technological Educational Institute of Thessaloniki (ATEITH)
Petros Katsimardos, Department of Nutrition and Dietetics, Alexander Technological Educational Institute of Thessaloniki (ATEITH)
Apostolos I Hatzitolios, Department of Internal Medicine, University Hospital Ahepa, Thessaloniki (AHEPA)
Konstantinos Bouas, Department of Internal Medicine, University Hospital Ahepa, Thessaloniki (AHEPA)
Nikolaos Kakaletsis, Department of Internal Medicine, University Hospital Ahepa, Thessaloniki (AHEPA)
5
IRELAND (UCCC-CHDR + IPH IRL)
UCC CHDR
Ivan J Perry, Department of Epidemiology & Public Health, UCC (Lead)
Fiona Geaney, Department of Epidemiology & Public Health, UCC
Laura Carter, Department of Economics, NUI Galway
Anne Dee, HSE Department of Public Health, Limerick
Edel Doherty, Department of Economics, NUI Galway
Douglas Hamilton, HSE Department of Public Health, Limerick
Laura McCarthy, Department of Epidemiology and Public Health, UCC
Grace O’Malley, Temple Street Children’s University Hospital, Dublin
Maura O’Sullivan, Department of Epidemiology and Public Health, UCC
Michelle Queally, Department of Economics, NUI Galway
IPH IRL
Kevin Balanda, Director of Research and Information, Institute of Public Health in Ireland
Jude Cosgrove, Senior Statistical Researcher, Institute of Public Health in Ireland
Lorraine Fahy, Information Analyst, Institute of Public Health in Ireland
ITALY (ISS)
Angela Spinelli, Instituto Superiore di Sanita, Roma (ISS)
Chiara Cattaneo, Instituto Superiore di Sanita, Roma (ISS)
Barbara De Mei, Instituto Superiore di Sanita, Roma (ISS)
Paola Nardone, Instituto Superiore di Sanita, Roma (ISS)
Laura Lauria, Instituto Superiore di Sanita, Roma (ISS)
PORTUGAL (MS)
Gisele Câmara, New University of Lisbon
Pedro Graça, Directorate-General of Health, Lisbon (MS)
Filipa Pereira, Directorate-General of Health, Lisbon (MS)
Miguel Telo de Arriaga, Directorate-General of Health, Lisbon (MS)
Andreia Jorge Silva, Directorate-General of Health, Lisbon (MS)
ROMANIA (IMCP)
Michaela Iuliana Nanu, Institute for Mother and Child Protection, Bucharest (IMCP)
Ioana Nanu, Institute for Mother and Child Protection, Bucharest (IMCP)
SLOVENIA (NIJZ)
Mojca Gabrijelcic Blenkus, National Institute of Public Health, Ljubljana (NIJZ)
Aleš Korošec, National Institute of Public Health, Ljubljana (NIJZ)
With thanks to Gregor Starc, Faculty of Sport, University of Ljubljana, for providing SLOfit data
Collaborating Partners
Ursula O’Dwyer, Department of Health, Ireland
Cliodha Foly-Nolan, safefood, Ireland
EU Joint Research Centre (EU JRC)
WHO(Europe)
7
GOVERNANCE
Expert International Scientific Advisory Committee (ISAC)
An expert International Scientific Advisory Committee (ISAC) guides the scientific aspects of JANPA
WP4.
The ISAC:
Gives scientific advice to WP4 Lead Team.
Reviews background materials and draft reports
Attends three face-to-face meetings in Ireland
Participates in one or two telecalls (if needed)
Members of the ISAC are:
Associate Prof Jennifer Baker, Institute of Preventive Medicine in Denmark and the
University of Copenhagen. Denmark
Dr Margherita Caroli, Nutrition Unit, Department of Prevention, Azienda Sanitaria Locale
Brindisi. Italy
Dr Anne Dee, Health Service Executive. Ireland
Dr Tony Fitzgerald, Department of Statistics and & Department of Epidemiology & Public
Health. University College Cork. Ireland
Prof David Madden, School Of Economics, University College Dublin. Ireland
Dr Martin O’Flaherty, University of Liverpool. England
Dr Pepijn Vemer, Department of Pharmacoepidemiology & Pharmacoeconomy, University of
Groningen. Netherlands
Prof Kevin Balanda, Institute of Public Health in Ireland. Ireland
Study principles
The seven principles that underpin the design, implementation and reporting of JANPA WP4 are
outlined in the Table below.
Table. Principles underpinning JANPA WP4
1. Relevance to JANPA WP4 countries and EU
2. Societal economic perspective that, in addition to health impacts and healthcare costs, includes important aspects of public health and impacts and costs experienced by society and its communities
3. Transparency that explains strengths but recognises assumptions limitations
4. Capacity building in JANPA WP4 countries and EU (research and information)
5. Identifying gaps in research and information
6. Stimulating further developments in research and information
7. Health equity
9
CONTENTS
CONTRIBUTORS ............................................................................................. 4
GOVERNANCE ................................................................................................ 7
Expert International Scientific Advisory Committee (ISAC) ................................................................ 7
Study principles ................................................................................................................................... 7
CONTENTS ..................................................................................................... 9
ACRONYMS AND ABBREVIATIONS ............................................................... 15
CONVENTIONS AND DEFINITIONS ................................................................ 17
SUMMARY: EVIDENCE.................................................................................. 22
Background ....................................................................................................................................... 22
Prevalence of child overweight and obesity ..................................................................................... 22
Child impacts of childhood overweight and obesity ......................................................................... 23
Adult impacts of childhood overweight and obesity ........................................................................ 27
SUMMARY: STUDY PROTOCOLS ................................................................... 31
Existing studies of lfetime cost of childhood overweight and obesity ............................................. 31
JANPA WP4 aims and objectives ....................................................................................................... 33
General issues ................................................................................................................................... 33
Model metrics ................................................................................................................................... 33
Diseases and other impacts .............................................................................................................. 34
Modelling .......................................................................................................................................... 34
Reporting .......................................................................................................................................... 35
Validity and generalisability .............................................................................................................. 35
CHAPTER 1: OUTLINE OF THE EVIDENCE PAPER ............................................ 38
1.1. Development .......................................................................................................................... 38
1.2 Chapters relevant to the Evidence Paper ................................................................................... 39
Tables ................................................................................................................................................ 40
CHAPTER 2: EVIDENCE: PREVALENCE OF OVERWEIGHT AND OBESITY .......... 42
2.1. Measurement of overweight and obesity ................................................................................. 42
2.1.1. Introduction ........................................................................................................................ 42
2.1.2 Waist Circumference and its relationship to BMI ............................................................... 42
2.1.3. Body Mass Index (BMI) ....................................................................................................... 43
2.2. International/European evidence .............................................................................................. 45
2.2.1. Introduction ........................................................................................................................ 45
2.2.2. Current child prevalence ..................................................................................................... 46
2.2.3. Recent trends in child prevalence ....................................................................................... 49
2.2.4. Inequalities in child prevalence .......................................................................................... 52
2.3. Evidence from JANPA WP 4 countries ....................................................................................... 54
2.3.1. Current child prevalence ..................................................................................................... 54
2.3.2. Recent trends in child prevalence ....................................................................................... 58
2.3.3. Inequalities in child prevalence .......................................................................................... 62
Tables ................................................................................................................................................ 66
CHAPTER 3: EVIDENCE: CHILDHOOD IMPACTS OF CHILDHOOD OVERWEIGHT
AND OBESITY ............................................................................................... 84
3.1. International/European evidence .............................................................................................. 84
3.1.1. Introduction ........................................................................................................................ 84
3.1.2. The weight of the evidence ................................................................................................. 85
3.1.3. Cardio-metabolic and cardio-vascular risk factors ............................................................. 85
3.1.4. Type 2 diabetes ................................................................................................................... 87
3.1.5. Type 1 diabetes ................................................................................................................... 87
3.1.6. Asthma ................................................................................................................................ 87
3.1.7. Dental health ....................................................................................................................... 88
3.1.8. Orthopaedic and musculoskeletal problems ...................................................................... 88
3.1.9. Sleep disorders and sleep problems ................................................................................... 89
3.1.10. Other physical co-morbidities ........................................................................................... 89
3.1.11. Self-esteem and quality of life .......................................................................................... 90
3.1.12. Depression/low mood....................................................................................................... 90
3.1.13. Educational achievement and attainment ........................................................................ 91
3.2. Evidence in JANPA WP4 countries ............................................................................................. 92
3.2.1. Overview ............................................................................................................................. 92
3.2.2. Croatia ................................................................................................................................. 94
3.2.3. Greece ................................................................................................................................. 94
3.2.4. Ireland ................................................................................................................................. 95
3.2.5. Italy...................................................................................................................................... 96
3.2.6. Portugal ............................................................................................................................... 97
11
3.2.7. Romania .............................................................................................................................. 97
3.2.8. Slovenia ............................................................................................................................... 98
Tables ................................................................................................................................................ 99
CHAPTER 4: EVIDENCE: ADULT IMPACTS OF CHILDHOOD OVERWEIGHT AND
OBESITY ..................................................................................................... 119
4.1. Introduction ............................................................................................................................. 119
4.2. Child or adolescent overweight and obesity and adult morbidities ........................................ 120
4.2.1. Type 2 diabetes ................................................................................................................. 120
4.2.2. Coronary heart disease (CHD) and ischaemic heart disease (IHD) ................................... 120
4.2.3. Stroke ................................................................................................................................ 121
4.2.4. Cancers .............................................................................................................................. 121
4.2.5. Metabolic syndrome ......................................................................................................... 122
4.2.6. Components of metabolic syndrome ............................................................................... 122
4.2.7. Asthma .............................................................................................................................. 124
4.2.8. Musculo-skeletal problems ............................................................................................... 125
4.2.9. Reproductive health .......................................................................................................... 125
4.3. Adult overweight and obesity .................................................................................................. 126
4.4. Adult mortality ......................................................................................................................... 126
4.5. Other adult outcomes .............................................................................................................. 127
4.5.1. Sick leave ........................................................................................................................... 127
4.5.2. Disability pension .............................................................................................................. 127
4.5.3. Lifetime productivity losses .............................................................................................. 127
4.5.4. Educational attainment..................................................................................................... 128
4.5.5. Income .............................................................................................................................. 128
4.5.6. Psychological health .......................................................................................................... 129
Tables .............................................................................................................................................. 130
CHAPTER 5: OUTLINE THE STUDY PROTOCOLS ........................................... 150
5.1 Development ............................................................................................................................. 150
5.2 Chapters relevant to the Study Protocols ................................................................................. 150
CHAPTER 6: EXISTING STUDIES OF LIFETIME COST OF CHILDHOOD
OVERWEIGHT AND OBESITY ....................................................................... 151
6.1. Approaches used to estimate costs ......................................................................................... 151
6.2. International/European reviews .............................................................................................. 154
6.2.1. Studies of (direct) healthcare costs based on US data (Hamilton et al (2016) review) .... 156
6.2.2. Studies of (direct) healthcare costs based on European data (Hamilton et al (2016)
review) ........................................................................................................................................ 165
6.2.3. Studies of (indirect) societal costs based on US data (Hamilton et al (2016) review) ...... 168
6.2.4. Studies of (indirect) societal costs based on European data (Hamilton et al (2016) review)
.................................................................................................................................................... 168
Tables .............................................................................................................................................. 171
CHAPTER 7: OVERVIEW OF MODELLING METHODOLOGY ........................... 176
7.1 EU countries participating in JANPA WP4 ................................................................................. 176
7.2 Governance ............................................................................................................................... 176
7.2.1 Expert International Scientific Advisory Committee (ISAC) ............................................... 176
7.2.2 Study principles .................................................................................................................. 177
7.3 JANPA WP4 aims and objectives ............................................................................................... 178
7.4 Challenges ................................................................................................................................. 178
7.5 General issues .......................................................................................................................... 179
7.5.1 Incorporating children ....................................................................................................... 179
7.5.2 Incorporating societal impacts .......................................................................................... 180
7.6 Model inputs, outputs and metrics........................................................................................... 181
7.6.1 Impact-cost indicators ....................................................................................................... 181
7.6.2 Excess Metrics .................................................................................................................... 181
7.6.3 Effect Metrics ..................................................................................................................... 182
7.7 Modelling .................................................................................................................................. 182
7.7.1 Cohort simulation studies .................................................................................................. 182
7.7.2 Modelling steps .................................................................................................................. 183
7.7.3. Adaptation of UKHF’s modelling software ....................................................................... 184
7.8 Reporting................................................................................................................................... 184
7.9 Validity and generalisability ...................................................................................................... 185
7.9.1 Validity ............................................................................................................................... 185
7.9.2 Generalisability .................................................................................................................. 185
CHAPTER 8: MODEL INPUTS, OUTPUTS AND METRICS ............................... 186
8.1 Research and data domains ...................................................................................................... 186
8.2 Population estimates ................................................................................................................ 188
8.3 BMI ............................................................................................................................................ 188
8.3.1 Current BMI distribution and trends.................................................................................. 188
13
8.3.2 Reductions in childhood obesity ........................................................................................ 189
8.4 Health impacts ......................................................................................................................... 189
8.4.1 Childhood disease risks ...................................................................................................... 189
8.4.2 Adult disease risks .............................................................................................................. 191
8.4.3 Disease incidence/prevalence ....................................................................................... 193
8.4.4 Morality ............................................................................................................................. 193
8.5 Direct healthcare costs ............................................................................................................ 194
8.6 Societal impacts and costs ........................................................................................................ 194
8.6.1 Childhood ....................................................................................................................... 194
8.6.2 Adulthood ....................................................................................................................... 194
8.7 Data collation ....................................................................................................................... 194
8.7.1 Data collation workflow .................................................................................................... 194
8.7.2 Use of proxy data .............................................................................................................. 195
8.7.3 Top-down and Bottom-up approaches .............................................................................. 195
8.7.4 Data cleaning...................................................................................................................... 196
8.8 Impact-cost indicators, excess metrics and effect metrics ...................................................... 197
8.8.1 Impact-cost indicators ....................................................................................................... 197
8.8.2 Excess metrics ................................................................................................................... 197
8.8.3 Effect metrics.................................................................................................................. 198
Tables .............................................................................................................................................. 199
CHAPTER 9: MODELLING ............................................................................ 201
9.1 Modelling software ................................................................................................................... 201
9.1.1 Existing modelling software .......................................................................................... 201
9.1.2 Adaptation of UKHF’s modelling software ........................................................................ 202
9.2 Summary of modelling steps .................................................................................................... 203
9.3 Step 1: Initialising the virtual cohort ......................................................................................... 204
9.4 Steps 2a and 2b: Forecasting BMI distributions and simulating lifetime BMI trajectories ...... 204
9.5 Steps 3a – 3d: Simulating impacts and estimating costs .......................................................... 205
9.5.1 Step 3a: Simulating health impacts ................................................................................... 205
9.5.2 Step 3b: Estimating direct healthcare costs ....................................................................... 206
9.5.3 Step 3c: Simulating societal impacts .................................................................................. 206
9.5.4 Step 3d: Estimating societal costs ..................................................................................... 206
9.6 Producing Model Output Tables .............................................................................................. 206
CHAPTER 10: REPORTING ........................................................................... 208
10.1 Reporting work flow ............................................................................................................... 208
10.2 Calculating excess metrics , effect metrics and producing graphical outputs ....................... 208
CHAPTER 11: ASSESSING VALIDITY AND GENERALISABILITY ....................... 209
11.1 Validity .................................................................................................................................... 209
11.1.1 Comparison of model-based and research-based relative risks ...................................... 209
11.1.2. Methods of modelling lifetime BMI trajectories ............................................................ 213
11.1.3 The independent disease processes assumption ............................................................ 213
11.1.4. Sensitivity analysis .......................................................................................................... 213
11.2 Generalisability ....................................................................................................................... 213
11.2.1 EConDA online tool .......................................................................................................... 214
11.2.2 Modelling resources ........................................................................................................ 214
REFERENCES ............................................................................................... 217
APPENDIX 1: POSSIBLE NON-MODELLING PROJECTS .................................. 236
A1.1 Conditions that could not be included in the modelling ........................................................ 236
A1.2 Experiences of morbidly obese children and their families ................................................... 236
A1.3 Childhood obesity and educational outcomes ....................................................................... 236
A1.4 Inequalities ............................................................................................................................. 236
APPENDIX 2: LIMITATIONS IN EVIDENCE, DATA AND MODELLING
(preliminary list) ........................................................................................ 238
A2.1 Evidence .................................................................................................................................. 238
A2.2 Data ......................................................................................................................................... 238
A2.3 Modelling ................................................................................................................................ 239
15
ACRONYMS AND ABBREVIATIONS
ACR Albumin to Creatinine Ratio
ADHD Attention Deficit Hyperactivity Disorder (also referred to as ADD)
ALSPAC Avon Longitudinal Study of Parents and Children (UK)
AO Abdominal Obesity
BMI Body Mass Index
BP Blood Pressure
CASP Critical Appraisal Skills Programme (checklist)
CCHS Canadian Community Health Survey
CDC Centers for Disease Control (US)
CHD Coronary Heart Disease
CI Confidence Interval
CIMT Carotid Intima-Media Thickness
CLVH Concentric Left Ventricular Hypertrophy
CMAP Central Mean Arterial Pressure
COI Cost Of Illness
COPD Chronic Obstructive Pulmonary Disease
COSI Childhood Obesity Surveillance Initiative
CVD Cardio-Vascular Disease
DCD Developmental Co-ordination Disorder
EU European Union
ENERGY European Energy Balance Research to Prevent Excessive Weight Gain Among Youth
FACCT Fluoride and Caring for Children’s Teeth
GDP Gross Domestic Product
GPA Grade-Point Average
GUI Growing Up in Ireland
HBSC Health Behaviour in School-aged Children
HDI Human Development Index
HDL High-Density Lipoprotein (cholesterol)
HELENA Healthy Lifestyle in Europe by Nutrition in Adolescence
HOMA-IR Homeostatic Model Assessment-Insulin Resistance
HR Hazard Ratio
HSE Health Survey for England
HW Healthy Weight
ICD International Statistical Classification of Diseases and Related Health Problems
IDEFICS Identification and Prevention of Dietary- and Lifestyle-Induced Health Effects in Children and Infants
IDF International Diabetes Federation
IGT Impaired Glucose Tolerance
IHD Ischaemic Heart Disease
IIH Idiopathic Intracranial Hypertension
IOTF International Obesity Task Force
IR Insulin Resistance
ISAC International Scientific Advisory Committee
JANPA Joint Action on Nutrition and Physical Activity
KNHANES Korean National Health and Nutrition Examination Survey
LDL Low-Density Lipoprotein (cholesterol)
LE Life Expectancy
MEPS Medical Expenditure Panel Survey (US)
MS Metabolic Syndrome
MSC Musculo-Skeletal Complaints
MSKI Musculo-Skeletal Impairments
NAFLD Non-Alcoholic Fatty Liver Disease
NHANES National Health and Nutrition Examination Survey (US)
NHS National Health Survey (Australia)
NPHS National Population Health Survey (Canada)
NR Not Reported
NUI National University of Ireland
OB Obese
OR Odds Ratio
OECD Organisation for Economic Co-operation and Development
OW Overweight
P Probability level
PAF Population Attributable Fraction
PCOS Polycystic Ovary Syndrome
PD Pre-Diabetes
PWV Pulse Wave Velocity
QALE Quality Adjusted Life Expectancy
QALY(s) Quality Adjusted Life Year(s)
RIVM-CDM National Institute for Public Health and the Environment Chronic Disease Model (Netherlands)
RR Risk Ratio / Relative risk
SD Standard Deviation
SES Socio-Economic Status
TG Triglycerides
UCC University College Cork (Ireland)
US United States
WC Waist Circumference
WHO World Health Organization
WOMAC Western Ontario and McMasters Universities Osteoarthritis Index
WP4 Work Package 4
zBMI Standardised BMI scores
17
CONVENTIONS AND DEFINITIONS
Adult 18 or more years except if age of majority is younger
Adulthood age categories for reporting
Age categories for adults that are used in the and table of model outputs:
18 – 24 years
25 years – 74 years
75+ years
Adult Healthy Weight (HW) 18.5 ≤ BMI < 25.0
Adult obesity (OB)1 Defined by WHO cut-off point (30.0 ≤ BMI)
Adult Overweight (OW) 25.0 ≤ BMI < 30.0
Advanced study A more involved participation in WP4
Adult Underweight (UW) BMI < 18.52
Basic study A less involved participation in WP4
Body Mass Index (BMI)
Three BMI categories will be used throughout the lifecourse:
Healthy weight (HW)
Overweight (OW)
Obese (OB)
Bottom-up methods Methods used to estimate impact-related and cost-related model inputs and outputs that are based on analysis of disease and healthcare data in cross-sectional studies or longitudinal studies that also include BMI data
Child 0-17 years except if age of majority is younger
Childhood age categories for reporting
Age categories for children that are used in the tables of model outputs:
Younger children: 0 – 6 years
Older children: 7 -11 years
Adolescents: 12 – 17 years
Childhood obesity Bases on an individual’s BMI at age 17 years as they enter their last year of childhood (using IOTF cut-off points).
Cohort simulation model A simulation model that takes an initial cohort (representative of the population at the time), ages them and simulates their experiences throughout their lives. No additional entries or exits from the cohort (except by death of existing cohort members) are
1 WHO defines three sub-categories of obesity: these are not considered in this study because of lack of data.
Obesity category I (OB-I): 30.0 ≤BMI < 35.0)
Obesity category II (OB-II): 35.0 ≤ BMI < 40.0)
Obesity category III (OB-III): 40.) ≤ BMI
2 Underweight individuals are included in the Healthy Weight (HW) category
allowed. A broad approach to burden of disease and cost of illness studies; their primary interest is in the current and future experiences of the initial cohort and not the whole population living in any future year.
Current year 2016
Current value Cost expressed in 2016 euros
Direct costs Costs that result from outpatient and inpatient health services (including surgery), laboratory and radiological tests, and drug therapy.
Discounting
Discounting of future disease and disability and costs (because people tend to devalue future disease and disability and costs compared to present) is considered to be best practice.
Effect metric
Describes the effect of a reduction in current childhood obesity rates on an excess metric
Excess metric
Describes an excess in some impact-cost indicator (e.g. direct healthcare costs) that can be associated with current childhood obesity.
Friction-cost approach An alternative approach for estimating value of productivity losses (see Human-capital l approach)
Human-capital approach The approach adopted for estimating value of losses (see Friction-Loss approach)
Impact-cost indicators Model outputs that capture the impacts and costs that are incur as a result of childhood obesity and overweight
Incremental lifetime costs All costs must be compared to a child without the condition
Indirect costs Also called societal costs (see societal costs)
Population simulation model Population simulation models allow new cohort members to be added or subtracted from the cohort, and between individual variation to be modelled. Their primary interest is in the experineces – impacts and costs – of the total population (current or future).
IOTF cut-off points
IOTF (now called World Obesity Federation) cut-off points will be used to categorise childhood BMI. They apply to 2–17 year olds and map to WHO’s adult BMI cut-off points
Lifetime BMI trajectory Lifetime trajectory of an individual’s annual BMI values throughout their life
Life Expectancy (LE) Can be measured at different ages
Obesity or overweight (OW/OB)
A generic term used for a group of individuals who are overweight or obese (Jonoula et al)
Obesity-related impacts
Two types of consequences of childhood obesity and overweight are considered:
Health impacts (diseases, disability and death)
Societal impacts (adult productivity losses and lifetime income loss)
per case
Based on the number of cases of a disease and not the underlying population size
Population Attributable Fraction (PAF)
The proportion of an impact that would be avoided if a particular risk factor was eliminated
Population simulation model
A simulation model that takes an initial cohort (representative of the population at the time), ages its members and simulates their experiences throughout their lives. Additional entries (births and immigration) and exits (emigration) are allowed to join as the
19
cohort ages so that the boosted cohort remains representative of the whole population living in any future years. A broad approach to burden of disease and cost of illness studies; the primary interest here is in the current and future experiences of the whole population.
Presenteeism Not covered in the modelling. Reduced productivity while attending work associated with obesity-related disease or disability.
Private costs Costs incurred privately by patients and their families and not by the health and social care system
Relative Risk (RR) Also Odds Ratio (OR)
Sensitivity analysis
To represent the uncertainties inherent in data and modelling assumptions
Societal costs A type of indirect cost. These are the other resources that society and its citizens and communities forego as a result of a health condition
Societal economic perspective
Includes impacts experienced and cost incurred by society and its communities
Start-year First year of the simulation (2016)
Stochastic models Statistical models that operate probabilistically with random model parameters having known distributions. For example:
The virtual individuals (virtual cohort) are sampled from a theoretical population that has a pre-specified population distribution. At least asymptotically, the sample and the population of interest have the same distribution
Transition probabilities and other model inputs are random variables unknown and sampled from pre-assigned distributions
Top-down methods Methods used to estimate impact-related and cost-related model inputs and outputs that are based on the application of Population Attributable Fractions (PAFs) to national disease and healthcare data
Years of Potential Life Lost (YPLL)
Years of life lost up to an individual’s national life expectancy in their birth year
zBMI scores Because cut-off points for overweight and obesity vary with age, gender-specific standardised z-score cut-off points will be used to define BMI status at different ages.
21
SUMMARIES
SUMMARY: EVIDENCE
Background
The Evidence Paper covers prevalence, health and societal impacts, healthcare and societal costs,
evidence and experience of socially disadvantaged in the EU as well as the availability and quality of
the data in the countries.
It is the product of collaboration between the JANPA WP4 Lead Team, the Irish National Team
supported by significant additional funding from the safefood and the National Teams in the JANPA
WP4 countries.
Prevalence of child overweight and obesity
127 publications on the prevalence of child overweight and obesity came from the JANPA WP4
countries: 32 publications examining trends, and 65 papers on inequalities is considered. These
‘local’ materials complement the international review by providing national contexts in which to
consider the international evidence. They also highlight some gaps in information in these countries.
Prevalence and inequalities
The most commonly used measure of overweight/obesity is body mass index (BMI). Various
cut-points for the classification of children BMI are in use; in recent years, the most widely-
used system in Europe is that of the International Obesity Task Force (IOTF).
Round 2 of the Childhood Obesity Surveillance Initiative (COSI), conducted in 2009-2010,
resulted in median values for overweight and obese boys in 13 European countries of 13.7%
and 6.7% respectively, and 15.7% and 6.7% respectively in girls aged 8 years of age (IOTF
cut-points) for 13 countries, five of which take part in JANPA (Wijnhoven et al., 2014a).
In 2013, the prevalence of overweight and obesity among children aged 2 to 19 years (IOTF
cut-points) was higher in Western Europe (including Mediterranean countries) (24.2% in
boys and 22.0% in girls) than in Eastern Europe (about 19% in both sexes) and Central
Europe (21.3% in boys and 20.3% in girls) (Ng et al., 2014).
Studies on inequalities in the prevalence of childhood overweight and obesity have generally
found inverse associations between measures of socio-economic status and prevalence:
o The most consistent associations are found with parental education (Shrewsbury &
Wardle, 2008).
o Children born outside the country under study tend to have a higher prevalence of
overweight and obesity, but the relationship between immigrant status and
overweight/obesity varies (Labree et al., 2011).
o Various other characteristics, some of which are confounded with socio-economic
status (SES), are consistently associated with overweight and obesity: these include
parental BMI, dietary intake, and physical activity.
Trends
The trends in overweight and obesity are mixed, but the overall picture is that, following a
rapid escalation during the 1990s, prevalence may be slowing down or stabilising since the
23
early- to mid-2000s. However, there is no consistent or long-term evidence that prevalence
in children is decreasing (Rokholm et al., 2010).
Trends in the prevalence of childhood overweight and obesity should be interpreted
cautiously, since sampling, measurement and reporting methods vary widely, and a focus on
BMI may mask increases in waist circumference (Visscher et al., 2015).
Evidence gaps
The main gaps in information relate to:
o The lack of standardised surveillance of BMI in pre-school children and adolescents
o The lack of data on waist circumference
o Difficulties in establishing trends over time with respect to socio-economic sub-
groups and individuals of different ethnic and migrant status.
Child impacts of childhood overweight and obesity
Evidence for child impacts of childhood overweight and obesity in adulthood comes from an
international review conducted by Queally et al. (2016). The review summarises evidence from 18
published reviews. Their review is supplemented by a consideration of the international evidence on
the associations between child or adolescent overweight/obesity and educational outcomes. JANPA
WP4 countries also submitted 81 sources of local evidence on this topic.
Underdeveloped evidence base
In the 18 review papers, it was very common for authors to cite the following challenges and
limitations:
o There is a lack of high-quality longitudinal data, which hampers the establishment of
cause-effect relationships, particularly for conditions such as asthma and
depression.
o There are large differences across individual studies in terms of how children’s
weight status has been classified.
o There is large variation in the extent to which studies controlled for confounders
such as socio-economic status.
o There are inconsistencies in the extent to which differences by gender are
examined.
o There is a lack of evidence and data on differences among ethnic/racial groups.
There are some gaps in the evidence base in both the international review and the local
materials from JANPA countries, mostly stemming from the relatively low number of studies
that have examined impacts of childhood overweight/obesity, particularly non-medical
impacts, longitudinally.
Range of impacts
Internationally, the bulk of studies that examine the impact of child/adolescent
overweight/obesity have focused on cardio-metabolic risk factors, psychological ill-health
and reduced quality of life (Pulgarón, 2013; Sanders et al., 2015).
A large majority – over 90% – of the 81 ‘local’ sources retrieved from the JANPA WP4
countries examined health impacts of overweight and obesity in childhood, while only about
10% considered other societal impacts.
Cardio-metabolic conditions
There is strong and consistent evidence for increased cardio-metabolic risk among children
and adolescents of higher weight status. For example, in a meta-analysis of 24 studies
(Friedemann et al., 2012), the mean values of diasystolic, systolic and ambulatory BP, total,
HDL and LDL cholesterol and triglycerides, and fasting glucose, fasting insulin and HOMA-IR,
and CIMT and left ventricular mass were computed for healthy weight, overweight and
obese groups. In all cases, differences were statistically significant, with larger differences in
comparisons of obese vs. healthy weight than in overweight vs. healthy weight.
There is also strong evidence for links between childhood overweight and obesity and risk of
both type 1 and type 2 diabetes, though relatively little is known about these conditions in
children compared with adults. In the case of type 2 diabetes, there is a scarcity of estimates
of risk associated with increased weight status in children in adolescents. One study
conducted in Israel with over 1 million 17 year-old adolescents receiving a medical
evaluation for military service found that obesity (compared to healthy weight) was
associated with type 2 diabetes (OR = 5.56 and OR = 4.42, for male and female subjects,
respectively) after controlling for origin, level of education and the year of recruitment
(Pulgarón, 2013). Verbeeten et al.’s (2011) meta-analysis suggests an odds ratio of 1.25 for
type 1 diabetes for every standard deviation increase in children’s BMI.
Of the ‘local’ materials examining health impacts, a majority of sources covered aspects of
cardio-metabolic health (69%), including multiple aspects of the metabolic syndrome
(36.5%), blood pressure (13%) and diabetes or blood glucose profiles (13%), with smaller
numbers of papers examining specific aspects of cardio-metabolic health or risk factors,
including liver abnormalities and arterial thickness.
Consistent with the international review, these local materials provide strong evidence for
negative impacts on child and adolescent cardio-metabolic profiles. There is also reasonably
consistent, though less widespread evidence, for negative impacts on child/adolescent
musculo-skeletal/motor and pulmonary/aerobic functioning.
Respiratory conditions
In a systematic review and meta-analysis of the association between asthma or wheezing
and childhood overweight/obesity (Mebrahtu et al., 2015), it was estimated that the risk was
increased by 23% among overweight and obese children. However the causal direction of
this association is unclear (Pulgarón, 2013).
Four studies (Sanders et al., 2015) examined associations between obstructive sleep apnoea
and child/adolescent weight status, and the association appears to be stronger among
adolescents than in younger children. Pulgarón’s (2013) review concluded that while there is
good evidence to show that sleep problems are more prevalent with increased weight
status, the long-term effects of this are unclear.
Dental health
25
Two systematic reviews on the associations between child/adolescent weight status and
dental health (number of caries) were included in Queally et al.’s (2016) review (Hayden et
al., 2013; Hooley et al., 2012). They suggest that these associations are stronger in high-
income countries, but further research which accounts for socio-economic and dietary
factors is needed. Hooley et al. (2012) reviewed the results of 48 studies and found that 23
studies reported no association, 17 reported a positive association, 9 reported an inverse
relationship, and 1 reported a U shaped pattern of association. Studies reporting a positive
association were from countries with a higher Human Development Index (HDI) score
(mainly Europe and US), higher quality dental services (more sensitive dental examination)
and a low percentage of underweight children in the population, while studies reporting a
negative association were from countries with a lower HDI score (mainly Asia and South
America), lower quality dental services (less sensitive dental examination), and more
underweight children.
Musculoskeletal conditions
Paulis et al. (2013) conducted a systematic review on the association between weight status
and musculoskeletal complaints (MSC) in children (aged 0-18 years). This provides medium
quality evidence that being overweight in childhood is positively associated with
musculoskeletal pain (RR = 1.26). There was also evidence of an association between
childhood weight and low back pain, as well as injuries and fractures, though evidence for
these associations was of lower quality.
Thivel et al. (2016) reviewed studies that examined the association between child weight
status and muscle strength and fitness. Although these studies varied in design and
comparisons of laboratory-based and field-based results were challenging, the review
provides strong evidence that children and adolescents with obesity have reduced muscular
fitness compared with children and adolescents of healthy weight. Thivel et al. (2016) call for
more research in this area, given the associations between muscular and musculo-skeletal
fitness with overall health.
About 9% of “local” materials sources examined aspects of children’s musculo-skeletal or
motor functioning, and 6% looked at pulmonary function or aerobic capacity. One or two
sources covered each of dental health, hormonal health (in girls), and idiopathic intracranial
hypertension.
Cognitive development
One of the 18 papers from Queally et al. (2016) consisted of a systematic review of
developmental co-ordination disorder (DCD; Hendrix et al., 2014). The prevalence of DCD
was estimated to range from 1.7% to 6%, and occurs four to seven times more often in boys
than in girls. All 21 studies in the review reported that children with DCD had higher BMI.
Mental health and quality of life
A systematic review by Griffiths et al. (2010) provides strong evidence that paediatric obesity
impacts on self-esteem and quality of life. Six of nine studies in their review found lower
global self-esteem in obese compared with healthy weight children and adolescents. Nine
out of eleven studies using child self-reports, and six out of seven studies using parental
reports, found significantly lower total quality of life scores in obese youth.
Mühlig et al.’s (2015) systematic review on associations between child/adolescent
overweight/obesity and child/adolescent depression/depressive symptoms found that
relationships were stronger in female adolescents and in cross-sectional studies compared
with longitudinal analyses. Out of 19, 14 cross-sectional studies confirmed a significant
association between obesity and depression. However, just three out of eight longitudinal
studies reported associations between obesity and subsequent depression. Mühlig et al.
(2015) proposed that overweight/obesity and depression may develop jointly over time, but
noted that longitudinal data on young people is too scarce to draw firm conclusions. A meta-
analysis of the relationships between adult depression and weight status (Luppino et al.,
2010) confirms that there is a reciprocal relationship between these two outcomes, which
may become reinforced over time.
A majority of the ‘local’ sources that examined other impacts covered aspects of
psychological or emotional wellbeing, while only one source examined the association
between child overweight/obesity and academic performance, and one examined subjective
quality of life.
The relatively small number of ‘local’ studies that examined emotional or psychological
impacts are difficult to compare due to differences in measures and analysis methods, but
they suggest negative associations (which are likely to operate bi-directionally) between
measures of psychological and emotional wellbeing and overweight and obesity.
Educational outcomes
There is evidence for a weak negative association between childhood overweight or obesity
and educational attainment, and much of this relationship can be accounted for by socio-
economic disparities between normal-weight and overweight or obese groups of children
(Caird et al., 2014). Few studies have examined these associations longitudinally, and those
that have provide conflicting evidence about the causal direction of this relationship (Sassi et
al., 2009; Booth et al., 2014).
Evidence gaps
There is strong evidence for associations between childhood overweight and obesity and risk factors
for cardio-metabolic morbidities. However, less is known about how these relationships accumulate
or change over time in adulthood:
Most of the studies reviewed in this section, both from the international evidence, and the
‘local’ evidence from the JANPA WP4 countries draw on cross-sectional data; therefore, the
causal direction of relationships cannot be determined.
Many potential confounders complicate researchers’ attempts to isolate the effects of
overweight and obesity in childhood.
There is also considerable interdependency among co-morbidities and outcomes. One
confounder that may need to be better accounted for in research in this area is puberty
onset.
27
Further work in this area might address these gaps through longitudinal analysis, particularly of
psychological and educational outcomes, where the longitudinal evidence is extremely scarce.
Queally et al. (2016) have noted the need for standardised approaches in analyses and reporting
of effect sizes (odds ratios; risk ratios) in terms of weight status as well as more uniform
reporting of adjusted and unadjusted effect sizes.
Adult impacts of childhood overweight and obesity
Evidence for impacts of child or adolescent overweight/obesity in adulthood come from an
international review conducted by McCarthy et al. (2016b). The review covers morbidity, mortality,
disability, and non-medical outcomes such as lifetime productivity, and is based on 13 systematic
reviews/meta-analyses, supplemented with data from 15 individual studies. A large majority of these
studies are longitudinal in design and incorporate measured (rather than self-reported) BMI, and so
are considered to be of high quality. None of the JANPA participants submitted local evidence on this
topic, which is illustrative of a gap in the evidence base.
Underdeveloped evidence base
Establishing firm evidence of a link between child/adolescent weight status and adult
outcomes is complex. For many of the outcomes considered, there is a scarcity of high-
quality longitudinal data. Also, studies varied in the extent to which adjustments were made
for potential confounding variables, effect estimates were reported in a variety of ways, and
BMI was also classified in a variety of ways.
The most challenging issue in considering the evidence base for impacts in adulthood arising
from childhood or adolescent overweight/obesity is the manner in which changes in
individuals’ BMI over time are incorporated into analyses. Several studies report that
associations between child/adolescent BMI status attenuate (reduce) once adjustments for
adult BMI are built into regression analyses. However, adult BMI status is likely to be causally
linked to child BMI status. Therefore, adjusting for adult BMI risks ‘over-adjusting’.
Mortality
There is limited evidence to support a link between all-cause mortality in adulthood and
overweight and obesity in childhood or adolescence (Park et al., 2012; Adami et al., 2008),
but a majority of these studies did not adjust for socio-economic status. A recent exception
is a study by Twig et al. (2016) which reported strong associations between overweight and
obesity in adolescence and cardiovascular mortality in adulthood in a cohort of 2.3 million
Israeli adolescents, after adjustments for age, sex, socio-economic status, education level
and country of origin.
Childhood BMI and adult BMI
Regardless of whether child/adolescent BMI status is an independent risk factor in the
outcomes considered, there is strong and consistent evidence for a link between child,
adolescent and adult BMI. Around 55% of obese children go on to be obese in adolescence,
80% of obese adolescents will still be obese in adulthood, and 70% will be obese over age 30
(Simmonds et al., 2016). However, 70% of obese adults were not obese in childhood or
adolescence, so overweight/obesity in childhood is only part of a larger problem.
Cardiovascular and metabloic conditions
There is some evidence for a relationship between child/adolescent weight status and
occurrence of metabolic syndrome in adulthood (Lloyd et al., 2012).
Studies on some of the components of metabolic syndrome (total cholesterol, HDL and LDL
cholesterol, triglycerides, insulin resistance, hypertension, carotid artery atherosclerosis, and
non-alcoholic fatty liver disease) were also included in this review:
o Evidence for a link between earlier BMI status and subsequent total cholesterol, HDL
and LDL cholesterol levels is mixed. There is evidence for an association between
child/adolescent BMI and adult triglyceride levels, but this is attenuated after
adjusting for adult BMI (Juonala et al., 2011; Lloyd et al., 2012).
o There is evidence for an association between childhood BMI and hypertension in
adulthood (Llewellyn et al., 2016; Part et al., 2012), though again, adjustments for
adult BMI status attenuate this association (Lloyd et al., 2012).
o Similarly, the evidence supports a positive association between BMI in childhood
and carotid artery atherosclerosis in adults, but the association tends to be
attenuated or inversed if adjustments are made for adult BMI status in analyses
(Juonala et al., 2011; Lloyd et al., 2010).
o Insulin resistance in adulthood is positively associated with elevated BMI in
childhood/adolescence, but this association tends to disappear with adjustments for
adult BMI status (Lloyd et al., 2012).
o There is a scarcity of evidence for associations between earlier BMI status and
subsequent risk of non-alcoholic fatty liver disease (NAFLD). Only one study in
McCarthy et al.’s (2016b) review examined this outcome, which found that increases
in child BMI over time, rather than absolute values of BMI, were associated with an
increased risk for NAFLD (Zimmerman et al., 2015).
There is also evidence from three systematic reviews (Llewellyn et al., 2016; Park et al.,
2012; Owen et al., 2009) for a link between childhood overweight/obesity and coronary
heart disease (CHD) in adulthood, though results suggest that higher BMI in later childhood
rather than early childhood is the greater risk.
Two systematic reviews (Llewellyn et al., 2016; Park et al., 2012) indicate that there is not
strong evidence for an association between childhood BMI and stroke in adulthood.
Evidence from two recent systematic reviews and one meta-analysis (Llewellyn et al., 2016;
Juonala et al., 2011; Park et al., 2012) supports a strong and consistent link between
child/adolescent BMI status and risk of type 2 diabetes in adulthood.
Cancers
The evidence for associations between BMI status in childhood or adolescence and cancers
is mixed. Findings depend on the type of cancer studied and also whether cancer incidence
or cancer mortality is considered. There also appear to be gender differences in terms of
29
impact for some forms of cancer (Llewellyn et al., 2016; Park et al., 2012; Aarestrup et al.,
2014, 2016; Kitahara et al., 2014a, 2014b).
Respiratory conditions
Just three individual studies identified by McCarthy et al. (2016b) examined asthma in
adulthood, and results from these studies are mixed (reviewed in Park et al., 2012; Reilly &
Kelly, 2011).
Musculoskeletal conditions
Associations between child or adolescent weight status and musculoskeletal problems in
adulthood vary depending on the type of problem or symptom; evidence is available from
only four sources. One study found no evidence for an association between lower back pain
in adulthood and childhood BMI, but this study examined younger adults only (age 32-33
years; Power et al., 2001). On the other hand, there is some evidence for an increased risk of
knee pain (Park et al., 2012; Antony et al., 2015), as well as arthritis (MacFarlane et al., 2011)
in adulthood.
Reproductive health
Reproductive health in adulthood and its relationship to earlier BMI status has not been
widely studied. McCarthy et al. (2016b) identified just two studies, both concerning only
women. Lake et al. (1997) reported that while adult obesity was associated with menstrual
problems, fertility rates, and hypertension during pregnancy, childhood obesity was
associated with menstrual problems and hypertension in pregnancy only. Polycystic ovary
syndrome (PCOS) in adulthood was associated with BMI at age 16 (Reilly & Kelly, 2011), but
family history of PCOS was not accounted for in this analysis, and PCOS is associated with
insulin resistance (Schwartz & Chadha, 2008).
Mental health
There is very little evidence from high-quality longitudinal studies on relationships between
adult psychological health and BMI status in childhood or adolescence. In the short term,
however, evidence suggests that young adults with elevated BMI in adolescence may have
lower self-esteem and are more socially isolated or lonely (Sikorski et al., 2015). A meta-
analysis of the relationships between adult depression and weight status (Luppino et al.,
2010) supports a reciprocal relationship between these two outcomes, which appear to be
reinforced over time.
McCarthy et al. (2016b) recommend the incorporation of measures of psychological
wellbeing, and more studies on disability, quality of life and productivity loss in this overall
area. They also emphasise the need for standardised, robust approaches to incorporate
changes in BMI over time in future work in this area.
Adult income and productivity
There are a limited number of studies, mainly based on males only, examining links between
BMI in late adolescence and work sick leave, disability pension status, and lifetime
productivity losses. However, the limited evidence supports an association between
adolescent BMI and more adverse outcomes on these measures, which is higher among
obese and morbidly obese than overweight individuals, after adjustments are made for
potential confounders (e.g. Neovius et al., 2012a).
There some evidence for associations between child BMI and later educational attainment
and income. A recent US study (Amis et al., 2014) showed that, after controlling for
demographics, family environment, prior academic achievement, behavioural and general
and mental health variables, obesity at ages 12-18 years was associated with a 9% reduction
in obtaining a college degree, and a 7.5% reduction in annual income 13 years later. These
effects were stronger in women, consistent with earlier studies (Gortmaker et al., 1993;
Sargent & Blanchflower, 1994; Viner et al., 2005).
Gaps in the evidence base
The lack of local evidence on this topic reflects a general gap in knowledge in this area. The material
in this chapter is a summary of an international review of best quality available evidence (McCarthy
et al., 2016b), and within this, there are also gaps:
The most prominent gap in knowledge is in the area of ‘non-medical’ outcomes, where very
few longitudinal studies have examined the long-term impacts of child overweight/obesity
in adulthood in the areas of psychological health, educational attainment, income, disability
status, and lifetime productivity.
In contrast, at first glance, there appears to be strong evidence for associations between
child/adolescent weight status and a number of adult morbidities, particularly type 2
diabetes, CHD, and some components of the metabolic syndrome.
For others, such as some cancers, evidence is mixed, and for still others, such as stroke, the
evidence is weak or absent.
Of course, a lack of evidence does not mean that a link does not exist, since there is not a sufficient
body of evidence in any of the areas considered to conclude definitively that high child/adolescent
BMI is associated with adverse outcomes (or not). Moreover, the evidence base will remain obscure
until more studies take more nuanced approaches to incorporate the complexities of BMI
trajectories over time into analyses.
31
SUMMARY: STUDY PROTOCOLS
Existing studies of lifetime cost of childhood overweight and obesity
Simulation methodology
The use of simulation models of obesity is an area of research at a “nascent stage of
development” (Levy et al., 2011, p. 389).
There are two categories of cost: direct costs and indirect costs.
Comparisons across studies are hampered by differences in modelling methods and
assumptions. Key differences relate to whether or not costs during childhood (<18 years) are
incorporated into the models, methods for calculating costs and cost components included,
incorporation of transitions in BMI status over time, adjustments for differential mortality
rates by BMI status, and whether or not results are reported by age, gender and
race/ethnicity (Finkelstein et al., 2014).
Bierl et al. (2013) have shown that differences between studies in cost outcomes relate
largely to study design parameters, and that modelling (simulation) studies tended to
provide the most conservative estimates.
Regardless of the type of model and costs, each relies on its own particular assumptions,
which should be clear to the users of the results:
o Results of forecasting models should therefore reflect this uncertainty by reporting
confidence intervals and/or sensitivity analyses (Astolfi et al., 2012).
o Regarding simulation models, there should be a clear statement of model structure
(i.e. assumptions, equations and algorithms), data used, and results of validation
exercises (Levy et al., 2011).
o This underlines the importance of decision-makers’ awareness of different purposes,
strengths and weaknesses of different studies when interpreting cost outcomes.
Paucity of evidencce
There is limited published evidence on the direct (healthcare) lifetime costs associated with
child or adolescent overweight/obesity. A recent systematic review (Finkelstein et al., 2014)
retrieved just six US studies on this topic.
There is even more limited evidence on indirect lifetime costs with no existing systematic
reviews on this topic.
No studies to date have modelled indirect costs incurred during childhood.
No materials from JANPA WP4 countries covered lifetime indirect or direct costs associated
with high BMI status in childhood or adolescence.
A recent systematic review of existing studies
A systematic review of the international literature on lifetime costs (Hamilton et al., in
preparation) yielded 13 studies.
o A majority of studies (8) were conducted in the US, with just 5 from Europe (two
from Germany, two from Sweden, and one from the Netherlands).
o Most studies (8) examined direct costs only; four examined indirect costs, and just
one examined both direct and indirect costs.
o Two of the studies were observational cohort studies, and the remaining 11 used
forecast modelling.
o Seven of these 11 studies predominantly used micro-simulation modelling, the other
four cohort modelling.
o Only two studies of 13 identified modelled direct costs incurred during childhood
(Ma & Frick, 2011; Trasande, 2010). Moreover, both of these studies were
conducted in the US and may not be generalizable to the European context.
The systematic review also provide strong evidence that obesity in adolescence has a
negative impact on later adult earnings, independent of a range of potential confounders.
Current estimates of lifetime costs
Finkelstein et al. (2014) estimated that, in a US context, the lifetime excess costs (discounted
at an annual rate of 3%) associated with obesity in a 10 year-old child amount to somewhere
between $12,660 to $19,630 (2012 values). This estimate cannot be generalised to European
contexts due to differences in the costing and treatment structures of health-care systems.
No pooled estimate for indirect costs is available.
Two recent German studies (Sonntag et al., 2015, 2016) have used micro-simulation to
estimate lifetime direct (2015) and indirect (2016) costs. Discounted (3%) lifetime excess
direct costs (incurred after 18 years) were estimated to amount to €4,262 for men and
€7,028 for women. Overweight and obesity during childhood resulted in an excess indirect
lifetime cost, discounted at 3%, per person of approximately €4,209 (men) and €2,445
(women), with a majority of indirect costs incurred prior to age 60. However, it cannot be
assumed that these results are generalizable to other European countries.
Discounting of future imapcts and costs:
An area of debate is whether, and by how much, to discount future costs. Severens and
Milne (2004, p. 399) comment that “neither theoretical nor empirical arguments are
adequate to determine an optimal solution regarding which discounting method and/or
discount rate should be used.” The most commonly used method, uniform discounting
using a constant non-zero discount rate, tends to prioritise immediate treatment at the
expense of prevention, thereby working against long-term public health measures. Some
authors (e.g. Hollingworth et al., 2012) report both discounted and undiscounted rates in an
attempt to address this.
Further research and data needs
This review indicates that there is a strong need for further research in this area in general,
and particularly in the following:
o Direct and indirect costs incurred prior to age 18 years
o Lifetime indirect costs
33
o Application of finer gradings of obesity
o Differential mortality rates by BMI status
o Differences across racial/ethnic groups
o The European context.
JANPA WP4 aims and objectives
Seven European countries are participating in JANPA WP4: Croatia, Italy and Portugal are
participating in basic studies while Greece, Ireland, Romania and Slovenia are participating in
advanced studies.
The aim of JANPA WP4 is to “contribute to the evidence-based economic rationale for action on
childhood obesity”.
Its modelling objectives are, in the seven EU countries participating in JANPA WP4, to:
1.a Describe the current prevalence and trends in childhood overweight and obesity. 1.b Estimate the lifetime impacts and costs of current childhood overweight and obesity. 1.c Breakdown these impacts and costs according to the year they occur 1.d Assess the effect of reducing childhood obesity by 1% and 5% on these impacts and costs
2. Explore the feasibility of generalising the JANPA WP4 modelling methodology to other EU
countries.
JANPA WP4 is essentially a modelling project. However, during development of its Study Protocols it
became clear that several important issues could not be incorporated into the modelling. A number
of non-modelling projects that were not initially part of the work package may be developed (see
Appendix 1).
General issues The evidence (see Chapters 1-4) highlights the challenges of estimating lifetime impacts and costs.
These include:
Incorporating children into existing simulation models
Incorportaing childhood health impacts and direct healthcare costs of childhood obesity
Incorporating adult health impacts and direct healthcare costs of cildhood obesity
Incorporating societal impacts (adult productivity losses and lifetime income losses)
Incorporating acute conditions
Model metrics
Model outputs will be expressed in terms of excess metrics and effect metrics which are in turn
constructed from impact-cost indicators that describe various aspects of children’s lifetime
experiences such as number of new disease cases, direct healthcare costs, a ult productivity losses,
Quality Adjusted Life Years (QALYs), etc. These indicators are outputted by the modelling software.
Excess metrics describe excesses in these impact-cost indicators that are associated with current childhood obesity. They are differences between the value of an indicator amongst individuals who were overweight or obese as children and its value amongst individuals who were of healthy weight as children.
Corresponding to each excess metric there is an effect metric that describes the effect of a reduction in childhood obesity on the excess. Effect metrics are differences between the value of an excess metric in one of the reduction scenarios and its current value.
Diseases and other impacts
The (childhood and adult) diseases and societal impacts that are included in a county’s model are
determined in a two stage process:
1. Firstly, an initial list of diseases and impacts that are significantly associated with (childhood)
obesity and overweight are identified from a review of international and local materials.
2. Secondly; diseases and impacts for which inadequate local data or acceptable proxy data or
a variable are removed.
Modelling
Existing lifetime costing studies link childhood obesity to adult consequences through its link to adult
obesity within a simulation model that uses modelled individual lifetime BMI trajectories (see Figure
below)
Figure. Modelling approach to lifetime costing studies of childhood obesity
35
UKHF has been contracted to undertake the modelling for EU JANPA WP4.
Substantial adaptations to the UKHF’s modelling software will be necessary to accommodate:
• Use of cohort simulation models rather than population simulation models
• Incorporation of children with shorter term impacts into the models
• Use of a societal economic perspective rather than an exclusively health services perspective
• Use of more complex metrics and reporting associated with lifetime costing studies
Reporting
To manage budget, IPH IRL will undertake a number of the routine data collation and reporting tasks
including the calculation of model metrics and production of graphical outputs.
The virtual individuals’ simulated BMI trajectories and their impact and cost experiences will be
summarised by UKHF in Model Output Tables that will be used by IPH IRL to calculate relevant
excess and effect metrics.
Validity and generalisability
Validation studies will address the validity of country-specific findings as well as the effect of
research data and modelling assumptions on model outputs. These include comparisons of model-
based estimates of the RRs of adult diseases and societal impacts associated with childhood obesity
5This presentation is part of the Joint Action JANPA (Grant agreement n°677063) which has received funding from the European Union’s Health Programme (2014-2020)
Life mecostofchildhoodobesitystudies
Obesity-relatedtreatment&deaths
Obesity-relateddiseases
RRs
AdjustedQOLmeasures&costs
Deathsfromothercauses
Adultproduc vitylosses
Costsofadultproduc vitylosses
to the existing estimates in the research literature. Another is an exploration of the methods used to
model lifetime BMI trajectories and the independent disease processes assumption
There is interest in knowing if the JANPA WP4 modelling methodology can be extended to other EU
member countries. Amongst other things, we will compare basic models and advanced models in
JANPA WP4 participating countries in advanced studies and develop a toolbox of modelling
resources for undertaking the modelling in other EU countries.
37
EVIDENCE
Chapters 1 – 4
CHAPTER 1: OUTLINE OF THE EVIDENCE PAPER
1.1. Development The Evidence Paper covers prevalence, health and other impacts, healthcare and other costs,
evidence and experience of socially disadvantaged in the EU as well as the availability and quality of
the data in the countries.
This Evidence Paper is the product of collaboration between the JANPA WP4 Lead Team in the
Institute of Public Health in Ireland (IPH), the Irish national Team working on an advanced study
extended with significant funding from safefood “Lifetime Costs of Childhood Overweight and
Obesity and the National Teams in each of the JANPA WP4 countries. Details of these groups are in
the Contributors section at the beginning of this document.
This Evidence Paper is based on:
1. Three systematic reviews conducted by the Irish Team
2. A Local Materials Survey of the JANPA Wp4 countries
3. A “local” Google Scholar search of extra material
4. Four international systematic review (Irish study)
5. Supplemented by local materials gathered in “Local Material Survey”
6. Summarised into Evidence Paper
7. Feedback form participating countries, consultations with expert groups and ISAC
The backbone of the Evidence Paper are three systematic reviews conducted by the Irish National
Team working on the safefood-funded project:
Prevalence of overweight and obesity in children in Ireland and Northern Ireland
(supplemented by IPH with an international review)
Impacts of childhood/adolescent overweight and obesity in childhood/adolescence
Impacts of childhood/adolescent overweight and obesity in adulthood
In a Local Materials Survey, we gathered evidence from WP4 countries by contacting national teams
with a request to supply any studies on the following topics, whether in English or local language.
1. Prevalence of childhood overweight and obesity
2. Health impacts of childhood overweight or obesity that occur in childhood
3. Other impacts of childhood overweight or obesity that occur in childhood (e.g. parental work
absenteeism, lower school attendance and performance)
4. Health impacts of childhood overweight or obesity that occur in adulthood (e.g. Type 2
diabetes, hypertension, coronary heart disease, stroke, cancers, osteoarthritis)
5. Other impacts of childhood overweight or obesity that occur in adulthood (e.g. work
absenteeism, disability benefit, premature mortality)
39
6. Current and future healthcare costs of childhood overweight or obesity (e.g. hospital
inpatient and outpatient costs, drugs and prescriptions)
7. Current and future other costs of childhood overweight or obesity (e.g. absenteeism, lower
educational attainment)
8. Any local studies or reports that relate to methods used to assess costs of overweight or
obesity.
Advanced countries (Greece, Ireland, Romania and Slovenia) were asked for materials on 3, 5, 7 and
8 while all countries (advanced plus Croatia, Portugal and Italy) were asked for materials on 1, 2, 4
and 6.
IPH IRL then supplemented these local materials with searches in Google Scholar (first 20 pages,
using key words child, obesity, [country name]). In all, this yielded 277 articles, presentations and
reports, covering the following:
Prevalence of child overweight/obesity: 45.5% of all materials
Trends in child overweight/obesity over time: 10.8%
Health impacts of overweight/obesity in childhood: 27.1%
Other impacts of overweight/obesity in childhood: 2.9%
Health impacts of child overweight/obesity in adulthood: 0.0%
Other impacts of child overweight/obesity in adulthood: 0.0%
Healthcare costs of child overweight/obesity: 0.4%
Other costs of child overweight/ obesity: 0.0%
Costing methods: 5.1%
Inequalities in prevalence of child overweight/ obesity: 23.8%
Other topic(s) (e.g. policy brief, adult prevalence): 25.6%.
A total of 277 local materials were identified by this process. Table T1.1 at the end of this chapter
shows the distribution of these materials by country and provides more specific details on these
studies. In summary 127 related to prevalence of childhood obesity and overweight, 32 to trends in
childhood obesity, 75 to childhood impacts, 69 related to inequalities in childhood obesity, 14
related to costing methods, and another 71 to other topics.
1.2 Chapters relevant to the Evidence Paper
The relevant chapters of this document are:
Chapter 2 examines the prevalence of child and adolescent overweight and obesity, and
considers trends over time in prevalence, as well as inequalities in prevalence. It begins with
a brief description of the measurement of overweight and obesity. The chapter draws on
best international evidence as well local evidence (127 sources examining prevalence, 32
examining trends, and 65 on inequalities).
Chapter 3 summarises a systematic review by Queally et al. (2016) which describes the
evidence on impacts of overweight and obesity in childhood/adolescence from 18 published
reviews, along with evidence from 81 ‘local’ sources on this topic.
Chapter 4 draws on a systematic review undertaken by McCarthy et al. (2016b) on the
impacts of child/adolescent overweight and obesity in adulthood, which comprises 13
review papers and 15 individual studies. None of the JANPA participants submitted materials
on this topic.
Chapter 5 summarises a review by Hamilton et al. (in preparation) on lifetime direct and
indirect costs of child/adolescent overweight and obesity, which covers 13 studies. Again,
none of the JANPA participants submitted materials on this topic.
Tables
41
Table T1.1: Areas covered by local materials from JANPA WP4 countries
Country
Topic area
Prevalence of child overweight/ obesity
Trends in child overweight/obesity over time
Health impacts of overweight/ obesity in childhood
Other impacts of overweight/ obesity in childhood
Health impacts of child overweight/ obesity in adulthood
Other impacts of child overweight/ obesity in adulthood
Healthcare costs of child overweight/ obesity
Other costs of child overweight/ obesity
Costing methods
Inequalities in prevalence of child overweight/ obesity
Other topic(s) (e.g. policy brief, adult prevalence, adult impacts)
Total number of discrete sources
Croatia 8 3 9 0 0 0 0 0 0 5 6 24
Greece 34 5 22 4 0 0 0 0 1 24 8 64
Ireland 18 7 6 1 0 0 1 0 6 9 12 44
Italy 16 4 15 0 0 0 0 0 0 9 4 33
Portugal 24 4 6 2 0 0 0 0 4 13 14 50
Romania 13 2 14 1 0 0 0 0 0 4 11 34
Slovenia 14 7 3 0 0 0 0 0 3 1 16 29
All countries 127 32 75 8 0 0 1 0 14 65 71 277
CHAPTER 2: EVIDENCE: PREVALENCE OF OVERWEIGHT AND OBESITY
2.1. Measurement of overweight and obesity
2.1.1. Introduction
Methods for the measurement of overweight and obesity can be categorised as direct or indirect.
Direct measures provide estimates of total fat mass while indirect, or anthropomorphic measures of
adiposity include waist, hip and other girth measurements, skin-fold thickness, and indices derived
from measured height and weight, the most common index being BMI (Body Mass Index) (Lobstein
et al., 2004). Anthropometric measures are cheaper and more convenient to obtain but are also less
accurate. For practical reasons, many surveys use self-reported height and weight as a means to
estimate BMI. This section describes the most commonly-used anthropomorphic measure of
adiposity in children and adolescents: Body Mass Index3.
2.1.2 Waist Circumference and its relationship to BMI
Waist circumference is the circumference of the body half way between the hip bone and the lowest
rib, on normal exhalation and against bare skin (WHO, 2011)4. The World Health Organization (2000,
p. 7) notes that “Obese individuals with excess fat in the intra-abdominal depots are at particular risk
of the adverse health consequences of obesity. Therefore, measurement of waist circumference
provides a simple and practical method of identifying overweight patients at increased risk of
obesity-associated illness due to abdominal fat distribution.”
Although it is the most commonly used measure of overweight and obesity, a main limitation of BMI
is that it does not distinguish between elevated body fat and elevated lean muscle mass (WHO,
2000; Maffeis et al., 2008; Must & Anderson, 2006), and this may be particularly true of adolescent
boys (Demerath et al., 2006). Children’s waist circumference may be a better indicator of total body
fat than BMI (Daniels et al., 2000). A very high correlation (r = .92) between central adiposity and
waist circumference in children has also been found (Taylor et al., 2000). This is of importance, since
some evidence suggests that central adiposity in children is more relevant than BMI to health
outcomes (Freedman et al., 1999; Rodriguez-Rodriguez et al., 2011). For example, Katzmarzyk et al.
(2004) concluded that waist circumference combined with BMI was a better assessment for
cardiovascular disease risk among 5 to 18 year-olds than BMI alone.
Some studies have found a disproportionate increase in intra-abdominal fat distribution in children
and adolescents in recent years, relative to changes in BMI, and this increase may be more marked
among girls than boys (McCarthy et al., 2003; Freedman et al., 2015; Kolle et al., 2009; Visscher et
al., 2015). Therefore, a reliance on BMI alone to monitor trends in the prevalence of overweight and
obesity in children and adolescents may underestimate the scale of the problem.
3 Lobstein et al. (2004), Krebs et al. (2007) and Horan et al. (2015) describe various methods to measure adiposity.
4 In comparing waist circumference measures across studies, note should be taken of where on the body the measure was
taken: some studies take the waist measurement at the umbilicus rather than midway between the lowest rib and the hip bone (Aeberli et al., 2011; McCarthy, 2007).
43
Waist circumference percentiles for children/adolescents have been established in several regions5.
Unlike BMI, there is no internationally-agreed set of cut-points to identify overweight/obesity in
children/adolescents, although various percentile cut-points have been suggested (Aeberli et al.,
2011; Krebs et al., 2007)6. Similarly, there is no internationally agreed set of cut-points for adults,
since populations differ in waist circumference distributions (Messiah et al., 2011)7.
2.1.3. Body Mass Index (BMI) BMI is defined as weight in kilogrammes divided by the square of height in metres. According to the
World Health Organization (2000, p. 7), BMI “…provides the most useful, albeit crude, population-
level measure of obesity. It can be used to estimate the prevalence of obesity within a population
and the risks associated with it.” In adults, cut-points of 25 kg/m2 and 30g/m2 are widely used to
define overweight and obesity with additional sub-classification of obesity into Classes I, II and III
(World Health Organization, 2000) (Table 2.1).
Table 2.1. Adult BMI Categories and Co-morbidity risk categories
Classification BMI Range Risk of Co-morbidities
Underweight <18.50 Low (but risk of other clinical problems increased)
Normal 18.50-24.99 Average
Overweight 25.00-29.99 Increased
Obese Class I 30.00-34.99 Moderate
Obese Class II 35.00-39.99 Severe
Obese Class III > 40.00 Very Severe Source: Adapted from World Health Organization (2000, Table 2.1).
In children and adolescents, BMI must be assessed against a reference-standard that accounts for
the child’s age and sex since, broadly speaking, BMI increases substantially with children’s age and
varies by sex (Cole et al., 2000; Krebs et al., 2007). These reference-standards are developed on the
basis of representative survey samples. Various reference-standards are in use8, for example:
the World Health Organization (WHO) Child Growth Standards which has separate systems
for birth to age 5 years (WHO Multicentre Growth Reference Study Group, 2006) and ages 5
to 19 years (de Onis et al., 2007, 2012);
the International Obesity Task Force (IOTF) BMI cut-points (centile curves linked to adult BMI
values of 25 and 30 for ages 0 to 18 years using data from Brazil, Great Britain, Hong Kong,
the Netherlands, Singapore, and the United States) (Cole et al., 2000, 2007);
5 These include European estimates (based on data from Sweden, Germany, Hungary, Italy, Cyprus, Spain, Belgium, and
Estonia) (Nagy et al., 2014), Bulgaria (Galcheva et al., 2009), Germany (Haas et al., 2011), Greece (Bacopoulou et al., 2015), Italy (Zannolli & Morgese, 1996), the Netherlands (Fredriks et al., 2005), Norway (Brannsether et al., 2011), Poland (Jaworski et al., 2012), Spain (Moreno et al., 1997), Switzerland (Aeberli et al., 2011), Turkey (Mazicioglu et al., 2010), the UK (McCarthy et al., 2001), and the USA (Cook et al., 2009). 6 de Moraes et al.’s (2011) systematic review of abdominal obesity in adolescents identified 18 different sets of cut-points.
7 Cut-points of 94 cm and 102 cm for Caucasian adult males, and 80cm and 102cm in Caucasian adult females, are
frequently used as indicators of increased and substantially increased risk of metabolic complications, respectively (WHO, 2000). Additional sex- and ethnicity-specific cut-points have been proposed (Katzmarzyk et al., 2011). 8 Some authors (e.g. Reilly, 2002) argue in favour of the use of national BMI reference-standards rather than international
ones. However, this prevents comparisons across countries and studies using different, local reference curves (de Onis & Lobstein, 2010).
the US 2000 Centers for Disease Control (US-CDC) growth charts (for boys and girls from
birth to age 2 and from ages 2 to 19 years, based on survey data collected between 1963
and 1995) (Kuczmarski et al., 2002);
Table 2.2. Definitions of child/adolescent overweight and obesity used by the WHO, US-CDC and
IOTF
Organization Definition of Childhood Obesity
World Health Organization (WHO)
WHO Child Growth Standards (birth to age 5) (de Onis et al., 2007)
Obese: Body mass index (BMI) > 3 standard deviations above the WHO growth standard median
Overweight: BMI > 2 standard deviations above the WHO growth standard median
Underweight: BMI > 2 standard deviations below the WHO growth standard median
WHO Reference 2007 (ages 5 to 19) (WHO Multicentre Growth Reference Study Group, 2006)
Obese: Body mass index (BMI) > 2 standard deviations above the WHO growth standard median
Overweight: BMI > 1 standard deviation above the WHO growth standard median
Underweight: BMI > 2 standard deviations below the WHO growth standard median
US Centers for Disease Control and Prevention (US-CDC)
CDC Growth Charts (Kuczmarski et al., 2000)
In children ages 2 to 19, BMI is assessed by age- and sex-specific percentiles:
Obese: BMI >95th percentile
Overweight: BMI > 85th and < 95th percentile
Healthy weight: BMI > 5th and < 85th percentile
Underweight: BMI < 5th percentile
In children from birth to age 2, the CDC uses a modified version of the WHO criteria (Grummer-Strawn et al., 2010)
International Obesity Task Force (IOTF)
Provides international BMI cut-points by age and sex for overweight and obesity for children age 2 to 18 (Cole et al., 2000)
The cut-points correspond to an adult BMI of 25 (overweight) and 30 (obese)
Source: Adapted from http://www.hsph.harvard.edu/obesity-prevention-source/obesity-definition/defining-
childhood-obesity/ (accessed December 16, 2015).
Use of BMI as a measure of total body fat in children is imperfect. Growth charts were not
developed as standards of how healthy children should grow, but are, rather, normative referents
(Krebs et al., 2007; Lobstein et al., 2004). Multiple anthropometric measures (BMI along with others
such as waist circumference and waist-hip ratio) may yield a more accurate measure of total body
fat than BMI alone (Lei et al., 2006). However, the use of BMI is generally supported, not solely
because it is relatively easy to obtain, but also because children’s and adolescents’ BMI is strongly
associated with both total body fat and percentage body fat in both boys and girls (Mei et al., 2002;
Pietrobelli et al., 1998; Coe et al., 2010), and children’s BMI is also highly correlated with other risk
factors for obesity-related adult morbidity (Must et al., 1999; Reilly et al., 2003).
Table 2.2 shows three widely used systems of classifying overweight and obesity among children and
adolescents on the basis of BMI. Each uses slightly different methods for classifying overweight and
45
obesity and each results in somewhat different prevalence estimates (Gonzalez-Casanova et al.,
2013; Shields & Tremblay, 2010). As a general rule, the IOTF cut-points result in more conservative
estimates of overweight and obesity than those of the WHO and US-CDC (Lobstein et al., 2004). In
recent years, the most commonly-used reference system in European studies is that of the IOTF.
Research comparing self-reported with measured height and weight among children and
adolescents suggests that although the two measures tend to be highly correlated overall, self-
reports tended to underestimate BMI, and this bias tended to be larger among girls, and adolescents
with higher BMIs (Aasvee et al., 2015; Gorber et al., 2007). The consequence of this is that
adolescent self-reported BMI is likely to underestimate the ‘true’ prevalence of overweight and
obesity (Sherry et al., 2007). There is evidence that the size of this bias varies across countries
(Gorber et al., 2007; Lobstein, 2015), suggesting a cultural component. A high rate of missing data is
also a problem in self-reports in some studies (and may be negatively related to age; Aasvee et al.,
2015), leading to potential bias in estimates for some sub-groups of interest9. Nonetheless, self-
reported data on height and weight for children and adolescents have value, particularly if they are
the only source of data available for a particular survey or population (Sherry et al., 2007). Because
of the limitations associated with self-reported height and weight, the evidence considered here
focuses on measured BMI.
2.2. International/European evidence
2.2.1. Introduction
This section describes data on the prevalence of overweight and obesity among children and
adolescents on the basis of published international results, with a focus on Europe. First, recent
international prevalence estimates are presented. Then, trends in overweight and obesity in children
over time are examined. Finally, studies on inequalities (sub-group variations) in the prevalence of
overweight and obesity are summarised.
The material for this section does not comprise a systematic review. Rather, it is a summary of
reviews on these topics conducted since 2000. Preference is given to measured rather than self-
reported BMI in reviewing prevalence and trends. Where studies have reported prevalence
estimates using multiple cut-points, results based on IOTF cut-points are described here.
There are numerous sources of data on prevalence of child overweight and obesity. However,
comparing and combining them is difficult and complex due to differences in survey and sampling
methods, representativeness and quality of samples, methods of measurement and classification,
and reporting of results. In 2007, a WHO-Europe report (Branca et al., 2007) concluded that
objectively-measured and valid data on BMI in children and adolescents were lacking for around half
of European countries. Efforts to pool European data (e.g. Cattaneo et al., 2010; Pigeot et al., 2009)
confirmed a need for standardised approaches in assessing and monitoring overweight and obesity
among children. This led to the Childhood Obesity Surveillance Initiative (COSI), beginning in
2007/2008, which surveys children aged 6 to 9 years, providing standardised data to monitor the
9 Some authors have suggested methods to correct for self-report bias (Ellert et al., 2014), but this requires strong linkages
between self-reported and measured BMI, which are not always available.
prevalence of overweight and obesity. There is currently no cross-national surveillance of measured
BMI for preschool-age children or adolescents.
2.2.2. Current child prevalence
In a review of all available prevalence data worldwide (collected between 1987 and 2003), Wang and
Lobstein (2006) estimated the prevalence of overweight and obesity in school-aged children (boys
and girls combined) in Europe10 at 25.5% (IOTF cut-points, 20.1% overweight, and 5.4% obese). This
is similar to their estimate for the Americas (27.7%) and the Eastern Mediterranean region (25.5%),
and much higher than in Africa (1.6%), South East Asia (10.6%), and the West Pacific region (12.0%).
In a large study that estimated the prevalence of and trends in overweight and obesity globally
among children and adults during the period 1980-2013, Ng et al. (2014) reported higher prevalence
of overweight and obesity in 2013 among children and adolescents aged 2 to 19 years in Western
Europe (including Mediterranean countries) (24.2% in boys and 22.0% in girls) than in Eastern
Europe (about 19% in both sexes) and Central Europe (21.3% in boys and 20.3% in girls) (Table 2.3)11.
In JANPA countries shown in Table 2.3, the combined prevalence of overweight and obesity in boys
ranges from 11% (Romania12) to around 33% (Greece, Slovenia). In girls, it ranges from about 20%
(Croatia, Romania) to around 29% (Greece). Table A1 (Appendix 2) shows Ng et al.’s (2014) estimates
of the prevalence of overweight and obesity in 2013 among children and adolescents aged 2 to 19
years in individual European countries.
Table 2.3. Prevalence of overweight and obesity among children and adolescents aged 2 to 19
years, by sex, in JANPA countries and European regions, 2013 from Ng et al. (2014) (IOTF cut-
points)
Region/Country % Boys Overweight % Boys Obese % Girls Overweight % Girls Obese
Croatia 21.9 7.6 14.1 5.6
Greece 23.2 10.5 21.2 7.9
Ireland 19.7 6.9 19.3 7.2
Italy 21.5 8.4 18.1 6.2
Portugal 19.8 8.9 16.5 10.6
Romania 2.4 8.6 14.6 5.7
Slovenia 25.9 7.2 18.7 5.3
Central Europe 13.8 7.5 14.0 6.3
Eastern Europe 11.9 6.8 10.9 2.8
Western Europe 17.0 7.2 15.6 6.4
Source: Ng et al. (2014, Table).
Table 2.4. Estimates of the prevalence of overweight and obesity (IOTF cut-points) among children
aged 2 to 4 years in European countries from the systematic review by Cattaneo et al. (2010)
Country Year of survey Sample % Overweight % Obese
10
Using survey data from the Czech Republic, Finland, France, Germany, Greece (Crete), Iceland, the Netherlands, Poland,
Russia, Serbia, Spain, Sweden, Switzerland, and the UK. 11
Ng et al. (2014) used numerous data sources and various search strategies to arrive at these estimates, and included only nationally representative data; they also applied a correction to self-reported BMI estimates. Their final dataset included 19244 data points for 183 countries (for adults and children combined). Spatiotemporal Gaussian process regression was used to estimate prevalence from 1980-2013. Results on trends from Ng et al. (2014) are discussed later in this chapter. 12
Note that a limited number of data sources were used/available for Romania (Ng et al., 2014, Webtable 7).
47
size 2 years
3 years
4 years
2 years
3 years
4 years
Belgium* 1998–1999 970 4.8 2.2
Cyprus* 2004 647 7.7 2.9
Czech Republic 2001 5456 8.5 8.3 8.2 2.1 2.0 3.7
England 2001–2002 1723 19.6 15.2 15.5 2.3 4.6 5.7
France 2006–2007 191
10.1 13.8
1.3 4.1
Greece 2003–2004 2154 15.1 16.6 16.2 5.8 7.2 11.1
Ireland 2001–2002 1352
20.5
7.0
Italy 2005 1230 10.2 13.5 14.4 3.1 4.5 7.8
Netherlands 2002–2004 1781
12.2
2.8
Northern Ireland 2001–2002 104
19.0
2.0
Poland 2000 139 26.0 4.9 10.4 4.0 12.2 12.5
Portugal 2001 1557
15.4 16.9
5.1 6.2
Romania 2004 1826 9.2 6.8 6.7 4.5 4.6 5.1
Scotland 2003 407 13.5 16.0 15.1 3.3 4.1 4.4
Spain 1998–2000 268 8.9 16.7 24.7 6.3 11.5 7.5
Sweden 2002 183
19.0
6.0
Source: Cattaneo et al. (2010, Table 2 – studies of measured BMI only; studies of reported BMI are not included in this table) *Data were not disaggregated by age. JANPA countries are highlighted.
Cattaneo et al. (2010) synthesized the available data on prevalence of overweight and obesity
among pre-school children (aged 2 to 4 years; prevalence by sex was not reported) in the European
Union13,14. The results of studies with measured BMI are shown in Table 2.4 (including five of the
seven JANPA countries). The combined prevalence of overweight and obesity in 4-year-olds (the age
group for which data are available for all five JANPA countries) ranges from 11.8% (Romania) to a
little over 27% (Greece, Ireland). Cattaneo et al. (2010) concluded that countries in the
Mediterranean region and British Isles had the highest rates of child overweight and obesity, while
countries in central, eastern and northern regions of Europe had lower prevalence rates. It should be
noted that the sample sizes for many of the studies identified by Cattaneo et al. were small.
Del Mar Biblioni et al. (2013) conducted a systematic review of the worldwide prevalence of
overweight and obesity measured by BMI among adolescents (aged 10 to 19 years) for surveys
conducted between 1999 and 201115. Table 2.5 shows estimates for the national studies identified,
including four of the seven JANPA countries (regional estimates are not shown here). Note that
estimates for Italy are based on self-reported data. Prevalence of overweight and obesity combined
in boys ranged from about 18% to 28% and in girls it ranged from about 9% to 25%. There is a fairly
consistent pattern of higher rates of overweight and particularly obesity among boys than girls, with
13
De Onis et al. (2010) have also provided prevalence estimates for overweight and obesity worldwide among preschool children (aged up to 5 years). 14
Cattaneo et al. (2010) identified studies in each of the 27 EU countries combined with a systematic review. Both measured and reported estimates of BMI were included. Data on preschool children’s BMI were not available in Austria, Denmark, Estonia, Finland, Latvia, Luxembourg, Malta, the Slovak Republic or Slovenia at the time of their review. Where multiple studies were available for a country, the one with the most robust sample was selected. 15
Del Mar Biblioni et al.’s (2013) search retrieved 40 studies. They included studies that were based on samples that were nationally or regionally representative and which used definitions of overweight and obesity developed by the US-CDC, IOTF or WHO, and with separate estimates for males and females. The most recent national or regional study was included in preference to older national or regional studies. Lien et al. (2010) have also reviewed the availability of objectively measured height and weight in nationally representative samples of adolescents aged 10 to 18 years and found that data were available for only 18 of 30 countries examined (EU-27 plus Iceland, Norway and Switzerland).
the exceptions of Ireland and Sweden. Higher prevalence is found in Cyprus, Germany, Greece,
Ireland and Portugal.
Table 2.5. Estimates of the prevalence of overweight and obesity in European countries from the
systematic review by del Mar Biblioni et al. (2013)
Source: del Mar Biblioni et al. (2013, Table 2). National estimates only are reported here; the authors also reported
regional estimates. All studies in the table used IOTF cut-points.
Countries taking part in JANPA are highlighted.
De Moraes et al. (2011) conducted a systematic review of the prevalence of abdominal obesity in
adolescents (aged 10 to 19 years) conducted between 2003 and 200916. A majority of the studies
identified were conducted in the USA, Central and South America, with only five European studies
identified. De Moraes et al. confirmed that there is no consensus on the identification of abdominal
obesity among adolescents (and in fact identified 18 different sets of criteria across the studies).
Consequently, prevalence estimates of abdominal obesity range from about one in 10 to one in
three adolescents. Ng et al. (2014) have also commented on the lack of cross-nationally comparable
data on abdominal obesity.
A key data source on the prevalence of child overweight and obesity is the WHO European
Childhood Obesity Surveillance Initiative (COSI), which was established as a response to the
European Ministerial Conference on Counteracting Obesity (2006), when Member States recognised
the need for harmonised surveillance systems, providing measured and comparable data on rates of
overweight/obesity among primary-school children. The aim of COSI is to “…fill the gap in
longitudinal information on anthropometry in primary-school children by routinely measuring their
body weight and body height” (Wijnhoven et al., 2013, p. 80). Countries participating in COSI assess
overweight and obesity in children aged 6-9 years using objective measures, in order to monitor
trends, compare progress with other countries, and inform action to reverse the trend17.
16
A total of 29 studies of the general population that were cross-sectional in design and with measured abdominal obesity, verified by waist circumference, for males and females separately, were included. 17
The number of countries taking part has increased since the first round, conducted in 2007-2008: http://www.euro.who.int/en/health-topics/disease-prevention/nutrition/activities/monitoring-and-surveillance/who-european-childhood-obesity-surveillance-initiative-cosi
Country Year of survey Population Age
% Overweight % Obese
All Boys Girls All Boys Girls
Cyprus 1999-2000 School-based survey 12 to 17 18.9 21.3 16.5 5.8 7.1 4.5
Czech Republic 2005 Lifestyle and obesity study 6 to 17 12.3 16.6 8.0 1.4 1.7 1.0
Germany 2008 CrescNet database 12 to 16 18.2 19.3 17.0 6.2 7.6 4.6
Greece 2003 School-based survey 13 to 19 18.3 23.3 14.0 4.3 6.1 2.7
Italy 2002 HBSC study (self-reported) 11, 13 and 15 15.6 20.9 10.6 2.3 3.5 1.2
Ireland (Rep of) 2003 School-based survey 11 to 16 18.5 17.8 19.2 5.8 5.6 6.1
Ireland (Northern) 2003 School-based survey 11 to 15 18.2 18.5 17.8 5.9 6.0 5.7
Portugal 2008 School-based survey 10 to 18 17.4 17.7 17.0 5.2 5.8 4.6
Sweden 2001 School-based survey 10, 13 and 16 15.8 14.6 16.9 4.4 5.0 3.6
49
To date, the international results of the first two rounds of the COSI data collection have been
published (2007/2008 and 2009/2010) (Wijnhoven et al., 2013, 2014a)18. In both rounds of COSI, the
majority of countries selected nationally representative samples. It should be noted that the
international results for COSI rounds 1 and 2 are unweighted (Wijnhoven et al., 2013, 2014a, 2014b).
Table 2.6 shows the percentages of children classified as overweight or obese on the basis of IOTF
cut-points for round 2 of COSI (2009/2010). Median values for overweight and obese boys were
13.7% and 6.7% respectively, and they were 15.7% and 6.7% respectively for girls. Prevalence of
overweight and obesity exceeded 33% in boys and girls in Greece, Italy and Spain, and was lower,
around 16-18%, in Latvia and Lithuania. Prevalence of overweight and obesity was on average higher
among girls than boys, with the largest gender differences in Belgium, Hungary, Ireland and Norway.
Table A2 (Appendix 2) shows other recent European data sources for the prevalence of child and
adolescent overweight and obesity.
Table 2.6. COSI round 2: Percentages of underweight/healthy weight, overweight and obese boys
and girls, IOTF cut-points
Country
Boys Girls
% Normal/ Underweight
% Overweight
% Obese
% Normal/ Underweight
% Overweight
% Obese
Belgium (Flanders) 84.6 10.3 5.1 79.7 13.5 6.8
Czech Republic 82.5 12.5 5.0 80.7 13.4 5.9
Greece 61.9 24.5 13.6 60.1 25.6 14.3
Hungary 81.1 12.2 6.7 76.1 15.7 8.2
Ireland (Rep of ) 84.3 11.6 4.1 75.1 20.2 4.7
Italy 66.1 22.3 11.6 65.6 23.1 11.3
Latvia 84.1 10.7 5.2 82.2 12.5 5.3
Lithuania 84.0 11.0 5.0 82.3 12.1 5.6
FYR Macedonia 73.8 15.3 10.9 78.5 18.0 3.5
Norway 81.4 13.7 4.9 69.5 20.9 9.6
Portugal 77.3 14.8 7.9 78.8 14.5 6.7
Slovenia 78.9 14.0 7.1 78.5 15.7 6.7
Spain 66.5 22.0 11.5 63.4 26.1 10.5
Median 81.1 13.7 6.7 78.5 15.7 6.7
Source: Wijnhoven et al. (2014a, Table 4). All children are aged 7, except in Latvia and Portugal (age 8) Countries taking part in JANPA are highlighted.
2.2.3. Recent trends in child prevalence
A number of caveats should be borne in mind when interpreting trends in the prevalence of
overweight and obesity:
18
Results for about 169,000 children in 12 countries in round 1 have been published (Wijnhoven et al., 2013). In round 2, results based on data from about 220,000 children in 13 countries have been reported (Wijnhoven et al., 2014a). Nine countries (Belgium (Flanders), Czech Republic, Ireland, Italy, Latvia, Lithuania, Norway, Portugal, and Slovenia) took part in rounds 1 and 2 of COSI. Some countries have published national reports on the results of round 3 of COSI (see Section 2.3). The international report on round 3 of COSI is not yet available (Breda, personal communication, February 19, 2016).
There are several possible reasons for a stabilisation in trends which are difficult to
disentangle, e.g. systematic sampling or non-response bias, issues with smaller sample sizes,
reported rather than measured BMI, and/or time-lagged changes associated with
interventions (Visscher et al., 2015; Olds et al., 2011; Rokholm et al., 2010).
Failure to detect change in mean BMI or prevalence of obesity could be masking large
differences in the right extreme of the distribution (i.e. increases in morbid obesity) (Lissner
et al., 2013)19. This indicates a necessity to examine changes across the entire distribution of
BMI over time, though in reality, trend analyses are largely based on the proportions of the
population classified as underweight or healthy weight, overweight, and obese.
In several countries20, increases in children’s waist-hip ratio have been recorded in the
presence of stable BMI and/or over and above changes in BMI, and this suggests that
reliance on BMI alone may not be sufficient to track trends in obesity. In fact, “…focusing on
trends in waist circumference rather than BMI leads to a less optimistic conclusion: the
public health problem of obesity is still increasing” (Visscher et al., 2015, p. 189).
The monitoring of sub-groups of the population is essential to better understand trends in
prevalence over time, and at present, data and research in this area are rather limited (Olds
et al., 2011; Lien et al., 2010).
The identification of trends is reliant on the method of statistical modelling used, as well as
the survey design. Trend analysis that assumes a linear pattern does not allow the
identification of the changes in prevalence which follow a non-linear or phased sequence, as
has appeared to have occurred in Denmark, for example (Rokholm et al., 2010)21.
Wang and Lobstein (2006) reviewed the evidence on trends in overweight and obesity among
children and adolescents (up to the age of 18) globally on the basis of papers published between
1980 and 200522. Annualised trends in obesity among school-aged children ranged from about +0.1%
to +0.7% across 14 European countries with available data23. Among school-aged children,
annualised trends for overweight and obesity combined tended to be larger, ranging from about
+0.5% to 2.3%24. Exceptions were Poland and Russia, where annualised trends in overweight and
obesity combined were negative. Among infants and preschool children, positive annualised trends
in overweight and obesity (ranging from about +0.1% to +0.9%) were observed in Croatia, Germany,
the Netherlands, FYR Serbia, and the UK. Wang and Lobstein (2006) concluded that the rate of
increase was higher in countries undergoing rapid socio-economic development25.
19
This kind of change has been detected in Swedish children, for example (Ekblom et al., 2004). 20
Including England (McCarthy et al., 2003), Norway (Kolle et al., 2009), Spain (Moreno et al., 1997) and the USA (Freedman et al., 2015). 21
It should be noted, though, that Ng et al.’s (2014) study used analyses that allowed for non-linearity in trends. 22
With data collected between 1970 and 2002; some of the data sources used for these estimates were not based on nationally representative samples. 23
Crete, Czech Republic, England, Finland, France, Germany, Iceland, Northern Ireland, the Netherlands, Poland, FYR Serbia, Spain, Sweden, and Switzerland. 24
In Crete, Czech Republic, England, Finland, France, Germany, Iceland, Northern Ireland, the Netherlands, FYR Serbia, Spain, Sweden, and Switzerland. 25
A study on patterns of the prevalence of overweight and obesity among children and adolescents in the ‘transition countries’ of Central and Eastern Europe where rapid changes starting in the early 1990s (Bodzsar & Zsakai, 2014) supports the link between macroeconomic development and trends in overweight/obesity, in that prevalence of overweight and obese children was similar in those countries whose societies had similar values on economic, nutritional and health indicators.
51
Rokholm et al. (2010) reviewed studies examining trends in overweight and obesity in children and
adolescents published between 1999 and 2010 which contained information on the prevalence of
overweight and obesity for at least two time points26. Results were presented (where available)
according to age group (children aged 2-12 years, adolescents aged 13-18 years, and adults aged 18+
years), sex, region and socio-economic group. Among children in Europe, stabilization/levelling off
or a decrease was observed in 11 countries27. Among European adolescents, stabilization/levelling
off or a decrease was observed in all seven countries included in the review28. Olds et al. (2011) also
reviewed evidence from nine countries with high-quality trend data29 from children age 2 to 19 years
collected between 1995 and 2008. They estimated that the annual rate of change of overweight and
obesity combined was 0.0%. They also found that the rate of change for girls (-0.08%) showed more
of a flattening than for boys (+0.08%) and that flattening was more marked among younger children.
Ng et al. (2014) drew on information from almost 1800 studies across 183 countries in order to
estimate trends in the prevalence of overweight and obesity in children (aged 2 to 19) and adults
(aged 20 and older). Table 2.7 shows their estimates for the percentages of overweight and obese
males and females aged 2 to 19 years across four time points from 1980-2013 for Central, Eastern
and Western Europe. In both genders and across all three regions in Europe there has been an
increase in the prevalence of both overweight and obesity. For example, in Western Europe, the
prevalence of overweight and obesity combined has increased from 19.6% to 24.2% in males, and
from 17.6% to 22.0% in females in the time period studied.
Ng et al. (2014) noted that in developed countries, across adults, the greatest rates of increase
occurred between 1992 and 2002, followed by a slowing down in the rate of increase from 2002
onwards. However, trends in prevalence across cohorts in developed regions indicated that
successive cohorts were gaining weight at all ages, including childhood and adolescence, with more
rapid gains between ages 20-40 years. In developed countries, peak prevalence was moving to
earlier ages with time. Ng et al. further noted that important sub-national variations (for example by
socio-economic group, ethnic or racial group, and urban and rural areas) were not captured in their
analyses. Table A3 and A4 (Appendix 2) show estimates of the prevalence of overweight and obesity
from 1980 to 2013 for each European country separately from Ng et al. (2014).
Section A2.3 (Appendix 2) provides a brief overview of recent trends in adult overweight and
obesity.
Table 2.7. Trends in prevalence of overweight and obesity among children and adolescents aged 2
to 19 years, by sex, in European regions (1980-2013), from Ng et al. (2014) (IOTF cut-points)
Region Males 1980
Males 1990
Males 2000
Males 2013
Females 1980
Females 1990
Females 2000
Females 2013
Overweight
Central Europe 10.7 10.8 12.0 13.8 11.4 11.0 12.0 14.0
26
A total of 52 sources were identified: 30 of these included children, and 14 included adolescents. All age groups and samples (whether regional or national) were included. 27
Denmark, England, France, Greece, the Netherlands, Norway, Russia, Scotland, Spain, Sweden, Switzerland. 28
Denmark, England, France, Iceland, the Netherlands, Sweden and Switzerland. 29
Australia, China, England, France, the Netherlands, New Zealand, Sweden Switzerland and the USA. There is some overlap in the studies examined by Rokholm et al. (2010) and Olds et al. (2011).
Eastern Europe 7.1 8.1 10.2 11.9 9.4 10.6 10.9 12.4
Western Europe 14.1 14.8 16.2 17.0 12.5 13.1 14.7 15.6
Obese
Central Europe 6.5 6.8 7.1 7.5 5.3 5.8 6.1 6.3
Eastern Europe 5.9 7.0 6.5 7.1 5.4 6.3 5.9 6.4
Western Europe 5.5 5.8 6.4 7.2 5.1 5.3 5.8 6.4
Overweight and obese
Central Europe 17.2 17.6 19.1 21.3 16.7 16.8 18.1 20.3
Eastern Europe 13.0 15.1 16.7 19.0 14.8 16.9 16.8 18.8
Western Europe 19.6 20.6 22.6 24.2 17.6 18.4 20.5 22.0
Source: Ng et al. (2014, Webtables 9 and 10).
In summary, in the absence of strong evidence supporting a decrease in the prevalence of
overweight and obesity rates among children and adolescents in Europe, a realistic assessment is
that the situation is not improving over time, and could be worsening, if waist circumference trends
are considered.
2.2.4. Inequalities in child prevalence
This section reviews the available evidence from recent international reviews on this topic, focusing
on socio-economic status and difference between immigrant and non-immigrant groups. It is outside
the scope of this review to consider the reasons for socio-economic disparities in any depth.
Knai et al. (2012, p. 1473) have commented that “A social gradient in overweight runs through
European and other developed countries, with those who are poorest the most likely to be
overweight”. Generally speaking, country-level data patterns indicate that overweight and obesity
prevalence tends to increase with per capita gross domestic product (GDP) up to a certain level30
after which it decreases (Pampel et al., 2012; Pomerleau et al., 2008). Also, countries with the
greatest inequality in wealth have the highest levels of both adult and child overweight and obesity
(Pomerleau et al., 2008; Knai et al., 2012). These between-country disparities are mirrored by socio-
economic disparities within countries: socio-economic status (SES) groups with the greatest access
to energy-dense diets are those with the highest levels of obesity. These tend to be low-SES groups
in developed countries (Due et al., 2009; Wang & Lim, 2012; Robertson et al., 2007). Women and
children in lower socio-economic groups may be more vulnerable than men to developing obesity
(Robertson et al., 2007; Pampel et al., 2012).
Knai et al. (2012) conducted a systematic review of socio-economic disparities in child overweight
and obesity across countries in Europe, and linked their search results with analyses of a country-
level indicator of relative inequality in household income31 and prevalence estimates of overweight
and obesity from the HBSC study (self-reported) and the OECD (measured). They reported a country-
level correlation between self-reported overweight/obesity prevalence and income inequality of .60,
and between measured overweight/obesity prevalence and income inequality of .55. Other evidence
30
Pomerleau et al. (2008) suggest that this level is around 10,000 USD around 2008. 31
This is the gap between the median and the 10th percentile of household incomes for households with children, expressed as a percentage of the median income value.
53
confirms that larger SES disparities are associated with higher overall levels of economic and social
development (Pampel et al., 2012).
Shrewsbury and Wardle (2008) reviewed the evidence from 45 cross-sectional studies conducted
between 1990 and 2005 for individual-level associations between measures of SES and overweight
and obesity in children and adolescents32. Various measures of SES were employed in the studies in
their review (education, occupation, family income, composite measures of SES, and/or
neighbourhood-level SES). Parental education level was the most frequently used across studies and
it was also the most consistently related to overweight/obesity. For the other measures of SES, quite
mixed results were found. Shrewsbury and Wardle (2008) also noted that inverse SES-adiposity
relationships were more consistently found in studies of younger children compared with
adolescents. The treatment of ethnicity varied across the studies reviewed (and the composition of
ethnic groups varies widely across countries). The median odds ratio across all studies examined for
risk of overweight/obesity was 2.04 between the lowest and highest SES groups. Sixteen of the 45
studies examined multivariate relationships between SES and child/adolescent adiposity and a
majority of these indicated an independent, significant association between SES and risk of
overweight or obesity.
Shrewsbury and Wardle (2008, pp. 281-2) commented that “The results may reflect the relative
stability and therefore greater validity of parental education as an indicator of SES, while parental
occupation and income could be more liable to change. … education is more than a marker for
parental overweight and probably exerts an independent effect on adiposity… [and] SES gradients in
adiposity develop early in the life course”. Other evidence has confirmed that SES gradients emerge
as early as age 3 or 4 years (Knai et al., 2012).
Labree et al. (2011) reviewed studies that compared within-country prevalence of overweight and
obesity among children and adolescents of migrant and native origin, and from various ethnic
groups, in Europe. They identified 19 cross-sectional studies conducted in 6 countries33. They found
that in most countries, children and adolescents in the minority groups were at higher risk for
overweight and obesity than their majority group counterparts. There were, however, some
exceptions. Labree et al. (2011) noted that the definition of the ‘migrant’ group varied across
studies. Studies also varied in the number of different sub-groups compared, and no studies for
some countries with both longer immigration histories (e.g. Belgium, Norway and Sweden) and more
recent immigration (e.g. Italy, Portugal and Spain) were identified in this review. Labree et al. were
unable to assess the extent to which ethnic variations could be attributed to socio-economic factors.
If prevalence of overweight and obesity is increasing to a greater extent among low-SES groups
compared with high-SES groups, this would be evident in a widening of SES-related disparities in
prevalence over time. Knai et al. (2012) reviewed studies which examined the extent to which socio-
economic disparities in the prevalence of child overweight and obesity may have changed over time.
32
This is an update of a review conducted previously by Sobal and Stunkard (1989), who examined 144 studies published before 1989 examining SES-overweight/obesity relationships among both children and adults. About a quarter of studies identified by Shrewsbury and Wardle (2008) were from the UK, 15% from each of Germany, the USA and Australia, 8% from Italy, and just one or two studies from each of France, the Netherlands, Belgium, Canada, the Republic of Ireland, Spain, Sweden and Switzerland. 33
Austria (1 study), Denmark (2), Germany (4), Greece (1), the Netherlands (4), and the UK (7) (studies published between 1999 and 2009, and data collected 1991-2007). Fourteen of these studies classified overweight and obesity using the IOTF cut-points.
Only seven European studies met their inclusion criteria34. Four of the seven studies reported a
widening in the social gradient over time, one reported a widening in disparities for only one sub-
group examined, and the remaining two did not find evidence for a change in disparities. None of
the seven reported a significant narrowing of SES-related disparities over time. Data from England
and France reviewed by Rokholm et al. (2010) are consistent with Knai et al.’s review in that they
suggested that the stabilisation of prevalence rates was more pronounced in medium and high
socio-economic groups than in low socio-economic groups.
A limitation with cross-sectional studies is that they only reveal information about relationships
between SES and overweight/obesity at single time-points; also, this may vary by ethnic or racial
groups. One study conducted in the USA (Jones-Smith et al., 2014) illustrates the complexity of this
issue. The study estimated the probability of overweight/obesity from birth until 5 years of age
according to socio-economic status (SES quintiles) for five race/ethnic groups (American
Native/Alaskan Native, African American, Asian, White). In general across race/ethnic groups, the
probability of being classified as overweight/obese increased until around age 2 and then decreased
until age 5. However, the trajectories varied substantially by ethnic group and SES quintile. Hispanic
and Asian children in the USA showed the largest SES disparities in risk of overweight or obesity,
while these were smaller among White children. Results for the African American and Native groups
indicated that risk of overweight and obesity was not consistently negatively related to SES.
Another study from the USA illustrates that it may be informative to track changes in both BMI and
SES over time to better understand the association between the two, since SES is not a fixed entity.
Kendzor et al. (2012) tracked the BMI and household income status by following children from birth
to age 15 years and identified different patterns in both BMI and income over time. They
characterised income trajectories into five patterns (stable low, stable adequate, unstable to low
(downward mobility), low to adequate, and unstable to adequate (the last two indicating upward
mobility)). They found that downwardly mobile children tended to have worse BMI trajectories than
upwardly mobile children. Also, downwardly mobile children and stable low income children had
similar trajectories, as did upwardly mobile and stable adequate income children.
In summary, “a growing body of literature suggests that the SES-obesity association is complex and
varies by several demographic (e.g. age, gender, ethnicity) or environmental (e.g. countries, SES)
factors.” (Wang & Lim, 2012, p. 185). As already noted in Section 2.2.3, several authors have
commented on the need for the development and inclusion of standard measures of both SES and
ethnicity in the surveillance and monitoring of overweight and obesity among children (Cattaneo et
al., 2010; Labree et al., 2011); this is evident in this section also, particularly in monitoring
inequalities in prevalence over time.
2.3. Evidence from JANPA WP 4 countries
2.3.1. Current child prevalence
In this section, estimates of prevalence of childhood overweight and obesity from each of the JANPA
WP4 countries is considered. Across all countries, 127 published estimates of overweight and
obesity in children, based on general populations and covering both regional and national samples
34
Measured BMI, indicators of SES and time trends since 2000 based on multiple cross-sectional data collections. One study was conducted in each of Belgium and Finland; two in France; and the remaining three in the UK.
55
have been published since 2000 (see Appendix 2). This section considers a selection of these
estimates, giving preference to studies with measured BMI, IOTF cut-points, more recent surveys,
and larger and nationally representative samples, where available. Details of each of these studies
are shown in Table A5 (Appendix 2). Unless otherwise stated, the IOTF cut-points have been used.
Estimates by gender for three broad age groups are provided for each country (preschool, primary
school age, adolescents) where available.
2.3.1.1. Croatia
Information on the prevalence of overweight and obesity in Croatia was found in eight sources.
Three are considered here. Sample sizes are generally small and samples are not nationally
representative. A study of children aged 3 to 7 in Osijek conducted in 2011 (Farkas et al., 2015)
estimated that about 24% of boys and 16% of girls were overweight or obese (WHO cut-points).
Among children aged 6-7 from a small nationally representative sample surveyed in 2003-2004,
about 22% of boys and 20% of girls were overweight or obese (Juresa et al., 2012). In Zagreb, a
survey of adolescents aged 15 to 19 resulted in estimates of overweight and obesity of 23% among
boys and 13% among girls (Petranowic et al., 2014). Croatia collected data for COSI for the first time
in the autumn of 2015; these results are not yet published.
2.3.1.2. Greece
Local materials from Greece resulted in 34 published sets of prevalence estimates since 2000. Four
are considered here. A systematic review by Kotanidou et al. (2013) identified 25 papers (covering
31 studies) that assessed prevalence of overweight and obesity in children aged 1 to 12 (using IOTF
cut-points, surveys conducted between 2004 and 2010). Meta-analysis indicated that 10.2% (CI95%:
9.8-10.7%) of Greek children were obese, 23.7% (CI95%: 22.7-24.8%) were overweight and the
combined prevalence of overweight and obesity was approximately 34% (CI95%: 32.7-35.3%).
Analysis by gender showed that 11.0% of boys and 9.7% of girls were obese, while 24.1% of boys and
23.2% of girls were overweight.
Among children aged 1 to 5 (from five regions in Greece surveyed in 2003-2004; Manios et al., 2007),
about 19% of boys and 24% of girls were overweight or obese. In a nationally representative sample
of 8 and 9 year-olds (surveyed in 2007; Tambalis, 2010), similar percentages of boys (39%) and girls
(38%) aged 8 to 9 were overweight or obese. These results are similar to those reported for round 2
of COSI (ages 7 and 9; Wijnhoven et al., 2014a)35. Among adolescents (aged 12 to 19; nationally
representative sample conducted in 2010-2012; Grammatikopoulou et al., 2014), the trend in
gender that was apparent in children aged 1 to 5 was reversed: 37% of boys, and 25% or so of girls,
were overweight or obese. Rates of overweight and obesity in adolescents were highest at ages 12-
14, but among adolescents at all ages, they exceeded 33% in boys and 20% in girls36.
2.3.1.3. Ireland
The literature from Ireland provided 18 sets of estimates of the prevalence of overweight and
obesity. Five are considered here. They come from four sources. The first is the Growing Up in
Ireland (GUI) study, a national longitudinal survey of representative samples of children in two
35
Results for round 3 of COSI are not yet published for Greece. 36
This study also included estimates of abdominal obesity (using cut-points of the International Diabetes Federation, IDF). Across all adolescents, about 9% of boys and 9% of girls were classified as being abdominally obese. Rates of abdominal obesity peaked in boys at age 13 and in girls at age 12, thereafter showing small declines with increasing age.
cohorts, and followed every 2-3 years. The Infant Cohort of about 11,150 children was first surveyed
at age 9 months in 2008-2009, while the Child Cohort of about 8,550 children was first surveyed at
age 9 years in 2007-2008. The second is the third round of COSI (2012). Estimates for adolescents
come from a study on second-level students’ participation in sport (Fahey et al. 2005), while the
most recent estimates, based on data collected in 2013-2014, come from the Fluoride and Caring for
Children’s Teeth (FACCT) study (McCarthy et al., 2016a)37.
Among infants aged 9 months who took part in GUI, on the basis of the UK-WHO growth charts,
24.8% of all children were classified as overweight and 15.7% as obese (Mangan & Zgaga, 2014).
Also based on GUI, at age 9, it was reported that 22% of boys and 30% of girls were overweight or
obese (Layte & McCrory, 2011). The FACCT study indicated that 21% of children aged 4-7 years (18%
of boys and 25% of girls), and 26% of adolescents (aged 11-14 years; 23% of boys and 28% of girls)
were overweight and obese.
The COSI results for Ireland for round 3, conducted in 2012 (Heinen et al., 2014) indicated that
among children aged 7, the prevalence of overweight and obesity was higher in girls (22%) than in
boys (17%). These estimates are lower than those from the FACCT study for children aged 4-7 years
but the gender difference is consistent. At age 9, COSI estimates indicated that prevalence was
similar for girls (22%) and boys (20%).
Fahey et al. (2005) surveyed a representative sample of adolescents aged 13 to 18 in 2004 and
estimated that about one in five (19.9% of boys and 20.4% of girls) was overweight or obese. The
pattern of prevalence followed a U-shape with age, being lowest among adolescents aged 15 and 16.
2.3.1.4. Italy
Estimates from Italy were retrieved from 16 sources. Five are described here. Turchetta et al.’s
(2012) systematic review of prevalence of overweight and obesity among children in Italy (aged 6-
11, studies published since 2000), as well as Italy’s child obesity surveillance system, OKkio alla
SALUTE (e.g. Spinelli et al., 2014, 2015; Nardone et al., 2015) confirm large regional variation in the
prevalence of overweight and obesity among children in Italy. The highest prevalence is found in the
South, and the lowest in the North of Italy. With the exception of OKkio alla SALUTE, studies of
measured BMI have tended to examine prevalence in specific regions, which means that arriving at
national prevalence estimates for preschool children and adolescents is difficult.
There are no nationally representative estimates of overweight and obesity among preschool
children. However, a study of two regions of children aged 2 to 6 from Northeast and Southern Italy
(Verona and Messina, surveyed in 2002), overweight and obesity prevalence was estimated at about
23% in boys and 28% in girls (Maffeis et al., 2006).
The most recent results from OKkio alla SALUTE, which informs COSI (collected in 2014; Spinelli et
al., 2015; Nardone et al., 2015) indicate that about 31% of all children aged 8 and 9 were overweight
and obese. Gender differences were not apparent, with 30-31% of both boys and girls classified as
overweight or obese.
37
The FACCT study estimates were obtained by McCarthy et al. (2016a) from the Oral Health Services Research Centre.
57
Italy does not have national data for prevalence of measured overweight and obesity in adolescents.
The only nationally representative data are self-reported, from the HBSC study (Cavallo et al., 2013;
Lazzeri et al., 2014). However, some regional studies provide estimates. In a study of children and
adolescents in North-Central Italy (surveyed in 2006; Lazzeri et al., 2008), marked gender differences
were apparent, and these increased with age. Prevalence of overweight and obesity combined in
boys at ages 11, 13 and 15, respectively, was 23%, 23% and 28%, compared with 16%, 13% and 12%
respectively, in girls. The prevalence of overweight and obesity in adolescents has also been
examined in three regions in Central Italy (surveyed 1993-2001; Celi et al., 2003). Prevalence of
overweight and obesity at ages 11, 13 and 15 was 35%, 29% and 25% in boys, respectively, and 31%,
28% and 21% in girls, respectively.
2.3.1.5. Portugal
Sources from Portugal included 24 sets of estimates of the prevalence of overweight and obesity;
four are described here. In the EPACI Portugal study, conducted in 2012, 31.4% of infants aged 12-36
months were overweight and 6.5% obese (WHO reference charts; Nazareth, 2013). Children aged 3
to 6 from Porto were surveyed in 2008-2009; of these, 37% of boys and 30% of girls were overweight
or obese (Vale et al., 2011). Rito and Graça (2015) reported the results for round 3 of COSI in
Portugal, conducted in 2013. They found that similar percentages of boys and girls aged 6 were
overweight or obese (21% and 22% respectively), while significantly more girls (27%) than boys
(24%) aged 7 were overweight or obese. In contrast, at age 8, significantly more boys (30%) than
girls (27%) were overweight or obese. Across all children in round 3 of COSI, 25.0% were overweight
and 8.2% were obese. Sardinha et al. (2011) surveyed a nationally representative sample of children
and adolescents aged 10 to 18 years from mainland Portugal in 2008 and reported that across all
ages, about 24% of boys and 22% of girls were overweight or obese. Generally, prevalence
decreased from ages 10 to 18, from 32% to 18% in boys and 28% to 16% in girls.
2.3.1.6. Romania
Data on prevalence were available from 13 studies identified in Romania. Seven are described here.
Note that reference standards to classify overweight and obese varies across studies.
In 2010, Romania’s national nutrition programme collected data from a representative sample of
infants aged 0-24 months. Nanu et al. (2011) reported that 5.4% of the infants assessed had high
weight for height (WHO growth standards). The results of a survey conducted in 2010-2012 in 14
counties of Romania indicated that 20% of children aged 6-7 years were overweight or obese, and
that 18% of children aged 13-14 were overweight or obese (WHO cut-points; Ardeleanu et al., 2015).
The results for the third round of COSI for Romania have been published in a national report
(Nicolescu et al., 2013) and discussed at the eighth international meeting for COSI (World Health
Organisation Regional Office for Europe, 2016). Among the 8-9 year-olds surveyed, prevalence of
overweight and obesity was higher among boys than girls: 14.8% of boys were classified as
overweight, and 15.0% obese, compared with 15.7% and 8.0% among girls, respectively (IOTF cut-
points).
The other four Romanian studies were conducted using regional samples. Mocanu (2013) reported
the results of a study carried out in Northeast Romania among children aged 6 to 10. Across all
children, about 25% of boys and 23% of girls were overweight or obese (IOTF cut-points). No
significant gender differences were found for any age, and prevalence was highest at age 9 in both
sexes. Valean et al. (2009) surveyed children in Northwest Romania and estimated, using the US-CDC
cut-points, that 25% of boys and 17% of girls aged 6-18 were overweight or obese. At all age groups,
there were significantly more overweight and obese boys than girls. There was a large decline in
prevalence with increasing age. For example, 34% of boys in grades 1-4, compared to 16% of boys in
grades 9-12, were overweight or obese. Chirita-Emandi et al. (2012) estimated that, in West
Romania, 30% of boys and 22% of girls aged 6-17 were overweight or obese (IOTF cut-points).
Estimates from South Romania (Bucharest) in children aged 7-19 are similar to those for the West
(IOTF cut-points), with prevalence of overweight and obesity around 29% among boys and 21%
among girls (Barbu et al., 2015).
2.3.1.7. Slovenia
Fourteen sources on the prevalence of child overweight and obesity were retrieved for Slovenia.
Three are described here. Slovenia is noteworthy in that it has a national monitoring and
surveillance system, SLOfit, which has collected data on children’s measured BMI, triceps skinfold,
and a battery of eight motor tests annually since 1987. It covers all of the population aged 6 to 19
years with sample sizes ranging from about 180,000-190,000 in recent years (Starc, personal
communication, February 17, 2016). Participation rates are generally a little over 90% of the
population. Based on SLOfit data from 201438 (SLOfit database 1989-2015, Laboratory for the
Analysis of Somatic and Motor Development, Faculty of Sport, University of Ljubljana), 26.4% of boys
and 22.2% of girls aged 7 to 18 were classified as overweight or obese. Gender differences became
apparent at around age 11 onwards (with higher prevalence among boys). Prevalence of overweight
and obesity peaked among girls at ages 9-10 (with 26-27% overweight or obese), and among boys at
ages 11-12 (with about 30% overweight or obese), generally decreasing thereafter. Prevalence
estimates from Kovac et al. (2012), also based on SLOfit data from 2011, show similar results to
those for 2014.
Among younger children in Slovenia, Sedej et al. (2014) estimated, on the basis of a representative
sample of 5 year-olds, that 17% of boys and a little over 21% of girls were overweight or obese.
Round 2 of the COSI study indicated that overweight and obesity combined ranged from 17% to 26%
in boys aged 6 to 9, and from 18% to 29% in boys aged 6 to 9, with prevalence increasing with age
(Wijnhoven et al., 2014a). The increase of prevalence with age in COSI is consistent with the SLOfit
results. Results for Slovenia for the third round of COSI are not yet published.
2.3.2. Recent trends in child prevalence
In this section, a description of recent trends in child prevalence of overweight and obesity is
provided. Across all countries, 32 analyses of trends in prevalence, based on general populations,
covering both regional and national samples, have been published since 2000 (see Appendix 2).
2.3.2.1. Croatia
Three published sources on trends in the prevalence of overweight and obesity in children were
retrieved for Croatia; none covers nationally representative samples.
38
With thanks to Dr Gregor Starc for providing the data.
59
Bralic et al. (2011) analysed trends in the prevalence of overweight and obesity among children aged
7 in Splitsko-Dalmatinska county on the basis of data collected in 1991, 1999 and 2008 (measured
BMI and IOTF cut-points). Between 1991 and 2008, prevalence of overweight in boys increased from
10.3% to 15.1%, and in girls, it increased from 7.4% to 19.0%. Obesity also increased significantly in
this time period, from 4.3% to 6.2% in boys and from 4.3% to 8.6% in girls. The rate of increase in
BMI was substantial between 1991 and 1999 and showed a trend towards levelling off from 1999 to
2008. In interpreting these results, Bralic et al. noted that the area in which the study was
undertaken was not directly hit by the war of independence (1991-1995).
Aberle et al. (2009) examined trends among four-year-old children in Slavonski Brod county by
comparing data for the years 1985 and 2005. BMI measures were not available for 1985, but
comparing height and weight of children in the two years, the authors reported that four-year-olds
in this region were 4 to 5cm shorter and about 500g lighter in 2005 than in 1985. Aberle et al. (2009)
discussed these changes in terms of the economic and demographic changes in the region
associated with the war of independence.
Petranovic et al. (2014) examined trends in prevalence of overweight and obesity among
adolescents aged 15 to 19 in Zagreb assessed in 1991, 1997 and 2010. The percentages of boys with
BMI at or above the 85th percentile on the US-CDC growth curves for 1991, 1997 and 2010,
respectively, were 6.5%, 4.8% and 13.3%, while in girls, they were 11.8%, 10.3% and 23.2%,
respectively. Petranovic et al. (2014) discussed these findings in the wider context of socio-economic
and political changes occurring at the time of the surveys.
In summary, there is insufficient evidence in Croatia to establish a clear picture of trends in
overweight and obesity in children. The Croatian war of independence will have impacted on
children’s nutritional status, with variations in the scale and kind of impact varying by region.
2.3.2.2. Greece
Five sources examining trends in prevalence were located for Greece (Papadimitriou et al., 2006;
Tambalis et al., 2010; Roditis et al., 2009; Kotanidou et al., 2013; Kleanthous et al., 2016; the
narrative review by Roditis et al. is not considered in detail here). One of these sources (Kotanidou et
al., 2013) is a systematic review of prevalence, which indicated, on the basis of 25 papers (31 sets of
estimates, all using the IOTF cut-points), that increases in prevalence between 2001-2003 have been
followed by a period of stabilisation from 2003-2010. The results of analyses by Tambalis et al.
(2010), which examined trends in overweight and obesity among 8 and 9 year-old children from
nationally representative surveys conducted annually between 1997 and 2007 are partially
consistent with those of Kotanidou et al. (2013). Tambalis et al. (2010) reported a stabilising in
prevalence of obesity (IOTF cut-points) among both boys (at around 12.2-12.3%) and girls (around
11.2-11.3%) from 2004 to 2007. However, prevalence of overweight showed an increasing trend
during the same time period (from 21.2% to 26.5% in boys and 22.1% to 26.7% in girls).
Trends in the Attica region have been examined in two papers. Papadimitriou et al. (2006) reported
increases in the prevalence of overweight and obesity among children aged 6 to 11 between 1994
and 2005, but these were not statistically significant. Kleanthous et al. (2016) compared the
prevalence of overweight and obesity among children in Grades 1, 4, 7 and 10 in 2009 and 2012
(using the IOTF criteria) and found that rates of overweight and obesity declined significantly in both
boys and girls over the 2.5 year time period.
In summary, trends in the prevalence of child overweight and obesity in Greece show rapid increases
during the 1990s and early 2000s, followed by a slowing down of increases in prevalence, and some
evidence of stabilisation, since around 2004.
2.3.2.3. Ireland
In Ireland, seven studies examining trends in the prevalence of overweight and obesity in children
have been published. One of these is a systematic review of prevalence and trends among children
aged 4 to 12 (Keane et al., 2014), which was updated by McCarthy et al. (2016a)39. Together, these
papers cover trends from 2002-2014. Keane et al. (2014) examined 16 sets of estimates from 15
papers published between 2002 and 2012 and using the IOTF cut-points; 6 of these were from
national samples, while 10 were from regional samples. Analyses indicated no significant trend in
the prevalence of overweight over time, and a borderline significant downward trend in the
prevalence of obesity. McCarthy et al. (2016a) reported a significant downward in obesity rates
among girls (p=0.02) and among boys (p=0.04). Trends in the prevalence of overweight and obesity
were also assessed for over time for younger children (aged 4-7 years) and older children (aged 8-14
years). No significant trends were observed in these sub-groups.
Analyses of rounds 1, 2 and 3 of the COSI data for Ireland (Heinen et al., 2014) support the findings
of Keane et al.’s (2014) and McCarthy et al.’s (2016a) systematic reviews, in that a small but
statistically significant downward trend was found in the prevalence of overweight and obesity in
both boys and girls at age 7 across successive rounds of COSI; however, no differences were found in
prevalence among 9 year-olds across successive rounds of COSI. Barron et al. (2009) compared
results from surveys of children aged 4 to 13 conducted in 2002 and 2007 and concluded that
prevalence in 2007 was the same as 2002.
Examining data from 1990 to 2005, O’Neill et al. (2007) reported a significant increase in the
percentage of overweight and obese children aged 8 to 12, from 12.2% to 22.1% (IOTF cut-points).
Similarly, Walton et al. (2014) reported significant increases in the mean BMI of children aged 8-12 in
comparisons of surveys conducted in 1988-1989 and 2005-2006. Going back even further, Perry et
al. (2009) have reported a dramatic increase in the BMI of 14-year-old children between 1948 and
2002 (from 17-18 to 21-22).
In summary, the evidence from Ireland indicates that there has been a rapid increase in rates of
overweight and obesity during the 1990s and up to around 2002, with some evidence of a stabilising
in trends, and indications of small decreases in prevalence among children, since 2002.
39
The review was completed by Laura McCarthy, Eimear Keane, Fiona Geaney, Maura O’Sullivan and Ivan J Perry. It forms
Deliverable D1 of the safefood project Lifetime Costs of Childhood Overweight and Obesity which covers the Republic of Ireland and Northern Ireland and which in turn helps to inform WP4 of JANPA for the Republic of Ireland. Principal Investigators of the safefood study are Dr Fiona Geaney and Professor Ivan J Perry, Department of Epidemiology & Public Health, University College Cork
61
2.3.2.4. Italy
Information on trends in prevalence of overweight and obesity in children comes from two sources:
OKkio alla SALUTE (which informs COSI), where prevalence among 8 and 9 year-olds in 2008, 2010,
2012 and 2014 have been compared (Spinelli et al., 2015), and a study examining trends in
overweight and obesity among children and adolescents in Tuscany (Lazzeri et al., 2015).
Spinelli et al. (2015) reported small but statistically significant decreases between 2008 and 2014 in
the prevalence of both overweight (from 23.2% to 20.9%) and obesity (from 12.0% to 9.8%) among 8
and 9 year-olds in the nationally representative OKkio alla SALUTE study (IOTF criteria). In the study
of children and adolescents in Tuscany (2002-2006), results were mixed, depending on the age-
group considered. Prevalence of overweight and obesity decreased only among 11 year-olds. Small
increases were recorded in children aged 9 (from 31.7% to 33.4%) and 13 (from 16.8% to 17.9%),
with the largest increases found among 15 year-olds (from 13.3% to 19.7%) (IOTF cut-points).
In summary, there is evidence in Italy for a small decrease in the prevalence of both overweight and
obesity rates among children since 2008. Evidence for trends in adolescent overweight and obesity
are limited and provide mixed results.
2.3.2.5. Portugal
Data on trends in prevalence for Portugal come primarily from COSI (Rito et al., 2012a, Rito & Graça,
2015), with earlier time trends from two papers by Padez and colleagues (2004, 2006).
Comparing the results from COSI rounds 1, 2 and 3, Rito et al. (2012a) and Rito and Graça (2015)
reported a small decrease in the prevalence of obesity among children aged 6 to 8, from 8.9% in
2008 to 8.2% in 2013 (IOTF criteria). A decrease in the prevalence of overweight, from 28.1% to
26.3%, was also found; however, neither of these slight decreases were not statistically significant.
Padez et al. (2004) have examined trends among children aged 7-9 using data collected in 1970,
1992 and 2002. They found statistically significant increases in BMI from 1970 to 1992 and also from
1992 to 2002. What is noteworthy about the trends is that the rate of increase in BMI was about the
same or even higher among children in the latter 10-year period (1992-2002) as in the earlier 22-
year period (1970-2002). Padez (2006) also examined trends in prevalence among Portuguese
conscripts (males aged 18) from 1986-2000. She reported a doubling in the prevalence of overweight
(from 10.5% to 21.3%; IOTF cut-points) and a fourfold increase in the prevalence of obesity (from
0.9% to 4.2%).
In summary, prevalence of overweight and obesity among children and adolescents in Portugal
accelerated quite rapidly from the 1970s to early 1990s, and continued to increase albeit at a slower
rate, with some evidence of a levelling off in prevalence rates from 2008 onwards.
2.3.2.6. Romania
There are two sources of information on trends in overweight and obesity among infants and
children from Romania. Nanu et al. (2011) reported a slight increase in prevalence of increased
weight for height in infants aged 0-24 months in 2010 compared with 2004 (5.4% compared with
4.2%). Rusescu (2006) estimated that 4.2% of children under the age of 5 years were overweight or
obese in 2005, and commented that this shows a favourable trend compared with 1998, when about
10% of children in this age group were overweight.
In contrast to other JANPA participants, this research has also highlighted the relatively high
prevalence of underweight children, noting relatively stable trends in underweight over time. For
example, Rusescu (2006) reported a median birth weight of 3,200g in Romania in 2005, which is
below that of other European countries (3,400g). This study also reported that 4.4% of children
under 5 had low weight for height, while about 5% of children aged 6-7 had low weight for height.
In summary, evidence on trends in overweight and obesity among children in Romania is limited,
with no clear pattern emerging.
2.3.2.7. Slovenia
As described in Section 2.3.1.7, Slovenia’s national monitoring and surveillance system, SLOfit,
provides annual data on the BMI of children aged 7-18 years since 1987. Several papers have been
published examining trends on the basis of the SLOfit data (Kovac et al., 2008, 2012, 2014; Leskosek
et al., 2010). For example, Kovac et al. (2012) examined trends from 1991-2011 on the basis of these
data. They found that the percentage of overweight (IOTF cut-points) increased substantially from
1991 to 2011, from 13.3% to 19.9% in boys and from 12.0% to 17.2% in girls. Prevalence of obesity
also rose even more dramatically, from 2.7% to 7.5% in boys and from 2.1% to 5.5% in girls. Based
on SLOfit data from 2010-201540 (SLOfit database 1989-2015, Laboratory for the Analysis of Somatic
and Motor Development, Faculty of Sport, University of Ljubljana), rates of overweight and obesity
(IOTF criteria) have remained quite stable in recent years with some evidence of a decline in
overweight and obesity since around 2010. During these years, between 19.0% and 20.4% of boys
aged 7 to 18 were overweight, and between 6.9% and 7.5% were obese. Among girls aged 7 to 18
during these years, between 16.6% and 17.8% were overweight, and 6.1% to 6.6% were obese.
Sedej et al. (2014) examined trends in prevalence among 5 year-olds in 2001, 2003-2005 and 2009
and found that rates of overweight and obesity were stable among both boys and girls during this
time period.
In summary, recent trends among both children and adolescents suggest a stabilisation in the
prevalence of overweight and obesity in Slovenia, along with some evidence of a slight decrease
since around 2010.
2.3.3. Inequalities in child prevalence
The evidence from local materials in JANPA WP4 countries is difficult to summarise, since theoretical
and analytical frameworks, as well as national priorities and patterns of variation, sampling and
survey designs, differ widely. Broadly speaking, the evidence can be divided into two categories:
inequalities arising from socio-economic/demographic and regional sources, and inequalities arising
from health-related behaviours. The latter may be confounded with the former. Also, some of the
studies examined variations in prevalence within a multivariate analysis, while others did not.
40
Kindly provided by Dr Gregor Starc.
63
A total of 65 sources examined inequalities in prevalence of overweight and obesity among children.
Across all countries, 36 included socio-economic characteristics (e.g. parental education), 25
examined demographic differences (e.g. country of birth, number of children in the family), 23
looked at regional variations, 29 included parental BMI, 8 examined breastfeeding, 28 included child-
related health behaviours (such as diet and physical activity) and 8 examined parent-related health
behaviours (such as maternal smoking during pregnancy). These are summarised in Table A6
(Appendix 2). A brief commentary for each JANPA participant is provided below.
2.3.3.1. Croatia
Five studies on inequalities were found for Croatia. Two of these confirmed associations between
parental and child BMI (Bralic et al., 2005; Petricevic et al., 2012), a further two indicated the
protective effects of breastfeeding (Mandic et al. 2011; Skledar & Milosevic, 2015), and the fifth
found a higher prevalence of overweight and obesity in children of lower birth order, and in families
with fewer children and lower levels of parental education (Juresa et al., 2012).
2.3.3.2. Greece
Twenty-four studies on this topic were found for Greece. Consistent findings emerged for higher
prevalence of overweight and obesity associated with lower parental education, boys, Greek-born
(rather than foreign-born) children, higher parental BMI, lower dietary quality, skipping breakfast,
less frequent meals, lower levels of physical activity, and higher levels of sedentary activity. Results
concerning variations in prevalence by rural and urban areas are not entirely consistent: this may be
due to the regional nature of some samples and/or more complex associations between local
environment and overweight/obesity. For example, Farajian et al. (2011) found that some aspects of
children’s diet were associated with rates of overweight/obesity, while dietary quality in turn varied
depending on urban/rural environment. Chalkias et al.’s (2013) analyses of children’s environments
indicated that areas characterised by low education and income levels, high population densities and
limited recreation facilities were associated with higher prevalence.
2.3.3.3. Ireland
Nine studies from Ireland that examined inequalities in prevalence were retrieved. One (Williams et
al., 2013) confirmed the presence of a socio-economic gradient at age 3 years, while another study
examining weight gain from birth to three years showed that lower SES was associated with lower
birth weights and highest gains in weight; higher gains in weight were associated with higher
maternal weight gain during pregnancy and no breastfeeding (Layte & Biesma-Blanco, 2014).
Multivariate analyses of children’s BMI at age 9 (Layte & McCrory, 2011; Keane et al., 2012; Perry et
al., 2015) indicated higher prevalence of overweight and obesity among girls, one parent families,
lower occupational class, lower parental education, lower rates of physical activity, poorer dietary
quality, and, in particular, among children with overweight or obese parents. Walsh and Cullinan
(2015) conducted an analysis of the relative contributions of a range of child and parent
characteristics to the socio-economic gradient at age 9 and found that parental characteristics
accounted for a large majority of this gradient, while child-related measures were not statistically
significant. Other studies confirmed an association between socio-economic deprivation and child
overweight/obesity (Heinen et al., 2014; O’Shea et al., 2014). One study (Fahey et al., 2005) did not
find a significant association between SES and rates of overweight or obesity among adolescents.
However, Fahey et al.’s (2005) analysis was bivariate and used a rather broad measure of SES
(parental occupation, split into 9 groups by sector).
2.3.3.4. Italy
Nine studies from Italy examined inequalities. There is strong and consistent evidence for regional
variation in the prevalence of overweight and obesity. Prevalence in the south is much higher than in
the north. Generally, prevalence tends to be highest in Campania and lowest in Bolzano (e.g. Spinelli
et al., 2014). Binkin et al. (2008) reported that this regional variation is not accounted for by
differences in levels of maternal education or employment, nor have any variations by urban/rural
location been reported (Spinelli et al., 2009, 2012). There is some evidence of higher prevalence of
overweight and obesity among boys, though this is not entirely consistent (Spinelli et al. 2009, 2012;
Lombardo et al., 2014). Research from Italy also confirms associations between parental education
and parental BMI and child prevalence of overweight and obesity (as well as rates of weight gain in
children over time: Lombardo et al., 2014; Nardone et al., 2015; Lazzeri et al., 2014; Valerio et al.,
2013). Prevalence is also higher among Italian-born than foreign-born children (Spinelli et al., 2012).
2.3.3.5. Portugal
Thirteen studies from Portugal described inequalities in prevalence. Some regional variations in the
prevalence of overweight and obesity have been reported for Portugal, but these are not as
dramatic as in Italy. Nazareth (2013) reported the highest levels of overweight and obesity among
infants in the North, with lowest rates in the Algarve in the South. Recent COSI results indicate that
prevalence of overweight is higher in Lisbon, the Tagus Valley and the Azores, while rates of obesity
were higher in central regions and Madeira (Rito et al., 2012b, Rito & Graça, 2015). Differences by
urban-rural location were not evident in the COSI results (Rito et al., 2012a). Research by Nogueira
et al. (2013) and Ferrao et al. (2013) suggest that inequalities in prevalence may be related to
specific aspects of children’s local communities, such as poorer built environments and less safe
neighbourhoods. Other studies from Portugal confirm associations between parental BMI, parental
education and children’s overweight and obesity (Bingham et al., 2013, Ramos et al., 2007; Padez et
al., 2005, 2009; Ferreira & Marques-Vidal, 2008), as well as among children who were first-born,
with fewer siblings, and whose mothers gained more weight during pregnancy (Moreira et al., 2007).
2.3.3.6. Romania
Four sources on this topic were located for Romania. Unlike other JANPA countries, a positive
association between SES (income) and prevalence of overweight/obesity has been reported in
Romania, after adjusting for aspects of children’s diets and physical and sedentary activity levels
(Mocanu, 2013). The COSI round 3 results for Romania (World Health Organization Regional Office
for Europe, 2016; Nicolescu et al., 2013) indicate quite large regional variations in the prevalence of
overweight and obesity among children in rural (21.6% overweight/obese), semi-urban (25.0%) and
urban (31.6%) areas. Cosoveanu (2011) reported higher rates of overweight and obesity among
children with higher parental BMI, who were not breastfed and introduced early to solid foods, and
with less healthy diets and less active lifestyles. Morea and Miu (2013) found a positive association
65
between childhood overweight and obesity and parental BMI (measured as pre-pregnancy maternal
overweight or obesity).
2.3.3.7. Slovenia
In Slovenia, inequalities in child overweight and obesity have not been studied extensively. Starc
(2014) reported regional variations in the prevalence of overweight and obesity among children and
adolescents. Rates were highest in Pomurska and Zasavska (two of Slovenia’s 12 regions). Starc
commented that reasons for these regional variations could be related to socio-economic,
educational or environmental differences, but these have not yet been analysed.
Tables Table T2.1. Prevalence of overweight and obesity among children and adolescents aged 2 to 19
years, by sex, in European countries, 2013 (IOTF classification)
Region/Country Males Overweight Males Obese Females Overweight Females Obese
Albania 21.3 11.5 13.9 12.8
Andorra 6.6 9.3 8.9 9.5
Austria 8.6 10.3 8.5 7.8
Belarus 11.6 3.8 13.2 4.2
Belgium 15.9 4.6 14.6 4.2
Bosnia and Herzegovina 7.1 10.1 11.1 11.6
Bulgaria 19.8 6.9 19.0 6.7
Croatia 21.9 7.6 14.1 5.6
Cyprus 17.7 8.0 15.1 7.4
Czech Republic 15.9 6.4 13.2 4.8
Denmark 11.0 8.7 13.5 5.9
Estonia 16.7 7.3 13.8 7.6
Finland 16.8 9.2 14.5 6.6
France 14.1 5.8 11.3 4.7
Germany 15.0 5.5 14.1 5.3
Greece 23.2 10.5 21.2 7.9
Hungary 22.3 7.9 18.8 6.1
Iceland 16.8 9.6 15.4 7.6
Ireland 19.7 6.9 19.3 7.2
Israel 17.1 13.9 15.3 11.3
Italy 21.5 8.4 18.1 6.2
Latvia 15.1 4.8 11.8 3.4
Lithuania 18.0 6.3 15.9 5.2
Luxembourg 18.2 11.1 4.2 13.5
Macedonia 15.1 8.6 16.9 5.4
Malta 21.1 12.5 17.4 7.9
Moldova 10.2 5.6 9.9 5.3
Montenegro 16.9 9.4 19.0 8.3
Netherlands 14.2 4.1 12.3 3.8
Norway 15.0 5.1 12.0 4.0
Poland 15.0 6.9 11.8 6.0
Portugal 19.8 8.9 16.5 10.6
Romania 2.4 8.6 14.6 5.7
Russia 14.4 7.3 12.0 6.6
Serbia 12.5 6.7 16.2 6.9
Slovakia 15.1 5.5 8.0 5.5
Slovenia 25.9 7.2 18.7 5.3
Spain 19.2 8.4 16.2 7.6
Sweden 16.1 4.3 15.3 4.0
Switzerland 14.1 6.6 10.7 5.5
UK 18.7 7.4 21.1 8.1
Ukraine 3.3 7.3 13.6 6.5
Central Europe 13.8 7.5 14.0 6.3
Eastern Europe 11.9 6.8 10.9 2.8
Western Europe 17.0 7.2 15.6 6.4
Source: Ng et al. (2014, Table).
JANPA WP4 countries are highlighted. Other major regions (bottom of table) are shown as comparisons.
67
Table T2.2. Trends in overweight and obesity (IOTF classification) for males and females aged 2-19
years, Europe, 1980-2013
Country/Region
Males Females
1980 1990 2000 2013 1980 1990 2000 2013
Albania 31.5 32.3 33.7 32.8 28.5 29.0 29.0 26.7
Andorra 13.5 14.3 15.0 15.9 14.8 15.7 17.5 18.4
Austria 16.7 18.6 19.6 18.9 14.3 16.1 17.2 16.3
Belarus 12.3 13.7 13.6 15.4 14.3 15.8 15.7 17.4
Belgium 19.1 20.0 19.7 20.5 16.7 17.8 18.5 18.8
Bosnia and Herzegovina 11.2 12.1 14.6 17.2 15.5 16.5 19.4 22.7
Bulgaria 33.1 29.5 25.6 26.7 32.5 29.2 25.4 25.7
Croatia 22.9 24.6 27.0 29.5 11.6 12.4 15.1 19.7
Cyprus 21.7 24.0 24.6 25.7 18.6 20.6 21.7 22.5
Czech Republic 20.3 21.1 21.7 22.3 18.2 18.8 18.5 18.0
Denmark 17.5 18.9 20.2 19.7 13.7 14.9 17.7 19.4
Estonia 14.0 15.0 20.2 24.0 19.5 20.0 20.2 21.4
Finland 20.8 21.3 22.5 26.0 16.7 17.2 18.3 21.1
France 17.8 19.7 21.2 19.9 14.3 15.7 16.7 16.0
Germany 16.6 16.9 18.8 20.5 15.2 15.8 18.0 19.4
Greece 20.4 23.4 28.6 33.7 19.0 20.9 24.2 29.1
Hungary 26.1 28.0 27.3 30.2 22.6 23.7 23.0 24.9
Iceland 21.3 21.2 22.9 26.4 18.0 18.0 19.6 23.0
Ireland 24.6 25.0 25.4 26.6 24.6 25.4 27.0 26.5
Israel 16.9 19.3 24.8 31.0 13.3 15.2 19.5 26.6
Italy 27.8 29.0 29.0 29.9 22.0 23.1 23.6 24.3
Latvia 16.3 18.5 18.3 19.9 13.4 14.8 14.0 15.2
Lithuania 16.2 18.5 21.4 24.3 13.3 15.0 17.8 21.1
Luxembourg 17.3 20.1 26.5 29.3 10.6 12.4 16.9 17.7
Macedonia 17.3 18.8 22.4 23.7 16.6 17.7 20.6 22.3
Malta 31.2 30.9 32.9 33.6 19.0 19.3 22.8 25.3
Moldova 11.9 13.1 13.7 15.8 12.0 13.0 13.4 15.2
Montenegro 19.3 21.2 23.6 26.3 20.9 22.6 24.6 27.3
Netherlands 14.4 15.2 16.5 18.3 14.1 14.9 15.3 16.1
Norway 17.5 18.1 19.2 20.1 13.3 13.8 14.6 16.0
Poland 17.2 17.6 19.8 21.9 13.6 14.1 15.6 17.8
Portugal 19.9 24.7 27.3 28.7 18.0 21.8 25.1 27.1
Romania 8.1 7.8 9.0 11.0 16.0 15.6 17.4 20.3
Russia 14.7 17.1 19.4 21.7 14.3 16.8 16.7 18.6
Serbia 8.2 9.0 13.1 19.2 10.5 11.3 16.0 23.1
Slovakia 14.2 16.4 18.0 20.6 10.3 11.6 12.4 13.5
Slovenia 24.4 26.2 29.6 33.1 14.1 14.6 18.1 24.0
Spain 20.2 23.9 28.3 27.6 17.6 20.2 23.3 23.8
Sweden 16.4 14.3 16.6 20.4 17.4 15.0 16.9 19.3
Switzerland 19.0 18.3 18.1 20.7 15.5 15.0 14.9 16.2
Ukraine 8.0 9.2 9.1 10.6 16.3 18.1 17.9 20.1
United Kingdom 17.6 17.5 21.9 26.1 21.0 21.0 26.2 29.2
Central Europe 17.2 17.6 19.1 21.3 16.7 16.8 18.1 20.3
Eastern Europe 13.0 15.1 16.7 19.0 14.8 16.9 16.8 18.8
Western Europe 19.6 20.6 22.6 24.2 17.6 18.4 20.5 22.0
Source: Ng et al. (2014, Appendix, Webtable 9)
JANPA WP4 countries are highlighted.
Table T2.3. Trends in obesity (IOTF classification) for males and females aged 2-19 years, Europe,
1980-2013
Country/Region
Males Females
1980 1990 2000 2013 1980 1990 2000 2013
Albania 15.2 15.8 15.1 11.5 14.2 14.6 14.4 12.8
Andorra 7.1 7.7 8.9 9.3 7.7 8.3 9.4 9.5
Austria 7.1 8.1 9.5 10.3 6.0 6.7 7.5 7.8
Belarus 3.1 3.4 3.3 3.8 3.6 3.9 3.8 4.2
Belgium 4.7 5.0 4.8 4.6 5.0 5.2 4.5 4.2
Bosnia and Herzegovina
6.7 7.2 8.8 10.1 8.2 8.8 10.5 11.6
Bulgaria 9.3 8.2 6.8 6.9 8.5 7.6 6.6 6.7
Croatia 4.2 4.6 5.7 7.6 2.9 3.2 4.1 5.6
Cyprus 6.9 7.8 7.8 8.0 6.1 6.8 7.0 7.4
Czech Republic 6.5 6.8 6.7 6.4 4.6 4.9 5.0 4.8
Denmark 5.9 6.6 8.0 8.7 3.9 4.3 5.1 5.9
Estonia 5.1 5.4 6.4 7.3 5.6 5.9 7.0 7.6
Finland 6.0 6.2 6.8 9.2 4.3 4.5 5.0 6.6
France 5.0 5.4 5.5 5.8 4.0 4.3 4.4 4.7
Germany 4.4 4.5 4.9 5.5 4.2 4.3 4.7 5.3
Greece 5.9 7.3 9.6 10.5 5.2 6.1 7.3 7.9
Hungary 9.8 10.4 8.5 7.9 7.6 8.0 6.5 6.1
Iceland 7.3 7.4 8.1 9.6 5.3 5.4 6.1 7.6
Ireland 5.7 5.8 6.3 6.9 6.5 6.7 7.2 7.2
Israel 9.1 10.5 12.9 13.9 6.3 7.2 9.1 11.3
Italy 7.8 7.9 8.1 8.4 6.4 6.5 6.4 6.2
Latvia 3.7 4.2 4.2 4.8 3.5 3.9 3.4 3.4
Lithuania 3.8 4.4 5.2 6.3 3.6 4.1 4.6 5.2
Luxembourg 6.0 6.8 9.2 11.1 10.2 11.6 12.8 13.5
Macedonia 5.6 6.1 7.3 8.6 3.8 4.1 4.8 5.4
Malta 13.1 13.0 13.3 12.5 5.9 6.0 7.2 7.9
Moldova 4.4 4.8 4.9 5.6 4.3 4.7 4.8 5.3
Montenegro 6.8 7.5 8.3 9.4 6.2 6.7 7.4 8.3
Netherlands 3.4 3.9 3.7 4.1 4.2 4.4 3.8 3.8
Norway 5.3 5.5 5.3 5.1 4.0 4.2 4.2 4.0
Poland 5.3 6.1 6.4 6.9 4.7 5.9 6.2 6.0
Portugal 6.8 8.5 9.3 8.9 6.9 8.5 10.0 10.6
Romania 6.6 6.4 7.2 8.6 4.4 4.2 4.7 5.7
Russia 6.2 7.5 6.9 7.3 5.6 6.7 6.3 6.6
Serbia 3.3 3.6 5.1 6.7 2.7 3.0 4.6 6.9
Slovakia 4.0 4.7 5.0 5.5 3.8 4.4 4.8 5.5
Slovenia 5.7 6.1 6.5 7.2 2.8 2.8 4.1 5.3
Spain 5.7 6.9 8.5 8.4 5.5 6.3 7.1 7.6
Sweden 3.0 3.0 3.6 4.3 3.4 3.9 4.0 4.0
Switzerland 6.0 5.8 5.7 6.6 4.5 4.5 4.7 5.5
Ukraine 5.8 6.5 6.4 7.3 5.3 5.9 5.8 6.5
United Kingdom 4.7 4.4 5.4 7.4 5.7 5.5 6.8 8.1
Central Europe 6.5 6.8 7.1 7.5 5.3 5.8 6.1 6.3
Eastern Europe 5.9 7.0 6.5 7.1 5.4 6.3 5.9 6.4
Western Europe 5.5 5.8 6.4 7.2 5.1 5.3 5.8 6.4
Source: Ng et al. (2014, Appendix, Webtable 10)
JANPA WP4 countries are highlighted
69
Table T2.4. Selection of published prevalence estimates of child and adolescent overweight and obesity from KANPA WP4 countries collected since 2000
Country First author, Year Year(s) of data
collection Sample
Age group
BMI Sub-group Boys Girls
Other relevant measures Comments
OW* OB* OW OB
Croatia Farkas, 2015 2011 Osijek region - 10 Kindergartens: 760
Ages 3 to 7 years
Measured, WHO
All 10.4 13.6 7.8 8.6 None
Prevalence is reported by discrete age groups but not tabulated here due to small sample size
Croatia Juresa, 2012 2003-2004
School health survey - Representative sample: 960
6.5 to 7.5 years
Measured, IOTF
All 13.8 8.3 12.6 6.9 None None
Croatia Petranovic, 2014 2010
Zagreb region - school sample, not necessarily representative: 804
Age 15 to 19 years
Measured, CDC
All 14.3 8.9 9.8 3.5 None None
Greece Kotanidou, 2013 2004-2010
Systematic review
covering about
430,000 children
aged 1-12
Ages 1 to
12
Measured,
IOTF All 24.1 11.0 9.7 23.2 None
Prevalence of OW and OB
for boys and girls
combined was estimated
at 23.7% and 10.2%
respectively.
Greece Manios, 2007 2003-2004
Broadly representative sample (5 regions) census-based sample: 2374
1 to 5 years
Measured, IOTF
All 12.9 6.2 15.5 8.1 None
Prevalence is reported by discrete age groups but not tabulated here due to small sample size
Greece Wijnhoven, 2014a
2009-2010 Representative sample (COSI round 2)
7 and 9 years
Measured, IOTF
Age 7 24.5 13.6 25.6 14.3
None None
Age 9 30.4 14.7 27.7 14.6
Greece Tambalis, 2010 2007 Representative sample: 71,227
8 to 9 years
Measured, IOTF
All 26.5 12.2 26.7 11.2 None None
Greece Grammati-kopoulou, 2014
2010-2012 ADONUT Study - Representative sample: 37,344
12 to 19 years
Measured, IOTF
Age 12 33.8 8.0 24.0 11.1 AO*: 11.5% boys, 15.5% girls Significantly more OW/OB boys than girls at all ages; significantly more girls with AO than boys at ages 12, 13 and 17 only; AO cutpoints from International Diabetes
Age 13 29.1 10.8 25.1 6.1 AO: 14.7% boys, 12.3% girls
Age 14 29.2 9.8 21.4 6.4 AO: 11.9% boys, 11.5% girls
Age 15 26.3 8.8 17.9 5.7 AO: 9.3% boys, 10.3% girls
Age 16 26.8 8.5 16.9 4.8 AO: 6.6% boys, 7.4% girls
Country First author, Year Year(s) of data
collection Sample
Age group
BMI Sub-group Boys Girls
Other relevant measures Comments
OW* OB* OW OB
Age 17 26.1 8.2 17.3 5.5 AO: 4.4% boys, 5.5% girls Federation (IDF), 2004
Age 18 25.9 7.5 15.5 4.7 AO: 3.9% boys, 3.2% girls
All 27.9 8.9 19.4 6.0 AO: 8.9% boys, 9.2% girls
Ireland Mangan, 2013 2008-2009
Nationally representative sample (GUI): 11,134
Age 9 months
Measured, UK-WHO
All 24.8 15.7 24.8 15.7 None
Results are not reported by gender: that is, 24.8% of all children were classified as overweight and 15.7% as obese.
Ireland Layte, 2011 2007-2008 Nationally representative sample: 8,568
Age 9 years
Measured, IOTF
All 17.0 5.0 22.0 8.0 None
Percentages have been rounded to the nearest whole number in the report.
Ireland McCarthy, 2016a 2013-2014
Fluoride and Caring for Children’s Teeth (FACCT) study, representative sample: 5232
Ages 4-14 years
Measured, IOTF
Age 4-7 13.0 4.0 19.0 6.0
None
Estimates are based on unpublished data, reported in McCarthy et al. (2016), who reported percentages to the nearest whole number
Age 9-14 19.0 4.0 21.0 7.0
All 16.0 4.0 20.0 7.0
Ireland Heinen, 2014 2012
COSI round 3, nationally representative sample: 1729 (age 7) and 1945 (age 9)
Ages 7 and 9 years
Measured, IOTF
Age 7 12.2 2.2 15.9 5.5
None None
Age 9 15.9 4.1 17.7 4.3
Ireland Fahey, 2005 2004 Nationally representative sample: 3051
Ages 13 to 18 years
Measured, IOTF
Age 13 16.4 6.1 22.9 3.5
None None
Age 14 17.7 4.4 19.1 3.5
Age 15 14.7 4.6 14.3 5.3
Age 16 13.1 3.4 15.9 1.7
Age 17 15.7 4.9 13.6 4.2
Age 18 18.0 6.0 11.1 9.3
All 15.4 4.5 16.6 3.8
71
Country First author, Year Year(s) of data
collection Sample
Age group
BMI Sub-group Boys Girls
Other relevant measures Comments
OW* OB* OW OB
Italy Tuchetta, 2012 Systematic review: 25 studies
published 2003-2010 Ages 6 to 11 years
Measured, IOTF
All 17.3-35.2
5.9-21.1
18.8-31.6
5.1-20.7
None
There is large regional variation. Taking only those studies with sample size >1000, in the North, OW ranged from 19%-21%; in the Centre, this ranged from 22-26%; and in the South, this was 25-26%; figures for OB for the North, Centre and South, respectively (sample sizes >1000) are 5.5-8.9%, 8.0-12.7%, and 16.6-20.9%, respectively. No gender breakdown by region is provided.
Italy Maffeis, 2006 2002
Regional samples from Verona (Northeast) and Messina (South): 2150
Ages 2 to 6 years
Measured, IOTF
All 14.5 8.1 19.8 7.9 None
Significantly higher prevalence rates in OW and OB in the South and North for both boys and girls; significantly more OW (but not OB) girls than boys. Percentages are estimates, read from Figure 1.
Italy Spinelli, 2015, Nardone, 2015
2014
OKkio alla SALUTE, nationally representative sample: 48,426
Ages 8 and 9 years
Measured, IOTF
All 20.7 10.3 21.2 9.4 None
Results by gender provided by the OKkio alla SALUTE national team (Nardone, personal communication, February 19, 2016). Across all children, 20.9% were classified as overweight and 9.8% as obese.
Italy Lazzeri, 2008 2006
Regional sample from Tuscany (North-Central): 1430 (age 9), 1178 (age 11-15)
Age 9, 11, 13, 15
Measured, IOTF
Age 9 24.0 8.8 26.2 7.8
None None Age 11 19.3 3.9 13.7 2.5
Age 13 18.7 4.1 10.5 2.3
Age 15 23.7 3.8 10.8 1.4
Italy Celi, 2003 1993-2001 Regional sample Ages 3- Measured, Age 3 13.1 2.9 4.9 4.4 None None
Country First author, Year Year(s) of data
collection Sample
Age group
BMI Sub-group Boys Girls
Other relevant measures Comments
OW* OB* OW OB
from Perugia, Terni and Rieti (Central): 44,231
17.5 IOTF Age 4 8.2 5.3 6.7 3.3
Age 5 14.3 7.5 9.9 4.6
Age 6 13.9 8.7 11.5 7.7
Age 7 15.9 8.4 16.9 9.3
Age 8 18.0 8.5 22.0 10.3
Age 9 22.9 7.5 22.5 9.3
Age 10 27.7 7.7 21.3 12.3
Age 11 27.5 7.0 23.5 7.3
Age 12 25.6 6.4 22.9 6.1
Age 13 23.3 5.5 21.4 6.4
Age 14 20.6 5.1 19.6 4.5
Age 15 19.3 5.2 16.6 4.6
Age 16 15.0 5.1 12.5 4.6
Age 17 15.0 4.3 11.1 4.2
All 20.9 6.7 18.9 6.5
Portugal Nazareth, 2013 2012 Representative sample: approx 2200
Age 12-36 months
Measured, WHO
12-23 months
32.9 6.2 32.9 6.2
None
Results were not reported by gender, that is, 31.4% of all children were overweight and 6.5% obese.
24-36 months
30.0 6.8 30.0 6.8
All 31.4 6.5 31.4 6.5
Portugal Vale, 2011 2008-2009 Kindergarten sample from Porto region: 625
Ages 3-6 Measured, IOTF
All 27.6 9.7 20.3 9.3 None None
Portugal Rito, 2015 2012-2013
COSI round 3, nationally representative sample: 7430
Ages 6-8 Measured, IOTF
Age 6 14.4 6.9 15.6 6.6
None
Significantly more girls than boys OW/OB at age 7, but significantly more boys OW/OB than girls at age 8.
Age 7 16.1 8.0 19.4 7.8
Age 8 18.3 11.7 17.9 8.9
73
Country First author, Year Year(s) of data
collection Sample
Age group
BMI Sub-group Boys Girls
Other relevant measures Comments
OW* OB* OW OB
Portugal Sardinha, 2011 2008
Nationally representative sample of mainland Portugal: 22,048
Ages 10-18
Measured, IOTF
Age 10 23.9 7.6 21.9 5.8
None None
Age 11 22.3 7.0 20.5 7.1
Age 12 20.8 6.9 20.5 4.8
Age 13 18.6 8.1 19.8 5.8
Age 14 18.8 5.2 18.0 4.3
Age 15 15.5 5.3 15.6 3.8
Age 16 15.3 4.4 15.4 4.5
Age 17 15.3 4.4 13.5 3.5
Age 18 13.4 4.3 12.7 3.3
All 17.7 5.8 17.0 4.6
Romania Nanu, 2011 2010
National nutrition programme: details on sample not available
Ages 0-24 months
Measured, WHO
All 5.4 None
OW and OB are not reported separately; this is an estimate of 'increased weight for height'. No significant gender differences were found.
Romania
Nicolescu, 2013; WHO Regional Office for Europe, 2016
NS COSI Age 8-9 Measured, IOTF
All 14.8 15.0 15.7 8.0 None None
Romania Ardeleanu, 2015 2010-2012
Broadly representative sample (14 counties of Romania): 14,030 aged 6-7 years, 13,385 aged 13-14 years
Ages 6-7 years, 13-14 years
Measured, WHO
Age 6-7 10.1 9.5 10.1 9.5 None
Results are not reported by gender: that is, 10.1% of all children were classified as overweight and 9.5% as obese.
Age 13-14 11.0 6.8 11.0 6.8 None
Results are not reported by gender: that is, 11% of all children were classified as overweight and 6.8% as obese.
Romania Mocanu, 2013 2008-2012 North-east Age 6-10 Measured, Age 6 13.2 5.3 16.7 7.1 None No significant gender
Country First author, Year Year(s) of data
collection Sample
Age group
BMI Sub-group Boys Girls
Other relevant measures Comments
OW* OB* OW OB
Romania (Iasi and Neamt): 3444
years IOTF Age 7 14.6 10.0 16.8 8.1 differences at any age.
Age 8 17.8 7.4 14.0 6.3
Age 9 19.1 9.4 19.1 6.5
Age 10 18.6 4.7 14.8 3.2
All 16.8 7.8 16.3 6.4
Romania Valean, 2009 Not stated North-west Romania (Cluj-Napoca): 7904
Age 6-18 years
Measured, CDC
Grades 1-4 16.8 16.9 15.1 9.8
None Significantly more OW and OB boys than girls at all age groups
Grades 5-8 15.6 9.1 14.1 6.2
Grades 9-12
9.9 6.0 5.6 1.2
All 10.8 14.2 5.9 11.5
Romania Chirita Emandi, 2012
2010-2011 West Romania (Timis): 3731
Age 7-19 years
Measured, IOTF
All 20.7 9.0 16.3 5.8 None BMI by age and gender is reported in the paper.
Romania Barbu, 2015 2010-2011 South Romania (Bucharest): 866
Age 6-17 years
Measured, IOTF
All 21.1 8.0 16.2 4.5 None Significantly more OW and OB boys than girls
Slovenia Sedej, 2014 2009 Representative sample: 5496
Age 5 Measured, IOTF
All 12.6 4.1 16.7 4.7 None Significantly more OW and OB girls than boys
Slovenia Wijnhoven, 2014a
2009-2010 Representative sample: COSI round 2
Ages 6-9 Measured, IOTF
Age 6 11.3 5.2 12.5 5..8
None None
Age 7 14.0 7.1 14.5 6.7
Age 8 18.0 8.5 18.1 8.5
Age 9 17.7 8.0 19.3 9.8
Slovenia Kovac, 2012 2011 Representative sample (SLOfit): >150,000
Ages 7-18
Measured, IOTF
Age 7 14.2 7.5 15.3 6.8
None
These percentages are approximations, read from Figures 2 and 3 in the article. Prevalence of OW and OB is higher in boys than girls at all ages (but not confirmed by
Age 8 17.0 8.0 17.4 7.0
Age 9 20.0 9.0 20.0 6.9
Age 10 20.5 9.5 20.0 6.8
Age 11 24.0 9.4 20.1 6.8
75
Country First author, Year Year(s) of data
collection Sample
Age group
BMI Sub-group Boys Girls
Other relevant measures Comments
OW* OB* OW OB
Age 12 24.0 9.5 19.5 5.3 significance tests).
Age 13 21.2 8.0 17.7 4.8
Age 14 20.0 7.4 16.0 4.7
Age 15 19.6 6.3 15.0 4.1
Age 16 20.2 6.1 13.0 3.9
Age 17 20.1 4.3 13.0 3.2
Age 18 21.0 4.1 12.5 3.9
All 20.0 7.5 17.4 5.5
Slovenia Starc, 2016 2014 Representative sample (SLOfit): 184,671
Ages 7-18
Measured, IOTF
Age 7 13.6 7.2 15.0 6.7
Triceps skinfold also measured, not reported here.
Source: Unpublished data provided by Dr Gregor Starc: SLOfit database 1989-2015, Laboratory for the Analysis of Somatic and Motor Development, Faculty of Sport, University of Ljubljana. Estimates for the two adjacent years, 2013 and 2015, show a very similar pattern as 2014.
Age 8 17.3 7.8 18.1 7.3
Age 9 19.7 7.3 19.5 6.7
Age 10 20.5 8.4 20.7 6.2
Age 11 22.1 7.9 19.3 5.8
Age 12 21.8 8.1 18.4 5.1
Age 13 20.0 7.9 16.9 4.9
Age 14 19.0 7.0 16.4 4.6
Age 15 19.5 6.2 14.3 3.7
Age 16 20.5 5.6 14.2 4.1
Age 17 19.8 4.5 13.0 3.9
Age 18 20.6 3.2 12.3 3.7
All 19.4 7.0 16.8 5.4
*OW = overweight, OB = obese, AO = abdominal obesity.
Table T2.5. Summary of publications examining inequalities in the prevalence of child overweight and obesity for JANPA participants
Country First author, year Age group Socio-economic
Demographic Regional or Geographic
Parental BMI
Breastfeeding
Child-related health behaviours
Parent-related health behaviours
Factors related to higher risk of OW/OB
Croatia Bralic, 2005 Mean age 11.3
x
Higher parental BMI
Croatia Petricevic, 20112 Median age 6.8
x
Higher parental BMI
Croatia Mandic, 2011 12 months
x
Not exclusively breastfed
Croatia Skledar, 2015 6-7 years
x
Not exclusively breastfed
Croatia Juresa, 2012 7 years x x
Fewer children, lower birth order, higher levels of parental education
Greece Tzotzas, 2008 13-19 years
x x
x
Greek compared with non-Greek (boys only); no differences across rural, semi-urban and urban areas; smoking and alcohol consumption (girls only)
Greece Patsopolou, 2015 12-18 years x x
x x Higher maternal education, boys, eating outside of kitchen/dining room, paternal worry
Greece Hassapidou, 2009 8-12 years x x
x x Less pocket money, Greek-born, higher energy intake and lower exercise, food prepared by grandmother
Greece Papadimitriou, 2006
6-11 years
x
Greek-born
Greece Manios, 2007 1-5 years x x x x
Parental BMI was the only significant predictor of OW/OB
Greece Malindretos, 2009 Mean age 12.2
x
Higher parental BMI
Greece Kyriazis, 2012 6-12 years
x
Skipping breakfast, not consuming fruits and vegetables, consuming bread and soft drinks; hours spent watching TV
Greece Roditis, 2009 N/A x x x
x x
This is a narrative review of OW/OB and characteristics associated with it, including nutrition, physical activity, socio-economic and demographic inequalities
Greece Tambalis, 2013 10-12 years
x
x
Rural; despite higher levels of physical activity
77
Country First author, year Age group Socio-economic
Demographic Regional or Geographic
Parental BMI
Breastfeeding
Child-related health behaviours
Parent-related health behaviours
Factors related to higher risk of OW/OB
Greece Jelastopulu, 2012 10-13 years x
x
x
Lower parental education, higher parental BMI, fewer daily meals and more time spent in front of the television and/or on the computer
Greece Kollias, 2011 Mean age 9.2
x
x
Higher parental BMI, consumption of sweets and fast-food, and decreased physical activity
Greece Farajian, 2011 10-12 years
x
x
Some aspects of child's diet but not overall dietary quality were associated with OW/OB. Dietary quality was better in rural and semi-urban compared to urban areas.
Greece Cassimos, 2011 11-12 years
x
x
Higher parental BMI, fewer meals in larger portions, more time watching TV
Greece Antonogeorgos, 2010
10-12 years x
x
x
After adjusting for parent BMI and education, less physical activity was associated with OW/OB
Greece Veltsitsa, 2010 7 and 18 years
x x x x
Lower paternal education, boys, urban residence, higher parental BMI
Greece Manios, 2011 10-12 years x x
x
Lower paternal education, boys, higher birth weight, higher parental BMI
Greece Kosti, 2007 12-17 years
x
Lower levels of physical activity, skipping breakfast
Greece Hassapidou, 2006 11-14 years
x
Higher consumption of unhealthy snacks, sugars and fats
Greece Angelopolous, 2006 11 years
x
Higher consumption of fast foods and lower levels of physical activity
Greece Chalkias, 2013 8-9 years x
x
Lower parental education was the strongest predictor of OW/OB; areas characterised by low educational levels, low income, high population density and limited recreation zones constitute obesogenic environments
Country First author, year Age group Socio-economic
Demographic Regional or Geographic
Parental BMI
Breastfeeding
Child-related health behaviours
Parent-related health behaviours
Factors related to higher risk of OW/OB
Greece Kontogianni, 2010 3-18 years x x
x
Skipping breakfast, less frequent eating, and lower dietary quality were associated with OW/OB after adjusting for age, sex and parental education
Greece Panagiotakos, 2008 10-12 years x x
x x
Boys, higher parental BMI, not breastfed, higher birthweight (girls only) (after adjusting for parental education which was not significant)
Greece Lagiou, 2008 10-12 years
x
x
Girls, Greek-born, physical inactivity
Greece Papandreou, 2010 7-15 years
x x x
Higher parental BMI, not breastfed, lower levels of physical activity, higher consumption of sugar-sweetened drinks
Ireland Fahey, 2005 13-18 years x
x
No associations between parental occupation or participation in sports and OW/OB
Ireland Heinen, 2014 7-9 years x x x
Girls (in some comparisons, but not others), trends towards decreasing OW/OB observed in non socio-economically disadvantaged schools but not in socio-economically disadvantaged schools, no significant differences by HSE (Health Services Executive) region.
Ireland Keane, 2012 9 years x x
x
Multinomial regression indicated that higher probability of OW/OB was associated with girls, one parent families, lower occupational class, lower maternal education, and particularly parental OW/OB
Ireland Layte, 2011 9 years x
x
Multivariate analysis shows that probability of OW/OB varies by occupation class, physical and sedentary activities, and quality of diet
79
Country First author, year Age group Socio-economic
Demographic Regional or Geographic
Parental BMI
Breastfeeding
Child-related health behaviours
Parent-related health behaviours
Factors related to higher risk of OW/OB
Ireland O'Shea, 2014 5-12 years x
Lower socio-economic status (medical card)
Ireland Walsh, 2015 9 years x x x x x x Results confirm strong socio-economic gradient, a majority of which is due to parent rather than child level measures
Ireland Perry, 2015 9 years X x x x x Lower parental education, girls, higher parental BMI, less physical activity, more TV viewing, and poorer dietary quality
Ireland Layte, 2014 0-3 years x x x x x x
Paper examined weight gain over three years: lower SES associated with lower birth weights and highest gains in weight; higher gains in weight associated with higher maternal weight gain during pregnancy and no breastfeeding
Ireland Williams, 2013 3years X Findings confirm presence of social gradient (occupation class) at age 3
Italy Spinelli, 2014 8-9 years
x
Prevalence varies dramatically by region, being highest in the South and lowest in the North. Prevalence was highest in Campania and lowest in Bolzano
Italy Spinelli, 2012 8-9 years x x x x
No gender difference, no variation by urban-rural environment but much higher prevalence in the South than the North, higher in Italian-born children, lower parental education, and higher parental BMI
Country First author, year Age group Socio-economic
Demographic Regional or Geographic
Parental BMI
Breastfeeding
Child-related health behaviours
Parent-related health behaviours
Factors related to higher risk of OW/OB
Italy Binkin, 2008 8-9 years x
x
Lower maternal education and employment rates are associated with OW/OB but do not explain the large regional differences in prevalence.
Italy Spinelli, 2009 8-9 years x x x
Southern region, boys lower maternal education and maternal employment rates (with no variation by urban-rural location)
Italy Lombardo, 2014 8-9 years x x x x
Males, southern region, lower parental education and higher parental BMI
Italy Nardone, 2015 8-9 years x
x
Southern region, lower parental education
Italy Cavallo, 2013 11, 13, 15 years
x
Self-reported OW/OB tended to be highest in Campania and lowest in Bolzano
Italy Lazzeri, 2014 11, 13, 15 years
x
x
x
Self-reported OW/OB was higher in the South, with lower parental education and less frequent eating of breakfast
Italy Lazzeri, 2011 8-9 years x
x
Lower parental education, higher parental BMI
Portugal Nazareth, 2013 12-36 months
x Rates of overweight or obesity were higher in the North than in the Algarve region
Portugal Rito, 2015 6-8 years
x
Prevalence of OW was higher the Lisbon and the Tagus Valley while the Centre had the highest prevalence of OB
Portugal Rito, 2012a 6-8 years
x
Similar geographic variations as reported in Rito et al. (2015); in addition, no variation in prevalence by rural, semi-urban or urban environments
Portugal Bingham, 2013 3-10 years x x
x x x x Females, not breastfed, maternal smoking during pregnancy, less physically active, lower parental education and higher parental BMI
81
Country First author, year Age group Socio-economic
Demographic Regional or Geographic
Parental BMI
Breastfeeding
Child-related health behaviours
Parent-related health behaviours
Factors related to higher risk of OW/OB
Portugal Rito, 2012b 6-8 years
x
Prevalence of overweight was highest in the Azores while prevalence of obesity was highest in Madeira
Portugal Nogueira, 2013 3-10 years x
x
Poor built environment and unsafe local neighbourhoods (higher OW/OB in girls, not boys)
Portugal Ramos, 2007 13 years
x
Higher parental BMI
Portugal Ferrao, 2013 3-10 years x x x
Unsafe and less well-maintained neighbourhoods (after adjusting for parental education and child age, gender and school)
Portugal Moreira, 2007a 6-12 years
x
x
More weight gain in mothers during pregnancy, not low birthweight, first-born, fewer siblings
Portugal Padez, 2009 7-9 years x
x
x
Shorter sleep duration, lower parental education, higher parental BMI and more time watching TV
Portugal Padez, 2005 7-9 years x x
x
x
Lower parental education, smaller family size, higher birthweight, higher parental BMI, and shorter sleep duration
Portugal Ferreira, 2008 6-10 years
x
Higher parental BMI
Portugal Li, 2015 3-10 years
x Maternal smoking during pregnancy
Romania Cosoveanu, 2011 2-14 years
x x x
Not breastfed, early introduction of solid foods, higher parental BMI, less healthy diet, eating food while watching TV/computer
Romania
WHO Regional Office for Europe, 2016; Nicolescu, 2013
8-9 years X Higher prevalence in urban compared with semi-urban and rural areas; lowest rates in rural areas.
Romania Mocanu, 2013 6-10 years x
x
Higher SES (positive relationship), high consumption of fried food
Romania Morea, 2013 6-19 years
x
Pre-pregnancy maternal OW/OB
Country First author, year Age group Socio-economic
Demographic Regional or Geographic
Parental BMI
Breastfeeding
Child-related health behaviours
Parent-related health behaviours
Factors related to higher risk of OW/OB
Slovenia Starc, 2014 7-18 years
x
Rates of overweight and obesity were highest in Pomurska and Zasavska regions; possible reasons for these differences, e.g. socio-economic, educational, environmental, are mentioned, but not analysed
83
Table T2.7. Trends in overweight and obesity (IOTF classification) for males and females aged 20
years and older, Europe, 1980-2013
Country/Region
Males Females
1980 1990 2000 2013 1980 1990 2000 2013
Albania 46.6 46.8 50.3 56.2 37.9 37.9 40.5 45.8
Andorra 40.2 41.2 35.5 34.4 36.3 37.3 34.7 36.1
Austria 45.0 49.7 59.2 59.7 27.1 30.4 40.5 42.8
Belarus 37.6 40.5 40.6 44.1 39.7 42.1 42.0 44.7
Belgium 50.6 53.1 55.9 58.0 40.9 43.0 45.1 47.1
Bosnia and Herzegovina 47.6 49.5 53.6 57.3 44.3 45.8 49.1 51.9
Bulgaria 68.0 64.1 58.9 59.7 56.0 52.4 47.9 48.8
Croatia 53.1 54.9 59.4 65.5 41.6 43.1 46.4 51.0
Cyprus 61.8 64.8 65.9 67.8 47.5 50.3 51.1 52.1
Czech Republic 60.1 61.1 63.0 65.5 51.4 51.9 50.4 50.0
Denmark 49.8 52.5 57.8 59.2 33.5 36.1 43.7 44.7
Estonia 49.3 50.6 54.5 59.3 50.1 50.7 52.0 54.3
Finland 54.9 56.6 59.5 62.2 43.2 44.0 45.8 50.4
France 47.0 50.1 54.6 55.9 32.4 35.4 41.3 42.8
Germany 57.5 60.3 63.8 64.3 39.6 42.7 48.0 49.0
Greece 56.2 61.9 71.7 71.4 43.0 45.8 48.2 51.1
Hungary 64.6 66.0 63.6 65.6 55.1 56.3 53.8 54.8
Iceland 66.7 67.1 70.3 73.6 52.1 52.4 56.4 60.9
Ireland 56.2 58.9 64.8 66.4 43.6 45.1 50.0 50.9
Israel 50.6 54.6 58.3 60.4 45.3 49.2 52.9 52.7
Italy 54.5 56.3 58.3 58.3 38.0 39.2 41.6 41.4
Latvia 51.6 54.8 53.6 56.3 54.5 56.8 54.9 55.8
Lithuania 48.0 51.8 57.9 63.9 46.0 49.0 52.9 56.2
Luxembourg 51.0 54.0 56.4 58.0 36.3 39.3 43.1 44.4
Macedonia 44.2 46.4 52.1 57.0 44.0 45.5 49.0 51.7
Malta 68.0 68.3 72.0 74.0 51.0 51.3 55.2 57.8
Moldova 37.0 39.5 40.5 44.7 53.1 55.1 56.2 58.8
Montenegro 51.6 53.9 56.6 60.1 49.9 52.0 54.2 57.0
Netherlands 48.9 50.5 49.9 53.2 42.4 43.1 40.8 44.9
Norway 48.9 50.3 54.2 58.4 39.0 40.6 45.0 47.3
Poland 61.1 61.0 62.3 64.0 47.3 46.7 47.6 49.4
Portugal 52.4 58.8 61.8 63.8 44.1 49.7 52.4 54.6
Romania 54.0 52.8 55.8 60.4 45.1 44.1 46.4 50.3
Russia 40.4 42.4 47.4 54.3 52.2 54.3 59.7 58.9
Serbia 47.6 49.8 52.5 55.7 43.3 45.3 47.8 50.4
Slovakia 58.6 61.9 63.0 64.4 47.4 49.9 50.3 51.5
Slovenia 57.1 58.9 62.1 65.1 44.6 46.1 49.4 52.1
Spain 51.5 55.5 63.4 62.3 40.8 44.2 48.1 46.5
Sweden 53.2 53.5 56.6 58.2 40.6 41.1 43.8 45.8
Switzerland 52.4 51.7 53.6 56.6 35.0 34.4 36.9 39.9
UK 53.8 55.9 64.4 66.6 43.8 45.3 53.7 57.2
Ukraine 52.1 55.2 55.1 59.1 52.1 54.7 54.5 57.4
Central Europe 58.1 58.2 59.4 62.2 47.8 47.6 48.3 50.4
Eastern Europe 43.5 45.7 49.1 55 51.5 53.8 57.4 57.8
Western Europe 52.9 55.6 60.4 61.3 39.2 41.5 46.2 47.6
Source: Ng et al. (2014, Appendix, Webtable 9)
JANPA WP4 countries are highlighted
CHAPTER 3: EVIDENCE: CHILDHOOD IMPACTS OF CHILDHOOD
OVERWEIGHT AND OBESITY
3.1. International/European evidence
3.1.1. Introduction
Queally et al. (2016) conducted a review of the international literature on the impacts of childhood
overweight and obesity on a range of medical and non-medical outcomes. This section summarises
the findings of 18 systematic reviews included in their review. Queally et al. (2016) did not include
educational achievement/attainment in their review, so recent systematic reviews in this area are
considered also. The subsequent section summarises 81 sources of information on this topic
retrieved from the seven countries JANPA WP4 countries.
Queally et al. (2016) note that much of the literature in this general area has focused on the long-
term consequences of obesity as it manifests in adulthood. However, overweight/obesity leads to a
number of adverse outcomes during childhood, and over the past decade, increasing numbers of
children manifest symptoms of what would previously have been considered adult diseases.
Queally et al. (2016) conducted their review in order to address four research questions:
1. What are the medical and non-medical consequences of childhood overweight and obesity
in childhood (age 0-18 years)?
2. What is the evidence base for each associated condition related to childhood
overweight/obesity?
3. Are there studies that have systematically reviewed one or more adverse outcomes and
their associations with childhood overweight/obesity?
4. Are there defined relative risks or odds ratios for these outcomes?
They conducted their search in three phases:
1. During February 2016, a scoping exercise on the American Academy of Paediatrics
(www.aap.org), Public Health England (www.noo.org.uk/NOO_pub) and Google Scholar was
conducted to establish a list of comorbidities to inform the search criteria.
2. Relevant systematic reviews were identified and, where available, relative risks or odds
ratios extracted.
3. Further searching of the literature was conducted for each of the co-morbidities identified in
1 above in order to update the evidence base and try to ensure that all relevant co-
morbidities and adverse outcomes were included.
In all, 327 studies were screened by title or abstract and 29 full texts subsequently assessed. Of
these, 18 articles were included in the review. Of these, 6 included meta-analysis. A summary of the
main findings of each of these articles is shown in the Appendix 2 (Table A8). Table A9 shows a
selection of the effect estimates extracted in the course of this review.
85
3.1.2. The weight of the evidence
The bulk of studies that examine the impact of child/adolescent overweight/obesity have focused on
cardio-metabolic risk factors, psychological ill-health and reduced quality of life. Two of the
systematic reviews identified by Queally et al. (2016) comprised the identification of multiple co-
morbidities, both physical and psychological (Pulgarón, 2013; Sanders et al., 2015), and these give a
very broad indication of the relative emphasis placed on particular conditions and outcomes in the
literature.
In Pulgarón’s (2013) review, 35 of 79 studies examined cardio-metabolic risk factors, and 10
examined anxiety/depression or behavioural or other externalising problems; in Sander et al.’s
(2015) review, cardio-metabolic factors were again the most frequently examined (15 of 47 studies),
12 examined health-related quality of life, and 9 examined mental health.
In the 18 review papers, it was very common for authors to cite the following issues:
There is a lack of high-quality longitudinal data, which hampers the establishment of cause-
effect relationships, particularly for conditions such as asthma and depression.
There are large differences across individual studies in terms of how children’s weight status
has been classified.
There is large variation in the extent to which studies controlled for confounders such as
socio-economic status.
There are inconsistencies in the extent to which differences by gender were examined.
There is a lack of evidence and data on differences among different ethnic or racial groups.
Aside from difficulties in establishing causality and reducing confounding effects, Pulgarón (2013)
sums up the challenges in this area of research as follows: "...there are so many potential confounds
and so much interdependency among the [physical] co-morbidities that it is difficult for researchers
to isolate the effects of childhood obesity." (p. 7).
3.1.3. Cardio-metabolic and cardio-vascular risk factors
Five review papers provide evidence for a link between overweight/obesity and cardio-
metabolic/cardio-vascular risk factors41.
Kelishadi et al. (2015) conducted a systematic review of the associations between abdominal obesity
in children and its associations with any of systolic BP diastolic BP, prehypertension, transient
hypertension, cholesterol, LDL-C, HDL-C, fasting blood sugar, insulin resistance, insulin dose per body
surface, carotid intima-media thickness, and alanine aminotransaminase, among children and
adolescents age 6-18 years. Sixty-one studies were identified in this review. They concluded that
"Whatever the definition used for abdominal obesity and whatever the methods used for
anthropometric measurements, central body fat deposition in children and adolescents increases
the risk of cardio-metabolic risk factors." Blood pressure was the most common measurement
among studies; most confirmed the association of abdominal obesity and elevated blood pressure.
41
Queally et al. (2016) also cite a study by Raj (2012) who conducted a narrative review covering metabolic syndrome/clustering of cardiovascular risk factors, insulin resistance/type 2 diabetes, inflammation and oxidative stress, artherogenic dyspipidemia and atherosclerosis, cardiac structure/function, and sleep disorders. A medical perspective is taken in this review. It is not discussed here in detail.
Reasonably consistent evidence was also found between abdominal adiposity and abnormal lipid
profile and fasting glucose.
Friedemann et al. (2012) conducted a systematic review covering 39 studies and a meta-analysis
covering 24 studies (including healthy children aged 5-15 years in developed countries) which
examined associations between weight status and one or more of systolic or diastolic BP, HDL, LDL
or total cholesterol, triglycerides, fasting glucose or insulin, HOMA-IR, carotid intima media
thickness, and left ventricular mass. In meta-analysis, the mean values of diasystolic, systolic and
ambulatory BP, total, HDL and LDL cholesterol and triglycerides, and fasting glucose, fasting insulin
and HOMA-IR, and CIMT and left ventricular mass were computed for healthy weight, overweight
and obese groups. In all cases differences were statistically significant, with larger differences in
comparisons of obese vs. healthy weight than in overweight vs. healthy weight.
Pulgarón’s (2013) review also provides solid evidence for associations between child/adolescent
weight status and cardio-metabolic risk factors, and the results suggest, consistent with the other
two reviews above, that there is a dose-response relationship between degree of
overweight/obesity and worsening of these risk factors.
Sanders et al. (2015) also comment on the strength of the evidence in this area. For example, one of
the 15 studies in their review found that, compared to normal-weight peers, obese adolescents
(aged 15.4±0.4 years) were significantly more likely to have two or more risk factors for heart
disease, type II diabetes and fatty liver disease (boys 73.5% vs 7.6%; OR, 34.0 [95% CI, 12.6-91.7]; p <
.001; girls 44.4% vs 5.4%; OR, 14.0 [95% CI, 4.1-47.5]; p < .001). All 5 studies examining non-alcoholic
fatty liver disease (NAFLD)42 reviewed by Sanders et al. found significant associations. For example,
one of these 5 studies found that NAFLD increased with increase of adiposity among normal-weight,
overweight and obese boys and girls aged 17 years (boys, 4, 15 and 65 %; girls, 10, 29, and 57%,
respectively). The severity of hepatic steatosis was also associated with BMI, waist circumference,
and subcutaneous adipose tissue thickness (p < .001) in this study.
Anderson et al. (2015) reviewed the evidence on associations between child/adolescent weight
status and NAFLD. They located 74 studies covering ages 1-19 years. The pooled prevalence of
NAFLD in general populations in this study was estimated at 9.0% males and 6.3% females, with a
clear increasing gradient in prevalence associated with weight status: 2.3% (healthy weight), 12.5%
(overweight), and 36.1% (obese). Meta-analysis of available within-study comparisons provided
consistent evidence that prevalence was higher on average in males compared with females and
increased incrementally with greater BMI, although the strength of these associations varied
considerably across studies.
Despite this large body of strong and consistent evidence, however, it is unclear whether the
magnitude of difference in cardio-metabolic/cardio-vascular risk in children of healthy weight,
overweight and obese, continues unchanged into adulthood (Friedemann et al., 2012).
42
NAFLD is included here as a cardio-metabolic risk factor, consistent with the discussion in Alberti et al. (2006).
87
3.1.4. Type 2 diabetes
There is strong evidence for an association between childhood obesity and risk of type 2 diabetes in
childhood, though relatively little is known about this condition in children compared with adults
(Queally et al., 2016).
Childhood obesity is associated with decreased insulin sensitivity, and increased circulating insulin
levels and insulin resistance is an important factor in the development of type 2 diabetes. Queally et
al.’s (2016) review indicates that the most clearly identifiable factor contributing to both type 2
diabetes and cardiovascular disease risk in children was increased body fat; they also note the link
between increasing trends in the prevalence of type 2 diabetes in children/adolescents with
increasing rates of obesity.
Lipid abnormalities and hypertension are two co-morbidities of type 2 diabetes, and evidence cited
by Queally et al. (2016) indicates that the onset of these is particularly marked in young people with
type 2 diabetes. They also point out that the management of type 2 diabetes poses particular
challenges in the paediatric population.
However, there are few estimates of the risk of type 2 diabetes associated with child/adolescent
overweight/obesity. One study conducted in Israel (discussed in Pulgarón, 2013), with about 1
million 17-year-old adolescents receiving a medical evaluation for military service, found that BMI
(i.e. comparing healthy weight with obese) was associated with type 2 diabetes (OR = 5.56; 95% CI
5.09–6.07 and OR = 4.42; 95% CI 3.90–5.00, for male and female subjects, respectively) after
controlling for country of origin, level of education and the year of recruitment.
3.1.5. Type 1 diabetes
There is reasonably strong evidence for an increased risk of type 1 diabetes in childhood associated
with childhood obesity.
Verbeeten et al.’s (2011) systematic review notes that the association between type 1 diabetes and
childhood obesity was first noted in published research about 40 years ago. Their meta-analysis of 4
of these studies yielded a pooled odds ratio of 2.03 for obesity compared to healthy weight (95% CI
1.46-2.80) and meta-analysis of 5 of these studies yielded a pooled odds ratio of 1.25 per 1 SD
increase in BMI (95% CI 1.04-1.51). The restriction of studies to those where weight status was
assessed prior to diagnosis in the inclusion criteria for this review provide confirmation of a causal
relationship. Studies varied in the age of obesity assessment, however.
3.1.6. Asthma
There is reasonably consistent evidence for a link between childhood obesity and risk of asthma or
wheezing, though the causal link is unclear (Pulgarón, 2013).
In a systematic review and meta-analysis of the association between asthma or wheezing and
childhood overweight/obesity (Mebrahtu et al., 2015) the summary odds ratios of underweight
(<5th percentile), overweight (>85th to <95th percentile) and obesity (≥95th percentile) were 0.85
(95% CI: 0.75 to 0.97; p = .02), 1.23 (95% CI: 1.17 to 1.29; p < .001) and 1.46 (95% CI: 1.36 to 1.57, p <
.001), respectively. Heterogeneity in effect estimates across studies was significant and substantial in
all three weight categories, and not accounted for by pre-defined study characteristics.
In Sanders et al.’s (2015) review, 4 of 6 studies examining asthma/asthma symptoms reported
significant associations. For example, in one of these 6 studies, parents of overweight (OR=1.30, 95
% CI= 1.16, 1.46) and obese (OR=1.36, 95 % CI=1.13, 1.62) children aged 4-6 years were significantly
more likely to report that they had asthma ever than parents of healthy weight children.
3.1.7. Dental health
Two systematic reviews on the associations between dental caries and child weight status were
identified by Queally et al. (2016). As Pulgarón (2013) and the authors of these reviews have noted,
more research that takes account of confounding factors such as age, diet and socio-economic
status are needed to better understand the associations between children’s weight status and
dental health.
Hooley et al. (2012) reviewed the results of 48 studies on this topic. They found that 23 studies
found no association, 17 found a positive association, 9 reported an inverse relationship, and 1
reported a U shaped pattern of association. They examined variations in these findings by study and
country characteristics, and found that studies reporting a positive association were from countries
with a higher Human Development Index (HDI) score (mainly Europe and the US), higher quality
dental services (more sensitive dental examination) and a low percentage of underweight children in
the population, while studies reporting a negative association were from countries with a lower HDI
score (mainly Asia and South America), lower quality dental services (less sensitive dental
examination), and more underweight children. Note that less sensitive dental examination can be
expected to result in an underestimate of the number of dental caries.
Hayden et al.’s (2013) systematic review and meta-analysis on this topic (based on 14 studies)
reported that overall, a significant relationship between childhood obesity and dental caries (effect
size = 0.104, p = .049). Results tended to be significant on the basis of standardised BMI comparisons
such as BMI-for-age centiles (effect size = 0.189, 95% CI: 0.060–0.318, p=.004) or IOTF cut-offs
(effect size = 0.104, 95% CI: 0.060–0.180, p=.008). Studies that used Zscores provided non-
significant findings (effect size = 0.147, p=.248), along with studies using non-standardized scales
(effect size = 0.030, p = .884). Consistent with Hooley et al. (2012), Hayden et al. (2013) found that
obese children from industrialized countries (effect size = 0.122, CI = 0.047–0.197, p=.001) had a
significant relationship between obesity and caries in contrast to those from non-industrialised
countries (effect size = 0.079, p=0.264).
3.1.8. Orthopaedic and musculoskeletal problems
There is evidence for an association between childhood obesity and musculoskeletal pain, and some
evidence, though of lower quality, for associations between childhood obesity and low back pain,
and injuries or fractures.
Paulis et al. (2013) conducted a systematic review on the association between weight status and
musculoskeletal complaints (MSC) in children (aged 0-18 years). Forty studies were included in this
review (7 longitudinal and 33 cross-sectional), which provided medium quality evidence that being
overweight in childhood is positively associated with musculoskeletal pain (RR=1.26; 95% CI 1.09–
1.45). There was also evidence of an association between childhood weight and low back pain, as
well as injuries and fractures, though evidence for these associations was of lower quality (RR [back
pain]= 1.42; 95% CI 1.03–1.97; RR [injuries/fractures]=1.08, 95% CI 1.03-1.14). They concluded that
"The link between overweight and MSC might induce a vicious circle in which being overweight,
89
musculoskeletal problems, and low fitness level reinforce each other." (p. 13) and recommended
more high-quality longitudinal research.
Sanders et al.’s (2015) review included two studies that examined musculoskeletal pain; both
reported significant associations. For example, one of these two studies reported odds ratios relative
to healthy weight (OR [overweight]=1.53; OR [obese]=4.09; p=0.010).
Malalignments (pes planus, scoliosis and tibia vara) were not considered in Paulis et al.’s review.
However, Stoltzman et al.’s (2015) systematic review on associations between pes planus (flat feet)
included 13 cross-sectional studies and across these, prevalence varied widely, from 14%-67%;
however, nearly all studies indicated increasing pes planus in children with increasing weight.
Queally et al. (2016) also included a systematic review of muscle strength and fitness and its
association with paediatric obesity (Thivel et al., 2016). The review included 36 studies examining
children and adolescents aged 6 to 18. These studies varied considerably in design (e.g. observation
vs intervention; field vs laboratory), definition of overweight/obesity, assessment instruments, and
samples. Comparisons of field and laboratory studies are complicated by the fact that many
laboratory tests isolate single movements during which body mass is not generally involved. Overall,
though, the review provides strong evidence that children and adolescents with obesity have
reduced muscular fitness compared with children and adolescents of healthy weight. Thivel et al.
(2016) conclude that “Improving muscular fitness and overall musculoskeletal fitness in children with
obesity is of crucial importance to favour their physical autonomy, promote engagement in daily
activities and physical activity-based weight management programmes, and subsequently improve
their health-related quality of life” (p. 61) and recommend further research in this area.
3.1.9. Sleep disorders and sleep problems
Four studies identified by Sanders et al. (2015) examined associations between obstructive sleep
apnea and child/adolescent weight status, and Sanders et al. suggest that the association appears to
be stronger among adolescents than in younger children. Pulgarón’s (2013) review included 12
studies that examined sleep problems or sleep duration, and she concluded that while there is good
evidence to show that sleep problems are more prevalent with increased weight status, the long-
term effects of this are unclear.
3.1.10. Other physical co-morbidities
One of the 18 review papers consisted of a systematic review of developmental co-ordination
disorder (Hendrix et al., 2014) (DCD; a chronic condition characterised by poor fine and/or gross
motor coordination). The prevalence of DCD was estimated to range from 1.7% to 6%, and occurs
four to seven times more often in boys than in girls. All 21 studies in this review reported that
children with DCD had higher BMI scores, larger waist circumference and greater percentage of body
fat compared with their typically developing peers. Eighteen studies (7 cohorts) found these
differences between groups to be statistically significant. Fourteen of 17 studies that used BMI
reported significant differences. There was some evidence of an increased risk of overweight/obesity
associated with DCD over time.
A small number of studies included in the 18 review papers covered other physical co-morbidities,
including acanthosis nigricans (hyperpigmentation), headaches and sexual maturation (see
Pulgarón, 2013; Sanders et al., 2015). These are not considered here in any detail.
3.1.11. Self-esteem and quality of life
There is evidence for lower self-esteem and quality of life among children and adolescents who are
overweight or obese. However, most of the evidence is based on cross-sectional studies, so the
direction of causality is difficult to establish and makes quantification of impact impossible. Also, a
variety of different measures of self-esteem and quality of life are used in the literature, making
direct comparisons across studies challenging (Pulgarón, 2013).
Taking these caveats into account, a systematic review by Griffiths et al. (2010) provides strong
evidence that paediatric obesity impacts on self-esteem and quality of life. Six of nine studies in their
review found lower global self-esteem in obese compared with healthy weight children and
adolescents. Similarly, four out of five studies that incorporated a self-esteem dimension within
quality of life scales reported significantly lower scores in their obese samples. Nine out of eleven
studies using child self-reports, and six out of seven studies using parental reports, found
significantly lower total quality of life scores in obese youth. Griffiths et al. (2010) concluded that
obesity had the greatest negative impact on physical functioning and physical appearance
perceptions, along with social functioning.
Consistent with this, the systematic review by Sanders et al. (2015) included 12 papers examining
health-related quality of life and in all 12 studies, overweight/obese children and adolescents
showed lower health-related quality of life than normal-weight peers. Furthermore, consistent
results emerged regarding worse outcomes for physical, emotional and social aspects of quality of
life. Evidence from a couple of these 12 sources is longitudinal, and suggests that the strength of this
association increases as a cumulative burden of overweight/obesity. Similarly, all four studies on
self-esteem described in Sanders et al. (2015) found significant associations with weight status. For
example, one study reported that obese children (aged 9.2–13.7 years) were between two and four
times more likely to have lower global self-worth than normal-weight peers.
Societal and cultural norms associated with weight and the resulting stigmatisation and negative
stereotyping of individuals with overweight or obesity is likely to have a significant negative impact
on the self-esteem of young people with overweight or obesity. Internalising or taking on these
norms and stereotypes can start from an early age. For example, Rees et al. (2011) conducted a
systematic review of studies on weight stigmatisation from the point of view of children and
adolescents. They included 28 UK-based studies that sought children’s (ages 4-14 years) views on
views on obesity, body size, shape or weight. This study provides evidence that young people are not
concerned with the health impacts of obesity, but rather the social ones. Rees et al. (2011, p. 9)
comment that "…being overweight was seen as a problem [among children] because of the impact it
could have on their lives as social beings, from reduced popularity through to discrimination. The
health consequences of obesity appeared to be largely irrelevant" (p. 9). They found that body
dissatisfaction and aspiration to thinness were extremely common, and more so in girls than boys. In
several of the studies, overweight or obesity was blamed on the individual and seen as something
for individual control. The very overweight children in the review described being teased and bullied
and reported how this impacted seriously on their wellbeing and behaviour.
3.1.12. Depression/low mood
Significant associations between depression or low mood in children and adolescents with a high
weight status have been reported, but a majority of the research in this area has drawn on cross-
91
sectional data. A meta-analysis of the relationships between adult depression and weight status
(Luppino et al., 2010) confirms that there is a reciprocal relationship between these two outcomes,
which may become reinforced over time (see Chapter 4).
Mühlig et al. (2015) conducted a systematic review on the associations between weight status and
depression/low mood among children and adolescents, and concluded that the evidence is mixed,
and firm conclusions are hampered by the methodological variations of the included studies.
Relationships appeared to be more readily detectable in female adolescents and in cross-sectional
studies compared with the longitudinal analyses. Out of 19, 14 cross-sectional studies confirmed a
significant association between obesity and depression. However, just three out of eight longitudinal
studies reported associations between obesity and subsequent depression in female children and
adolescents only. Mühlig et al. (2015) propose that the joint development of obesity and depression
in predisposed subjects may help to explain this discrepancy.
In the review by Sanders et al. (2015), nine studies examining mental health were consistent in their
reports of associations between child/adolescent weight status and depression. For example, one
study reported odds ratios of depression in overweight/obese children aged 6-13 years relative to
healthy weight as follows: (overweight, OR=8.95; obese, OR=18.8; p=.001). However, studies
examining gender differences in this review gave varying results, which is consistent with Pulgarón’s
(2013) observation that the association between weight status and depression varied by sub-groups
in the studies included in her review.
3.1.13. Educational achievement and attainment
There is evidence for a weak negative association between childhood overweight or obesity and
educational attainment, though much of this relationship can be accounted for by socio-economic
disparities between normal-weight and overweight or obese groups of children. The role of
psychological wellbeing in this relationship requires further research, and gender differences could
suggest that this association is stronger in girls than in boys. The main issue, perhaps, is that the
direction of causality in this association is not at all well understood (Caird et al., 2014; Booth et al.,
2014; Sassi et al., 2009).
Caird et al. (2014) conducted a systematic review that included 29 studies that examining the
associations between childhood overweight or obesity and educational attainment, defined as grade
point average (GPA) or other validated attainment measures (excluding standardised cognitive test
scores). These studies were conducted in high income countries and covered children and
adolescents between the ages of 6 and 16 years; a majority of analyses were cross-sectional.
Twenty-six of the 29 studies included adjustments for potential moderating/confounding variables.
When adjusted for SES, the negative relationship between overweight/obesity and educational
attainment was weaker and in many cases not statistically significant. Caird et al. (2014) suggest that
SES-adjusted differences are not socially or educationally important and that the results should be
considered indicative of broader inequalities in health and education.
Sassi et al. (2009) used data from national health surveys undertaken in four countries, including the
Australian National Health Survey (NHS) 1989-2005, the Canadian National Population Health Survey
– cross-section (NPHS) and the Canadian Community Health Survey (CCHS) 1995 -2005, the Health
Survey for England (HSE) 1991-2005 and the Korean National Health and Nutrition Examination
Survey (KNHANES) 1998-2005, to examine associations between education and obesity. The analyses
were conducted by applying the same (or similar) models to all countries’ data, in order to facilitate
comparisons across countries. While they found that the relationship is approximately linear, and
stronger in females than males, they noted that the potential of cross-sectional health survey data in
assessing the causal nature of links between variables is limited. They examined the direction of
causality using data from a French survey (Enquête Décennale Santé 2002-2003) which provided
information on weight status and later educational attainment and weight status. Their results
suggest that the direction of causality appears to run mostly from education to obesity, but they
commented that this conclusion cannot be made with any certainty, since it was only based on one
data source.
A recent study has examined longitudinal associations in child BMI and achievement in the UK on the
basis of the Avon Longitudinal study of Parents and Children (ALSPAC) (Booth et al., 2014). The
authors used national achievement test results for English, Mathematics and Science at ages 11, 13
and 16 (with scores ranging from 1-9) with measured BMI (overweight = zBMI 1.04-1.64; obese =
zBMI > 1.64). Results were adjusted for a number of confounders (e.g. age, birth weight, mother’s
age, maternal smoking during pregnancy, ethnicity, maternal educational attainment, physical
activity, depressive symptoms, and IQ scores). Analyses showed negative associations between
weight status and attainment which became attenuated with the inclusion of confounders, and
largely insignificant in the case of boys. In girls, overweight or obesity showed significant
independent effects, and long-term overweight or obesity rather than high weight status in the short
term, was particularly problematic. Booth et al. (2014) concluded that for girls, these results suggest
that the relationship between obesity and subsequent academic attainment is likely to be causal
(which stands in contrast to Sassi et al., 2009, above). They commented that the inclusion of
measures of self-esteem, and changes in psychological wellbeing over time, should be considered in
future analyses in this area.
3.2. Evidence in JANPA WP4 countries
3.2.1. Overview
In total, 81 sources were received from JANPA participants that examined health and other impacts
of childhood overweight/obesity occurring in childhood. The distribution of these papers across
countries is shown in Table 3.1. A large majority of these – 93% – examined health impacts, while
only 10% examined other impacts (four of the 81 sources covered multiple areas). Almost a third of
materials came from Greece, 18.5% from Italy, 17% from Romania, 11% from Croatia, about 9% from
each of Ireland and Portugal, and 4% from Slovenia.
Table 3.2 shows the specific topics covered in these 81 sources. A majority of sources covered
aspects of cardio-metabolic health (69%), including multiple aspects of the metabolic syndrome
(36.5%), blood pressure (13%) and diabetes or blood glucose profiles (13%), with smaller numbers of
papers examining specific aspects of cardio-metabolic health or risk factors, including liver
abnormalities and arterial thickness. About 9% of sources examined aspects of children’s musculo-
skeletal or motor functioning, and 6% looked at pulmonary function or aerobic capacity. One or two
sources covered each of dental health, hormonal health (in girls), and idiopathic intracranial
hypertension. A majority of the nine sources that examined other impacts covered aspects of
psychological or emotional wellbeing, while only one source examined the association between child
overweight/obesity and academic performance, and one examined subjective quality of life.
93
Table 3.1. Summary of materials on health and other impacts of child overweight and obesity in
childhood from JANPA participants, by country
Country Health impacts Other impacts Total sources
N % N % N %
Croatia 9 11.1 0 0.0 9 11.1
Greece 22 27.2 4 4.9 26 32.1
Ireland 6 7.4 1 1.2 7 8.6
Italy 15 18.5 0 0.0 15 18.5
Portugal 6 7.4 2 2.5 7 8.6
Romania 14 17.3 1 1.2 14 17.3
Slovenia 3 3.7 0 0.0 3 3.7
Total 75 92.6 8 9.9 81 100.0
Table 3.2. Summary of materials on health and other impacts of child overweight and obesity in
childhood from JANPA participants, by topic
Area/sub-area N %
Health Impacts 76 89.4
Cardio-metabolic health 59 69.4
Arterial thickness 2 2.4
Blood pressure 11 12.9
Diabetes/Glucose profile 11 12.9
Iron levels 2 2.4
Liver abnormalities 2 2.4
Metabolic syndrome 31 36.5
Dental health 2 2.4
Hormonal/Reproductive health 1 1.2
Idiopathic intracranial hypertension 1 1.2
Musculo-skeletal/Motor 8 9.4
Pulmonary/Aerobic 5 5.9
Other Impacts 9 10.6
Academic 1 1.2
Psychological/Emotional 7 8.2
Quality of life 1 1.2
Total 85 100.0
Four sources are counted twice in the table as they cover two of the topics listed.
Many of the studies relied on clinical samples (e.g. children referred to a nutrition or obesity clinic)
so small, non-representative samples are common in these sources. Consistent and strong evidence
for negative impacts on child and adolescent cardio-metabolic profiles is evident in almost all
countries. There is also reasonably consistent, though less widespread evidence, for negative
impacts on child/adolescent musculo-skeletal/motor and pulmonary/aerobic functioning. The
relatively small number of studies that examined emotional or psychological impacts are difficult to
compare due to differences in measures and analysis methods, but they suggest negative
associations (which probably operate bi-directionally) between measures of psychological and
emotional wellbeing and overweight and obesity. Table A10 (Appendix 2) provides details of each of
these studies. A brief summary for each country is provided below.
3.2.2. Croatia
Nine papers from Croatia examined health impacts of overweight/obesity during childhood. Two of
these examined trends in the incidence of type 1 diabetes in children. Stipancic et al. (2008)
estimated a 9% average annual increase in incidence for the period 1995-2003, while Putarek et al.
(2015) estimated a 6% average annual increase from 2004-2012. While not empirically linked to
rates of overweight or obesity, they nonetheless show a worrying trend.
Two papers examined cardio-metabolic risk factors. Ille et al. (2012) found, in a sample of children
and adolescents, all with BMI > 90th percentile, that 10.4% had impaired glucose tolerance, 17.3%
had increase cholesterol, and 30.1% had elevated triglyceride levels. Musil et al. (2012) reported
significant associations between raised blood pressure and overweight/obesity among 8th grade
adolescents.
A further three sources examined associations between BMI and musculo-skeletal/motor and
aerobic function. Delas et al. (2008) tested adolescents (mean age 13 years) on speed, power,
reaction time and balance. Among overweight/obese boys tested, motor performance was
significantly lower on all tests than healthy weight boys except balance, while in girls, only lower leg
repetitive movement was significantly lower. Bozanic et al. (2011) reported lower performance on
tests of speed and agility among overweight and obese 7 year-old children of both sexes compared
to healthy weight children. Kunjesic et al. (2015) found that higher BMI was significantly associated
with lower aerobic capacity among children aged 7 to 11 years.
Croatia is the only country for which information was located on two further areas – idiopathic
intracranial hypertension (IIH), and hormonal/reproductive health. In a clinical sample of children
(mean age 10.7 years), Sindicic Dessardo et al. (2010) reported that 75% of children suffering from
IIH were overweight or obese. Bralic et al. (2012) reported a significant association between early
onset of menarche and overweight/obesity43.
3.2.3. Greece
A relatively large number of sources on impacts of childhood overweight/obesity were retrieved for
Greece. Nineteen of these covered aspects of cardio-metabolic health. Of these, 10 examined
multiple risk factors associated with the metabolic syndrome (Kollias et al., 2011; Mazaraki et al.,
2011; Hatzis et al., 2012; Kollias et al., 2013; Mirkopoulou et al., 2010; Manios et al., 2004; Sakka et
al., 2015; Papadopoulou-Alataki et al., 2004; Lydakis et al., 2012; Sakou et al., 2015). For example, in
a sample of 17 year-olds, Mirkopoulou et al. (2010) reported that central obesity increased the
chances of impaired fasting glucose (eight-fold) and doubled the prevalence of dyslipidemia and
elevated serum cholesterol. Among younger children (age 4-7 years) all with a waist circumference >
90th percentile, Hatzis et al. (2012) found 77% had an increment in at least one risk factor for
atherogenesis. The factors with the highest prevalence were overweight (18.1%) and obesity (9.9%)
followed by hyperlipidemia (about 15%) and hypertension (7.7%). Mazaraki et al. (2011) reported a
43
Early onset of menstruation is a risk factor for breast cancer (Collaborative Group on Hormonal Factors in Breast Cancer, 2012).
95
significant negative relationship between BMI and albumin to creatinine ratio, (ACR, an indicator of
risk of diabetes and hypertension) among adolescents aged 12-17 years.
Three more of these 19 papers examined blood pressure (BP) (Mavrakanis et al., 2009; Angelopoulos
et al., 2009; Kollias et al., 2009). For example, Mavrakanis et al. (2009) reported that 7.9% of a
sample of children aged 4-7 years had elevated systolic or diastolic BP (≥95th percentile), and that
this was more common in obese children, from 17.8% to 27.5% depending on the method used to
define obesity.
Two papers examining the associations between insulin resistance/impaired glucose tolerance found
strong, significant associations with measures of adiposity in children (Xekouki et al., 2007; Manios
et al., 2007). For example, among children aged 10 to 12 years, insulin resistance (IR) was 5-10 times
higher in obese compared to healthy weight children (Manios et al., 2007).
The four remaining papers covering aspects of cardio-metabolic health examined liver abnormalities
and iron deficiency. Papandreou et al. (2008, 2012) reported a prevalence of around 42% for non-
alcoholic fatty liver disease among a sample of obese children and adolescents aged 8-15 years.
Moschonis et al. (2012) and Manios et al. (2013) reported associations between adiposity and iron
deficiency. For example, among children aged 9 to 13 years, Manios et al. (2013) reported odds
ratios for iron deficiency and iron deficiency anaemia were 2.46 and 3.13 in obese boys and 2.05 and
3.28 in obese girls relative to healthy weight children44.
Spathopoulos et al. (2009) examined lung function among children aged 6-11 years and found that
BMI remained an independent risk factor for reduced lung function, asthma and atopy. Trikaliotis et
al. (2011) found that overweight Greek pre-school children were at a significantly higher risk of
dental caries (with a mean of 1.88 caries in the overweight group compared with 0.74 caries in the
healthy weight group).
Three sources from Greece concerned psychological/emotional impacts of childhood
overweight/obesity (Pervanidou et al., 2013, 2015; Koroni et al., 2009). For example, in a sample of
110 obese and 31 healthy weight children (mean age 11.2 years), Pervanidou et al. (2013) found that
obese children were 3.1 and 2.3 times more likely to report state and trait anxiety, respectively, and
3.6 times more likely to report depressive symptoms, than healthy weight children of the same age.
Greece was the only country for which a study on academic performance was retrieved. In a sample
of children aged 10 to 12 and using multiple regression, Vassiloudis et al. (2014) found that academic
performance was significantly associated with BMI, dietary quality, TV viewing, sleep, physical
activity, parents' education, mother's ethnicity and family income. Note, however that academic
performance in this study was based on teachers’ ratings rather than standardised test results.
3.2.4. Ireland
Three papers from Ireland examined aspects of cardio-metabolic health. Finucane et al. (2008a)
reported that 51% of boys and 49% of girls had systolic BP in hypertensive range (> 95th percentile
44
It is thought that sub-clinical inflammation plays a central role in the association between iron deficiency and
overweight; i.e. hepcidin levels are higher in obese individuals and are linked to subclinical inflammation; this may reduce
iron absorption and blunt the effects of iron fortification (Cepeda-Lopez et al., 2010).
for age, sex and height). Results also showed a clear and continuous increase in systolic BP with
increasing BMI, particularly in boys. This is of significance, since 93% of this sample (aged 2-18 years)
was obese. Finucane et al. (2008b) reported significant associations between degree of obesity,
insulin sensitivity and markers of liver steatosis among a sample of obese children and adolescents
(mean age 15.5 years). Carolan et al. (2013) reported that obese children showed changes in
immune cell frequency, inflammatory environment, and regulation of metabolic gene expression
compared to children of healthy weight. These changes have been causally linked to adult onset of
metabolic disease and suggest a future trajectory for the development of type 2 diabetes and
premature cardiovascular disease.
Three further papers examined associations between overweight/obesity and musculo-
skeletal/motor function (O’Malley et al., 2012, 2015a, 2015b). For example, in a sample of obese
children and adolescents (mean age 12.2 years), O’Malley et al. (2012) reported moderate negative
correlations were found between body composition and range of motion, flexibility, and strength.
Genu valgum deformity was moderately positively correlated to body mass index.
One source from Ireland examined psychological/emotional impacts. On the basis of a
representative sample of 9 year-olds, Layte & McCrory (2011) reported that self-perceptions relating
to popularity and physical appearance were significantly negatively related to self-perceptions of
weight. The perception of overweight was also significantly associated higher levels of emotional
and behavioural problems.
3.2.5. Italy
Most of the sources from Italy – 14 of 15 – examined aspects of cardio-metabolic risk factors. Eight
of these looked at the metabolic syndrome (DiBonito et al., 2015; Capizzi et al., 2011; Caserta et al.,
2010; Valerio et al., 2013; Invitti et al., 2005; Calcaterra et al., 2008; Invitti et al., 2003; Ianuzzi et al.,
2004). For example, in a sample of children aged 0-14 years, Calcaterra et al. (2008) found that the
prevalence of metabolic syndrome (i.e. three or more of BMI > 97th percentile, triglyceride levels >
95th percentile, high density lipoprotein (HDL) cholesterol level < 5th percentile, systolic or diastolic
Blood pressure > 95th percentile, and impaired glucose tolerance) was 0% in normal and overweight
children, 12.0% in moderately obese and 31.1% in severely obese children. Ianuzzi et al. (2004)
reported that among children/adolescents aged 6-14 years, obese children had significantly higher
BP and plasma concentrations of tryglycerides, cholesterol, glucose, insulin, HOMA and C-reactive
protein than healthy weight children. Carotid intima-media thickness (CIMT) was also significantly
higher in obese children.
A further four papers examined blood pressure (Turconi et al., 2006, 2007; Barba et al., 2006;
Genovesi et al., 2005). For example, Barba et al. (2006) reported that BMI and waist circumference
were independently associated with systolic BP, after adjusting for parental education and children's
levels of physical activity (sample aged 6-11 years).
Bruno et al. (2010) examined trends in type 1 diabetes among children aged 0-14 years from 1990-
2003 and found that the incidence rate was 12.26 per 100,000 person years and significantly higher
in boys (13.13) than in girls (11.35). Incidence rates increased linearly by 15, 27, 35, and 40% across
four successive birth cohorts studied. Note that this trend is not empirically linked with trends in
prevalence of overweight or obesity in the article. In a sample of obese children and adolescents
aged 3-18 years, Brufani et al. (2010) found that glucose metabolism abnormalities were present in
97
12.4%. Impaired glucose tolerance (IGT) was the most frequent alteration (11.2%), with a higher
prevalence in adolescents than in children (14.8 vs. 4.1%).
The final source from Italy considered here examined pulmonary/aerobic function. Eight per cent of
all children (aged 6-7 years) reported current wheezing and 6.7% reported current asthma. Elevated
BMI (comparing highest quintile to others) was significantly associated with both current wheeze
(adjusted odds ratio=1.47) and current asthma (adjusted odds ratio=1.61) (Corbo et al., 2008).
3.2.6. Portugal
Seven sources on health and other impacts of child overweight and obesity in childhood were
retrieved for Portugal. Two of these examined the metabolic syndrome in children (aged 7-9 years;
Pedrosa et al., 2010) and adolescents (mean age 13.2 years; Teixera et al., 2001). Pedrosa et al.
(2010) reported that presence of metabolic syndrome (MS), i.e. three or more of abdominal obesity,
high fasting triglycerides, low HDL, high BP, and high fasting glucose, was significantly associated
with higher BMI, while Teixera et al. found that both direct and indirect measures of adiposity were
associated with serum cardiovascular risk factors in boys and girls. Leite et al. (2012) reported that,
among a sample of children/adolescents (mean aged 12.9 years), CIMT was positively associated
with higher BMI, even in moderately overweight ranges, independent of age, gender, systolic blood
pressure and plasma lipid concentrations. Ribeiro et al. (2003) found that systolic and diastolic blood
pressure were significantly and positively related to BMI among a sample of 8-16 year-olds (all at risk
of obesity).
Lopes et al. (2011) found that motor co-ordination was inversely associated with BMI: the strength
of the association increased during childhood but decreased into early adolescence; however at all
ages, overweight and obese children had significantly lower motor co-ordination than healthy
weight children.
Two sources from Portugal examined psychological/emotional impacts (Ferreira Felgueiras, 2011;
Moreira et al., 2013). The study by Moreira et al. is of note since it allows comparisons of healthy
children and adolescents with children/adolescents with various conditions including obesity.
Participants in their study were classified as healthy, with diabetes, asthma, epilepsy, or obesity.
Children with obesity and epilepsy reported the lowest quality of life and highest levels of
psychological problems, and parents of obese children reported the lowest quality of life.
3.2.7. Romania
Fourteen sources were retrieved for Romania. Eight of these examined aspects of the metabolic
syndrome (Morea et al., 2013; Pelin & Matasaru, 2012; Casariu et al., 2011; Popescu et al., 2013;
Gherlan et al., 2012; Valean et al., 2010; Chesaru et al., 2013; Brumariu et al., 2007). For example,
Chesaru et al. (2013) studied a sample of overweight/obese adolescents (mean age 13 years) and
found that 37.4% exhibited one metabolic syndrome (MS) diagnosis criterion, 27.6% had two MS
diagnosis criteria, 20.9% combined three criteria, while 8.6% had four or five of the criteria. The
most common risk factors were abdominal obesity (75.5%) and high blood pressure (41.1%),
followed by low HDL-cholesterol (35%), increased fasting blood glucose (23.3%) and
hypertriglyceridemia (17.8%).
A further four sources examined diabetes/glucose profiles of children/adolescents. One of these
consisted of an analysis of the paediatric diabetes registry 2002-2011 (Serban et al., 2015). The
incidence of type 1 diabetes increased significantly from 6.2 to 9.6 per 100,000. Note that this trend
is not empirically linked with trends in prevalence of overweight or obesity in the article. The other
three articles examined diabetes/glucose profiles and their associations with overweight/obesity
(Dumbrava et al., 2012; Mihai et al., 2011; Marginean et al., 2010). For example, in a sample of 9-18
year-olds, all overweight or obese, Dumbrava et al. (2012) reported that 41.4% had prediabetes and
that this was higher in obese (50.7%) than overweight (10.0%) children/adolescents, and also higher
in pre-teens than adolescents (44.8% vs 34.5%).
Chirita-Emandi et al. (2013) examined the blood pressure profiles of children/adolescents aged 7-18
years and found that three times as many obese participants (21.1%) than healthy weight
participants (7.1%) had hypertension, while hypertension was present in 12.8% of overweight
participants.
AnaMaria et al. (2015) examined the dental health of children (mean age 9.1 years) and reported
that dental caries were significantly higher among underweight than overweight children; they
commented that associations between malnutrition and dental caries should be examined further.
Finally, one study from Romania included information on the psychological/emotional health of
children and adolescents (Marginean, 2010). This is a poster presentation, so details are lacking, but
the study was based on all children admitted to a paediatric hospital 2004-2009 and indicates that
rates of depression and social isolation were very high among the obese adolescents admitted.
3.2.8. Slovenia
In Slovenia, three papers were retrieved. Two of these examined musculo-skeletal/motor
performance and pulmonary/aerobic function (Leskosek et al., 2007; Matejek et al., 2014). For
example, among 7-18 year-olds, Leskosek et al. (2007) reported that the performance in almost all
the fitness tests administered was substantially hindered (or at least had a negative correlation) with
obesity – regardless of the age or sex of the children. The greatest influence of obesity was found in
tests requiring movement of the whole body. The third study (Mocnik et al., 2015) reported an
association between less compliant arteries and childhood obesity and hypertension.
99
Tables
Table T3.1. Summary of main findings of 18 review studies on impact of overweight/obesity in childhood identified by Queally et al. (2016)
First author, year
Condition(s) Systematic
review Meta-
analysis Study parameters Main findings
Pulgarón, 2013
Multiple medical and psychological comorbidities
x
79 studies (majority cross-sectional): 2002-2012, age 0-18 years, any study examining association between OW/OB and comorbidity. Review is more narrative than systematic as effect sizes and other aspects of the studies are not reported in detail.
There is substantial support for cardio-metabolic risk factors, internalizing disorders, ADHD, and decreased health related quality of life as comorbidities to obesity in childhood. However, "...there are so many potential confounds and so much interdependency among the co-morbidities that it is difficult for researchers to isolate the effects of childhood obesity. ... one of the greatest challenges within the psychological domain is the definition of the psychological variable of interest." (p. 7). Of medical co-morbidities, 35 studies examined cardio-metabolic risk factors, 15 asthma/asthma symptoms, 3 dental health, 6 musculo-skeletal, 12 sleep disorders/problems, and 3 airway hyper-responsiveness/obstruction. Small numbers of studies examined other conditions such as athancosis nigricans, headaches and sexual maturation. Of psychological co-morbidities, 7 examined anxiety/depression, 5 ADHD, and 5 other behavioural or externalising problems. Smaller numbers of studies covered topics such as bullying and disordered eating. Some key additional points include: the association between obesity and asthma may be due at least in part to an increase in diagnosis and causal relationships are not clear; the degree of OW/OB is associated with degree of cardio-metabolic risk factors; the relationship between caries and OW/OB may vary by age and other measures such as diet and SES; associations between ADHD and OW/OB are stronger for clinical diagnoses than self-reports; the relationships between OW/OB and depression is not consistent across sub-groups; and the association between OW/OB and short sleep duration is consistent but the long-term effects of this are unclear.
First author, year
Condition(s) Systematic
review Meta-
analysis Study parameters Main findings
Sanders, 2015 Multiple medical and psychological comorbidities
x
47 Australian studies (29 cross-sectional, 17 cohort, 1 case control; 26 physical, 16 psychological, 5 both): 2004-2014, ages 0-18, investigating obesity-related co-morbidities with clearly reported metric for weight status.
Main conclusion: "Evidence suggests that overweight/obese Australian children and adolescents, compared to normal-weight peers, had more cardio-metabolic risk factors and higher risk factors of non-alcohol fatty liver disease and were experiencing more negative psychological outcomes (depression, low self-esteem and lower scores of health-related quality of life)." (p. 1) Many other conditions have not been studied extensively. Cardio-metabolic risk factors were the most frequently examined (15 studies). For example, one study found that, compared to normal-weight peers, obese adolescents (aged 15.4±0.4 years) were significantly more likely to have two or more risk factors for heart disease, type II diabetes and fatty liver disease (boys 73.5% vs 7.6%; OR, 34.0 [95% CI, 12.6-91.7]; P < .001; girls 44.4% vs 5.4%; OR, 14.0 [95% CI, 4.1-47.5]; P < .001). All 5 studies examining NAFLD found significant associations. For example, one of these 5 studies found that NAFLD increased with increase of adiposity among normal-weight, overweight and obese boys and girls aged 17 years (boys, 4, 15 and 65 %; girls, 10, 29, and 57%, respectively). The severity of hepatic steatosis was also associated with the body mass index, WC and subcutaneous adipose tissue thickness (p<0.001) in this study. Four of 6 studies examining asthma/asthma symptoms reported significant associations. For example, parents of OW (OR=1.30, 95 % CI= 1.16, 1.46) and OB (OR=1.36, 95 % CI=1.13, 1.62) children aged 4-6 years were significantly more likely to report that they had asthma ever than parents of HW children. Four studies examining obstructive sleep apnea indicated conflicting results, and the pattern suggests that the association may be stronger among adolescents than younger children. Two studies in this review examined musculoskeletal pain; both reported significant associations (e.g. relative to NW, OR OW=1.53; OR OB=4.09; p=0.010). Overweight/obese children and adolescents showed lower HRQoL than normal-weight peers in all 12 included studies (and consistent results emerged regarding physical, emotional and social quality of life). Evidence suggests that the strength of this association increases with age. Nine studies examining mental health were also consistent in their reports of associations between child/adolescent weight status and depression. For example, one study reported odds ratios of depression in OW/OB children aged 6-13 years relative to HW as follows: (OR=3.38, 95 % CI=1.13– 10.11; overweight, OR=8.95; obese, OR=18.8; p=0.001). However, studies examining gender differences had varying results. Only 1 of 3 studies examined reported significant associations between anxiety and weight status. All four studies on self-esteem found significant associations with weight status. For example, one study reported that obese children (aged 9.2–13.7 years) were between two and four times more likely to have lower global self-worth than normal-weight peers. The long-term impact of childhood OW/OB on physical comorbidities (i.e. blood pressure, diabetes, asthma, cardio-metabolic health and cardiac structure) in adulthood was reported in 5 studies. For example: Greater BMI z-score (odds ratio (OR)=1.48, 95 % confidence interval (CI)=1.11–1.96) or being overweight at 5 years (OR=2.23, 95 % CI=1.08–4.60) was found to increase the likelihood of type 1/2 diabetes at 21 years in one study; in another study, compared with children who were in the lowest 25% for WC, those in the highest 25% were 5-6 times more likely to be classified with metabolic syndrome at age of 26–36 years; in a third study, childhood BMI (male, β=0.41, 95 % CI=0.14–0.67; female, β=0.53, 95 % CI=0.34–0.72) and change in BMI from childhood to adulthood (male, β=0.27, 95 % CI=0.04– 0.51; female, β=0.39, 95 % CI=0.20–0.58) were positively associated with left ventricular mass in adulthood, which increases risk of myocardial infarction, congestive heart failure and cardiovascular disease mortality.
101
First author, year
Condition(s) Systematic
review Meta-
analysis Study parameters Main findings
Kelishadi, 2015
Cardio-metabolic risk factors
x
61 studies: Date parameters not specified; abdominal obesity (not secondary to other disease) and any of systolic BP diastolic BP, prehypertension, transient hypertension, cholesterol, LDL-C, HDL-C, fasting blood sugar, insulin resistance, insulin dose per body surface, carotid intima-media thickness, and alanine aminotransaminase, ages 6-18 years.
"Whatever the definition used for abdominal obesity and whatever the methods used for anthropometric measurements, central body fat deposition in children and adolescents increases the risk of cardio-metabolic risk factors." BP was the most common measurement among studies; most of them confirmed the association of abdominal obesity and elevated BP. Reasonably consistent evidence was also found between abdominal adiposity and abnormal lipid profile and fasting glucose.
Friedemann, 2012
Cardio-vascular risk factors
x x
39 studies in descriptive analysis and 24 studies included in meta-analysis: healthy children aged 5-15 years, in developed countries, minimum sample size of 20. Objective measure of weight and one or more of: systolic or diastolic BP, HDL, LDL or total cholesterol, triglycerides, fasting glucose or insulin, HOMA-IR, carotid intima media thickness, left ventricular mass.
In the 39 papers selected for descriptive analysis: general risk parameters for CVD were worsened with increasing BMI. BMI was positively associated with: systolic BP in five studies, diasystolic BP in four studies, total cholesterol in one study, LDL cholesterol in 3 studies, tricglycerides in three studies, and left ventricular mass in one study. BMI was negatively associated with HDL cholesterol in two studies. BMI was also associated with CVD risk clustering (6 studies). In meta-analysis, the mean values of diasystolic, systolic and ambulatory BP, total, HDL and LDL cholesterol and triglycerides, and fasting glucose, fasting insulin and HOMA-IR, and CIMT and left ventricular mass were computed for NW, OW and OB. In all cases differences were statistically significant, with larger differences in comparisons of OB-NW than in OW-NW. The evidence indicates that risk factors can track into adulthood. However, it is unclear whether the magnitude of difference in CVD risk in NW, OW and OB children continues unchanged into adulthood.
Verbeeten, 2011
Type 1 diabetes x x
9 studies (8 case control and 1 cohort study): up to February 2010, age 0-18 years, measurement of weight status prior to diagnosis of T1 DM.
Meta-analysis of 4 of these studies yielded a pooled odds ratio of 2.03 for obesity compared to healthy weight (95% CI 1.46-2.80) and meta-analysis of 5 of these studies yielded a pooled odds ratio of 1.25 per 1 SD increased in BMI (95% CI 1.04-1.51). The restriction of studies to those where weight status was assessed prior to diagnosis provide confirmation of a causal relationship. Studies varied in the age of obesity assessment, however.
Anderson, 2015
NAFLD x x
74 studies, divided into general and clinical (OW/OB) populations: Ages 1-19, all studies up to Oct 2013 measuring prevalence of NAFLD (multiple methods), excluding previous or existing liver disease.
Pooled prevalence in general populations: 9.0% males, 6.3% females, 2.3% NW, 12.5% OW, 36.1% OB. (Pooled prevalence in obese clinical populations: 35.3% males, 21.8% females.) Prevalence did not vary by diagnostic method, age of sample, publication year, location or sample size. Meta-analysis of available within-study comparisons provided strong evidence that prevalence is higher on average in males compared with females and increases incrementally with greater BMI. However, these associations also varied considerably across studies. Authors did not have sufficient information on the distribution of ethnicity in each study to assess whether NAFLD prevalence differed between ethnic groups.
First author, year
Condition(s) Systematic
review Meta-
analysis Study parameters Main findings
Hayden, 2013 Dental health x x 14 studies: 1980-2010, measured BMI and dental caries, ages 1-18.
Overall, a significant relationship between childhood obesity and dental caries (effect size = 0.104, P = 0.049) was found. Results tended to be significant on the basis of standardised BMI comparisons such as BMI-for-age centiles (effect size = 0.189, 95% CI: 0.060–0.318, P=0.004) or IOTF cut-offs (effect size = 0.104, 95% CI: 0.060– 0.180, P=0.008). Studies that used Zscores (effect size = 0.147, 95% CI: 0.396 to 0.102, P=0.248) provided non-significant findings, along with studies using non-standardized scales (effect size = 0.030, 95% CI: 0.436 to 0.375, P = 0.884). Obese children from industrialized countries (effect size = 0.122, CI = 0.047–0.197, P=0.001) had a significant relationship between obesity and caries in contrast to those from NIC countries (effect size = 0.079, CI = 0.106 to 0.264, P = 0.264). Authors link these findings to diet, including consumption of sugar-sweetened drinks, and highlight the need for future work to measure caries in standardised ways as well as analyse multiple confounding factors including SES.
Hooley, 2012 Dental health x
48 studies: 2004-2011, measured caries, measured BMI, ages 0-18 years.
Authors note that dental disease ranks as the second most expensive disease in Australia (with CVD as the most expensive). Given that BMI and dental health both relate to diet, an association between the two is not surprising. 23 studies found no association, 17 found a positive association, 9 reported an inverse relationship, and 1 reported a U shaped pattern of association. Studies reporting a positive association were from countries with a higher Human Development Index (HDI) score (mainly Europe and US), higher quality dental services (more sensitive dental examination) and a low % of UW children, while studies reporting a negative association were from countries with a lower HDI score (mainly Asia and South America), lower quality dental services (less sensitive dental examination), and more UW children. Authors recommend more longitudinal research that includes health and diet behaviours.
Hendrix, 2014 Developmental coordination disorder
x
21 studies (10 cohorts): ages 4-14 years, measured body composition (BMI, % body fat, waist circumference), comparison of DCD with ND children.
Authors note that prevalence of DCD (poor fine and/or gross motor coordination) is estimated to range from 1.7% to 6%, and in boys is found four to seven times more often than in girls. It is a chronic condition. All 21 studies in reported that children with DCD or pDCD had higher BMI scores, larger WC and greater percentage BF compared with their TD peers. Eighteen studies (7 cohorts) found these differences between groups to be statistically significant. Fourteen of 17 studies that used BMI reported significant differences. Evidence of gender differences was weak or inconclusive. There was some evidence of an increased risk of OW and OB associated with DCD over time.
103
First author, year
Condition(s) Systematic
review Meta-
analysis Study parameters Main findings
Mebrahtu, 2015
Asthma/wheezing disorders
x x 38 studies: original reports on childhood wheezing disorders and BMI, covering 0–19 year-olds, published until May 2014
Authors note that previous meta-analyses on this topic were based on studies from different age groups (child and adult combined) and different definitions of weight status. The summary ORs of underweight (<5th percentile), overweight (>85th to <95th percentile) and obesity (≥95th percentile) were 0.85 (95% CI: 0.75 to 0.97; p = 0.02), 1.23 (95% CI: 1.17 to 1.29; p < 0.001) and 1.46 (95% CI: 1.36 to 1.57, p < 0.001), respectively. Heterogeneity was significant and substantial in all three weight categories, and not accounted for by pre-defined study characteristics. Studies classified BMI differently - IOTF, US-CDC, WHO, and data-driven definitions. The summary ORs estimates tended to attenuate as the number of BMI categories used by study authors increased. Summary risk estimates of the cohort and cross-sectional studies are very similar, both for the overweight and obesity risk estimates. ORs did not vary by age.
Paulis, 2013 Musculo-skeletal complaints
x x
40 studies (33 cross-sectional, 7 longitudinal): up to May 2013, age 0-19 years, objective assessment of weight status and musculo-skeletal complaints (malalignments defined as pes planus, scoliosis and tibia vara not included).
There was moderate quality of evidence that being overweight in childhood is positively associated with musculoskeletal pain (risk ratio [RR] 1.26; 95% confidence interval [CI]: 1.09–1.45). In addition, low quality of evidence was found for a positive association between overweight and low back pain (RR 1.42; 95% CI: 1.03–1.97) and between overweight and injuries and fractures (RR 1.08; 95% CI: 1.03–1.14). Although the risk of developing an injury was significantly higher for overweight than for normal-weight adolescents (RR: 2.41, 95% CI: 1.42 to 4.10), this evidence was of very low quality. Authors comment that "The link between overweight and MSC might induce a vicious circle in which being overweight, musculoskeletal problems, and low fitness level reinforce each other." (p. 13) They recommend more high-quality longitudinal research.
Smith, 2014 Musculo-skeletal pain
x
10 studies: 2000-2012, ages 3-18, associations between weight status and musculo-skeletal pain (back, knee, hip, foot, and pelvic)
Narrative synthesis is provided and effect estimates are summarised. For example, one US study of children aged 3-18 reported the following: Knee: OR=1.13 per 10kg increase in weight, 95% CI: 1.01-1.29. OR=1.04 per unit increase in BMI, 95% CI: 1.01-1.08 Hip: OR=1.29 per 10kg increase in weight, 95% CI: 1.05-1.60. OR=1.09 per unit increase in BMI, 95% CI: 1.03-1.16. Main conclusion: "...emerging evidence suggests that being overweight or obese has a significant impact on the health and well-being of these young people and may contribute to ongoing health problems such as musculoskeletal pain and bone/joint dysfunction in later life. The cumulative effect of children being overweight or obese and experiencing musculoskeletal pain requires further investigation" (p. 15)
Stolzman, 2015
Pes planus (flat feet)
x
13 cross-sectional studies: to September 2013, ages 3-18 years
Prevalence of pes planus varied widely, from 14-67% in population studies, but all studies showed an increased prevalence of pes planus in obese or overweight children. However, "a longitudinal, randomized control trial is necessary to declare a causal relation between a high BMI and pes planus" (p. 5)
First author, year
Condition(s) Systematic
review Meta-
analysis Study parameters Main findings
Thivel, 2016 Muscle strength and fitness
x 36 studies of mixed design, to June 2015, ages 6-18 years
Laboratory results yield similar results to field studies when adjustments are made for body mass. Overall, review provides strong evidence that children and adolescents with obesity have reduced muscular fitness compared with children and adolescents of healthy weight. More research is needed on muscular and musculoskeletal fitness given their associations with overall health. “Improving muscular fitness and overall musculoskeletal fitness in children with obesity is of crucial importance to favour their physical autonomy, promote engagement in daily activities and physical activity-based weight management programmes, and subsequently improve their health-related quality of life” (p. 61).
Raj, 2012 Cardio-vascular risk
x (narrative)
N/A
The review considers some of the evidence for childhood OW/OB and hypertension and associations with: metabolic syndrome/clustering of cardiovascular risk factors, insulin resistance/type 2 diabetes, inflammation and oxidative stress, artherogenic dyspipidemia and atherosclerosis, cardiac structure/function, and sleep disorders. A medical perspective is taken. The author cites findings from the Avon longitudinal study of 5235 children (Lawlor et al. 2010) which reported that, in girls, a 1 standard deviation (SD) increase over mean BMI during 9–12 years was associated with cardiovascular risk factors at age 15–16 years in fully adjusted models, with odds ratio of 1.23 for high systolic BP (≥130 mm Hg); 1.19 for LDL-C (≥2.79 mmol/l); 1.43 for high triglycerides (≥1.7 mmol/l); 1.25 for low HDL-C (<1.03 mmol/l); and 1.45 for high levels of insulin (≥16.95 IU/l). The corresponding values in boys were 1.24 for systolic BP, 1.30 for LDL-C, 1.96 for triglycerides; 1.39 for HDL-C, and 1.84 for high insulin levels.
Griffiths, 2010
Self-esteem and quality of life
x
9 cross-sectional studies on self-esteem (5 children, 1 adolescents, 3 both), 15 studies on quality of life (6 children, 4 adolescents, 5 both): up to March 2009, comparisons with OW/OB and NW, ages 0-18 years, validated measures of self-esteem/quality of life, measured or reported BMI.
Six of nine studies found lower global self-esteem in OB compared to HW children and adolescents. Four of five studies that incorporated a self-esteem dimension within quality of life scales reported significantly lower scores in their OB compared to NW. Lower total quality of life was found in 9 of 11 studies using child self-report, and 6 of 7 studies using parental report. OB had the greatest impact on physical functioning and physical appearance perceptions, as well as social functioning. There was limited evidence that presence of OB with a severe medical condition did not impact significantly, while presence of OB with a less severe medical condition did impact significantly on quality of life. There was not sufficient information to make detailed comparisons between genders or ethnic groups.
Muhlig, 2016 Depression x
24 studies (19 cross-sectional and 8 longitudinal, some of the 24 mixed in design): up to August 2014, objectively measured weight status, validated measure of depression, age up to 18 years.
14 of the 19 cross-sectional studies confirmed an association between weight status and depression while 3 of 8 longitudinal analyses confirmed a significant association (in females and adolescents only). Authors hypothesise the joint development of OW/OB and depression over time. However, in children and adolescents, longitudinal studies are still too few to permit estimation of the effect sizes of bidirectional associations.
105
First author, year
Condition(s) Systematic
review Meta-
analysis Study parameters Main findings
Rees, 2011 Weight stigmatisation
x
28 UK studies (15 included in interpretative synthesis, 13 included in aggregative synthesis): 1997-June 2009, ages 4-14, views on obesity, body size, shape or weight from children.
The study provides a narrative synthesis and the authors note that many of the studies were not of high quality. Key conclusions: "being overweight was seen as a problem because of the impact it could have on their lives as social beings, from reduced popularity through to discrimination. The health consequences of obesity appeared to be largely irrelevant" (p. 9). Body dissatisfaction and aspiration to thinness were extremely common, and even more so in girls than boys. In several of the studies, OW/OB was blamed on the individual and seen as something for individual control. The very overweight children in the review described being teased and bullied and reported how this impacted seriously on their wellbeing and behaviour. Authors suggest that research in this area needs to be much more rigorous and representative of children's own views as well as a more diverse range of children.
Table T3.2. Summary of odds ratios and confidence intervals for various conditions extracted by
Queally et al. (2016): review on impact of child overweight/obesity during childhood/adolescence
Disease Outcome Odds Ratio 95% (Confidence Interval) Quality of evidence
Asthma Overweight: Adjusted risk: 1.23 (1.17–1.29) Obesity: Unadjusted risk: 1.43 (1.33, 1.54)
Moderate with conflicting findings
Wheezing disorders Overweight: Unadjusted risk: 1.23 (1.17–1.29) Adjusted risk: 1.30 (1.19–1.42) Obesity: Unadjusted risk: 1.46 (1.36–1.57) Adjusted risk: 1.60 (1.42–1.81)
Moderate with conflicting findings
Metabolic syndrome: Study 1: For every one unit increase in zBMI the odds ratio of meeting criteria for metabolic syndrome is 2.4 (1.21–4.63). Study 2: Compared with healthy weight children, overweight and obese children were more likely to have MetS (overweight: 67.33, (21.32–212.61); obesity: 249.99, (79.51–785.98)
Good but often defined by different criteria
High blood pressure Study 1: 4.11(3.89–4.34) and 5.56 (5.09–6.07) for obese male and female subjects, respectively Study 2: Obese youth are twice as likely to have hypertension (for SBP > 140, 2.24; (1.46 – 3.45), and for DBP 2.10: (1.063–4.17)
Good
Type 2 diabetes 5.56 (5.09–6.07) and 4.42 ( 3.90 – 5.00) for obese male and female, respectively)
Moderate
Hyperlipidemia 16.07 ( 8.29 – 31.15) and 9.00 ( 4.36–18.6) for male and female subjects, respectively
Moderate
Other
Depression Overweight/obese children (aged 6–13 years) more likely to suffer from depression than normal-weight children 3.38, (1.13–10.1)
Moderate
Musculoskeletal pain Risk ratio (RR) 1.26; (1.09-1.45). Good
Obstructive sleep apnoea
Adolescents at ages 12+ years 3.55, (1.30–9.71), but not among younger children
Moderate
Non-Alcohol Flamatory Liver Disease (NAFLD)
Overweight 13.36 (9.09- 18.02) and for obese compared with healthy weight 13.74 (9.92-19.03)
Moderate
107
Table T3.3. Summary of publications examining health and other impacts of child overweight/obesity occurring in childhood for JANPA participants
Country First author, year Age group/Sample Health impacts
Other impacts
Area Sub-area Key findings
Croatia Musil, 2012 8th grade x Cardio-metabolic health Blood pressure Prevalence of overweight was higher among boys and girls in high normal and elevated blood pressure (BP) category than in those with normal BP.
Croatia Bralic, 2012 9 to 16 years, females
x Hormonal/Reproductive health
Hormonal/Reproductive health
Girls who experienced early menarche were significantly more often overweight/obese. Overweight/obesity may be considered as one of the predictors for the early occurrence of menarche.
Croatia Stipancic, 2008 0 to 14 years x Cardio-metabolic health Diabetes/Glucose profile
Incidence of Type 1 diabetes for the whole age group was 8.87 per 100,000 person-years, which represents a 9% average annual increase 1995-2003. Note that this trend is not empirically linked with trends in prevalence of overweight or obesity in the article.
Croatia Putarek, 2015 0 to 14 years x Cardio-metabolic health Diabetes/Glucose profile
Incidence of Type 1 diabetes for the whole age group was 17.73 per 100,000 person-years, which represents a 5.9% average annual increase 2004-2012. Note that this trend is not empirically linked with trends in prevalence of overweight or obesity in the article.
Croatia Sindicic Dessardo, 2010
Mean age 10.7 years, clinical sample
x Idiopathic intracranial hypertension
Idiopathic intracranial hypertension
Idiopathic Intracranial Hypertension clinical sample, where prevalence of overweight and obesity was very high - about 75% - relative to the general population.
Croatia Delas, 2008 Mean age 13.1 years
x Musculo-skeletal/Motor Musculo-skeletal/Motor
Children were tested for speed, power, reaction time and balance. In boys, motor performance was lower on all tests other than balance, while in girls, only lower leg repetitive movement was significantly negatively affected, by overweight or obesity.
Croatia Bozanic, 2011 7 years x Musculo-skeletal/Motor Musculo-skeletal/Motor
In the motor areas of speed and agility, as well as in the strength area, significant differences were found between the overweight and obese, as well as between the healthy weight and obese groups of subjects.
Croatia Kunjesic, 2015 7-11 years x Pulmonary/Aerobic Pulmonary/Aerobic Higher BMI was significantly associated with lower aerobic capacity.
Croatia Ille, 2012 1-19 years, BMI > 90th percentile
x Cardio-metabolic health Diabetes/Glucose profile 10.4% of the sample had impaired glucose tolerance, 17.3% had increased cholesterol, and 30.1% had increased triglyceride levels.
Country First author, year Age group/Sample Health impacts
Other impacts
Area Sub-area Key findings
Greece Pervanidou, 2015 Mean age 11.3 years
x Psychological/Emotional Psychological/Emotional
Results show higher levels of depressive and anxiety symptoms and evidence of higher externalising behaviours among obese compared with healthy weight children.
Greece Mavrakanas, 2009
4-10 years x Cardio-metabolic health Blood pressure
7.9% of the sample had elevated systolic or diastolic BP (≥95th percentile). This was more common in obese children, from 17.8% to 27.5% depending on the method used to define obesity.
Greece Kollias, 2011 9 years x Cardio-metabolic health Metabolic syndrome
Overweight/obese children compared with normal-weight children had significantly higher BP, lower high-density lipoprotein cholesterol (HDL-C), and higher triglyceride levels.
Greece Moschonis, 2012 9-13 years x Cardio-metabolic health Iron levels
Percentage body fat and visceral fat mass were positively associated with iron deficiency in both sexes. These associations might be due to the chronic inflammation induced by excess adiposity.
Greece Mazaraki, 2011 12-17 years x Cardio-metabolic health Metabolic syndrome
There was a significant negative relationship between BMI and albumin to creatinine ratio (ACR, an indicator of risk of diabetes and hypertension) as well as between waist circumference and ACR.
Greece Angelopoulos, 2009
5th grade x Cardio-metabolic health Blood pressure
Intervention study: favourable effects were observed in the intervention group for both diastolic and systolic BP which was attributed to the reduction observed in BMI values.
Greece Hatzis, 2012 4-7 years, WC > 90th percentile
x Cardio-metabolic health Metabolic syndrome
77% of children in the sample had an increment in at least one risk factor for atherogenesis. The factors with the highest prevalence were overweight (18.1%) and obesity (9.9%) followed by hyperlipidemia (about 15%) and hypertension (7.7%).
Greece Kollias, 2009 12-17 years x Cardio-metabolic health Blood pressure
In multiple regression, BMI predicted high systolic BP in both boys and girls. Prevalence of high BP in 2007 was higher than in 2004 (and OW/OB higher but not significantly so).
Greece Xekouki, 2007 5-19 years, all obese
x Cardio-metabolic health Diabetes/Glucose profile Prevalence of impaired glucose tolerance in this sample of obese children and adolescents was 14.5%.
109
Country First author, year Age group/Sample Health impacts
Other impacts
Area Sub-area Key findings
Greece Kollias, 2013 8-18 years x Cardio-metabolic health Metabolic syndrome
Central obesity (WC) and systolic BP were independently associated, although modestly, with carotid intima-media thickness (CIMT) values (a marker of cardiovascular disease risk).
Greece Mirkopoulou, 2010
17 years x Cardio-metabolic health Metabolic syndrome Central obesity increased the chances of impaired fasting glucose eight-fold and doubled the prevalence of dyslipidemia and elevated serum cholesterol.
Greece Manios, 2004 Mean age 11.5 years
x Cardio-metabolic health Metabolic syndrome
Overweight and obese children had higher levels of plasma triglycerides (TG) and total cholesterol to HDL-cholesterol (TC/HDL-C) ratio and lower levels of HDL-C and physical fitness compared to their normal-weight peers. Risk factors were stronger in males than females.
Greece Papandreou , 2012
8-15 years, all obese
x Cardio-metabolic health Liver abnormalities
Fatty liver was found in 42.6% of this sample of obese children; BMI and WC were significantly higher among children with severe non-alcoholic fatty liver disease (NAFLD).
Greece Sakka, 2015 6-12 years x Cardio-metabolic health Metabolic syndrome
Plasma Lp-PLA2 levels were significantly higher in obese children compared with normal-weight ones. Lp-PLA2 concentrations were significantly correlated with the BMI z-score. All obese children had Lp-PLA2 levels > 200 ng/mL, which predicts atherosclerosis and a high thromboembolic risk in adults. The positive correlation of Lp-PLA2 with BMI suggests that Lp-PLA2 might be the link between obesity and increased cardiovascular risk.
Greece Papandreou, 2008
9-14 years, all obese
x Cardio-metabolic health Liver abnormalities
41.8% of children had fatty liver (FL). Severe hepatic steaosis was significantly associated with higher BMI. Insulin resistance was also higher in the group with FL (85%) than without FL (65%).
Greece Manios, 2007 10-12 years x Cardio-metabolic health Diabetes/Glucose profile
Insulin resistance (IR) was 5-10 times higher in obese compared to healthy weight children and IR indices were significantly correlated with BMI and WC. In a multiple regression with HOMA-IR as the outcome, significant predictors were sex, simple carbohydrate intake and WC.
Country First author, year Age group/Sample Health impacts
Other impacts
Area Sub-area Key findings
Greece Papadopoulou-Alataki, 2004
6-14 years x Cardio-metabolic health Metabolic syndrome
BMI was positively correlated with age, blood pressure (systolic as well as diastolic), TG, LDL-C, ALT, positive family history and negatively correlated with HDL-C and Apo; i.e. childhood adiposity was associated with the traditional components of metabolic syndrome in adulthood.
Greece Lydakis, 2012 12 years x Cardio-metabolic health Metabolic syndrome Obesity and adherence to the Mediterranean diet were independently related to indices of arterial stiffness.
Greece Manios, 2013 9-13 years x Cardio-metabolic health Iron levels Odds ratios for iron deficiency and iron deficiency anaemia were 2.46 and 3.13 in obese boys and 2.05 and 3.28 in obese girls relative to healthy weight children.
Greece Sakou, 2015 Unknown x Cardio-metabolic health Metabolic syndrome
Obesity was associated with insulin resistance (IR; adjusted OR=3.19). IR steadily predicted low HDL (adjusted OR=5.75), hypertriglyceridemia (adjusted OR=10.28), and systolic hypertension.
Greece Spathopoulos, 2009
6-11 years x Pulmonary/Aerobic Pulmonary/Aerobic
Lung function was significantly poorer in overweight and obese compared with healthy weight children after adjusting for gender, age and height. BMI remained an independent risk factor for reduced lung function, asthma and atopy (asthma in girls only).
Greece Koroni, 2009 10-11 years x Psychological/Emotional Psychological/Emotional Picture ranking exercise confirms high level of stigma associated with overweight and obesity in both healthy weight and overweight/obese children.
Greece Trikaliotis, 2011 3-5.5 years x Dental health Dental health
Overweight Greek pre-school children were at a significantly higher risk of dental caries (mean of 1.88 caries in overweight compared with 0.74 caries in healthy weight).
Greece Vassiloudis, 2014 10-12 years x Academic Academic
In a multiple linear regression, academic performance (as rated by teachers) was significantly associated with BMI, dietary quality, TV viewing, sleep, physical activity, parents' education, mother's ethnicity and family income.
111
Country First author, year Age group/Sample Health impacts
Other impacts
Area Sub-area Key findings
Greece Pervanidou, 2013 Mean age 11.2 years (110 obese, 31 healthy weight)
x Psychological/Emotional Psychological/Emotional
Obese children were 3.1 and 2.3 times more likely to report state and trait anxiety, respectively, and 3.6 times more likely to report depressive symptoms than healthy weight children of the same age group.
Greece Magkos, 2005 Mean age 12.1 years
x Cardio-metabolic health Metabolic syndrome
1982-2002, Crete: The prevalence of overweight and obesity has risen by 63 and 202%, respectively. The 2002 sample had 3.6% higher total cholesterol, 24.9% lower high-density lipoprotein-cholesterol (HDL-C), 25.3% higher low-density lipoprotein-cholesterol (LDL-C), 19.4% higher triacylglycerol, 36.6% higher TC/HDL-C ratio, and 60.3% higher LDL-C/HDL-C ratio compared with the 1982 sample. Results are indicative of a largely deteriorated CVD risk profile in Cretan children since 1982.
Ireland Carolan, 2014 Unknown age, 29 obese and 20 non-obese
x Cardio-metabolic health Metabolic syndrome
Relative to normal-weight children, obese children showed changes in immune cell frequency, inflammatory environment, and regulation of metabolic gene expression. These changes have been causally linked to the onset of metabolic disease in adulthood and suggest the future trajectory of obese children to the development of type 2 diabetes and premature cardiovascular disease.
Ireland Finucane, 2008a 2-18 years, 93% obese
x Cardio-metabolic health Blood pressure
51% of boys and 49% of girls had systolic BP in hypertensive range (> 95th percentile for age, sex and height). Results also showed a clear and continuous increase in systolic BP with increasing BMI, particularly in boys.
Ireland Finucane, 2008b Mean age 15.5 years, all obese
x Cardio-metabolic health Metabolic syndrome Within this obese sample, there were significant associations between the degree of obesity, insulin sensitivity and markers of liver steatosis.
Ireland O'Malley, 2012 Mean age 12.2 years, all obese
x Musculo-skeletal/Motor Musculo-skeletal/Motor
Moderate negative correlations were found between body composition and range of motion, flexibility, and strength. Genu valgum deformity was moderately positively correlated to body mass index.
Country First author, year Age group/Sample Health impacts
Other impacts
Area Sub-area Key findings
Ireland O'Malley, 2015a 3-18 years, all obese
x Musculo-skeletal/Motor Musculo-skeletal/Motor
Musculo-skeletal impairments (MSKI) were present in 89.9% of children, 51% reported pain, 45% had a radiological scan for MSKI, 69% had been referred to orthopaedics and 30% to A&E for MSKI. Difficulties with gait and function were observed in 19% and 9.6% respectively.
Ireland O'Malley, 2015b 3-18 years, all obese
x Musculo-skeletal/Motor Musculo-skeletal/Motor Balance impairment (BI) was observed in 80.2% of the group, and 87.2% of parents and 72.3% of children perceived that the child had an impaired quality of life.
Ireland Layte, 2011 9 years x Psychological/Emotional Psychological/Emotional
Self-perceptions relating to popularity and physical appearance were significantly negatively related to self-perceptions of weight. The perception of overweight was also significantly associated higher levels of emotional and behavioural problems.
Italy Turconi, 2007 Mean age 15.4 years
x Cardio-metabolic health Blood pressure BMI was significantly associated with both systolic BP and diasystolic BP.
Italy Turconi, 2006 Mean age 15.4 years
x Cardio-metabolic health Blood pressure There were significant positive correlations between BMI and blood pressure (diastolic and systolic) in males and females (r = 0.21 to 0.36)
Italy DiBonito, 2015 5-18 years, 78% obese
x Cardio-metabolic health Metabolic syndrome
Tg/HDL-C ratio discriminated better than non-HDL-C children with cardio-metabolic risk factors (CMRFs) or preclinical signs of liver steatosis, and increased carotid intima-media thickness (CIMT) and concentric left ventricular hypertrophy (CLVH). Also, higher BMI and WC were associated with significantly higher non-HDL-C and Td/HDL-C ratio, even within this sub-population of OW and OB children and adolescents.
Italy Capizzi, 2011 Mean age 10.3 years, referred to nutrition centre
x Cardio-metabolic health Metabolic syndrome
With fasting insulin, HOMA-IR, and triglycerides as the dependent variables, BMI was significantly associated with all three. Study proposes wrist circumference as an alternative anthropomorphic measure.
113
Country First author, year Age group/Sample Health impacts
Other impacts
Area Sub-area Key findings
Italy Caserta, 2010 11-13 years x Cardio-metabolic health Metabolic syndrome
The study demonstrates the association between overweight and obesity and cardio-vascular disease (CVD) risk factors. Subjects with lower levels of HDL and higher levels of triglycerides, insulin, and CRP plasma were observed more frequently among overweight and obese subjects than nonoverweight. At multivariate analysis, HDL cholesterol, insulin, and CRP were associated with overweight and obesity in girls, whereas in boys, insulin and CRP were associated with overweight and obesity, and LDL cholesterol with obesity. The association between overweight or obesity and increased CIMT was present in girls and was close to statistical significance in obese boys (p = 0.07).
Italy Valerio, 2013 5-18 years, referred to obesity treatment
x Cardio-metabolic health Metabolic syndrome
A range of cardiometabolic risk factors was examined. Results show that 32-34% of the overweight and obese sample had no risk factor, 39-40% had one risk factor, 20-24% had two risk factors, and 5-7% had three or more.
Italy Invitti, 2005 Attendees of obesity clinic 1979-2002
x Cardio-metabolic health Metabolic syndrome
Over the 24 year period studied, the degree of obesity has increased among attendees of the obesity clinic, glucose intolerance has decreased, traditional cardio-vascular risk factor profiles have improved, but non-traditional cardio-vascular risk profiles (CRP and uric acids) have worsened.
Italy Barba, 2006 6-11 years, southern Italy
x Cardio-metabolic health Blood pressure BMI and WC were independently associated with systolic BP, after adjusting for parental education and children's levels of physical activity.
Italy Brufani, 2010 3-18 years, all obese
x Cardio-metabolic health Diabetes/Glucose profile
Glucose metabolism abnormalities were present in 12.4% of this obese sample. Impaired glucose tolerance (IGT) was the most frequent alteration (11.2%), with a higher prevalence in adolescents than in children (14.8 vs. 4.1%).
Country First author, year Age group/Sample Health impacts
Other impacts
Area Sub-area Key findings
Italy Calcaterra, 2008 Mean age 11.2 years
x Cardio-metabolic health Metabolic syndrome
The prevalence of metabolic syndrome (i.e. three or more of BMI > 97th percentile, triglyceride levels > 95th percentile, high density lipoprotein (HDL) cholesterol level < 5th percentile, systolic or diastolic blood pressure > 95th percentile, and impaired glucose tolerance) was 0.0% in normal and overweight children, 12.0% in moderately obese and 31.1% in severely obese children.
Italy Invitti, 2003 6-18 years, all obese
x Cardio-metabolic health Metabolic syndrome
The prevalence of glucose intolerance was 4.5%. Insulin resistance, impaired insulin secretion, and diastolic BP were significantly and independently related to 2-h postload glucose values. The degree of obesity did not relate to insulin resistance but was independently correlated with inflammatory proteins, uric acid, and systolic BP.
Italy Bruno, 2010 0-14 years x Cardio-metabolic health Diabetes/Glucose profile
Diabetes registry, 1990-2003: The incidence rate was 12.26 per 100,000 personyears and significantly higher in boys (13.13 than in girls (11.35). Large geographic variations were present. Incidence rates increased linearly by 15, 27, 35, and 40% across four successive birth cohorts studied. Note that this trend is not empirically linked with trends in prevalence of overweight or obesity in the article.
Italy Ianuzzi, 2004 6-14 years, 100 obese and 47 healthy weight
x Cardio-metabolic health Metabolic syndrome
Obese children had significantly higher BP and plasma concentrations of tryglycerides, cholesterol, glucose, insulin, HOMA and C-reactive protein than healthy weight children. CIMT was also significantly higher in obese children.
Italy Corbo, 2008 6-7 years x Pulmonary/Aerobic Pulmonary/Aerobic
7.9% of all children reported current wheezing and 6.7% reported current asthma. Elevated BMI (comparing highest quintile to others) was significantly associated with both current wheeze (adjusted odds ratio=1.47) and current asthma (adjusted odds ratio=1.61).
115
Country First author, year Age group/Sample Health impacts
Other impacts
Area Sub-area Key findings
Italy Genovesi, 2005 6-11 years x Cardio-metabolic health Blood pressure
Four different methods were used to define OW and provided different estimates of OW prevalence (from 17.0 to 38.6%). The percentage of high BP subjects was significantly higher in OW than in normal-weight children regardless of the method used for the definition of the weight class.
Portugal Ferreira Felgueiras, 2011
12-15 years, all obese
x Psychological/Emotional Psychological/Emotional Higher BMI was associated with lower self-concept and the experience of being bullied further undermined self-concept.
Portugal Leite, 2012
Mean age 12.9 years, 50 healthy weight, 50 overweight, 50 obese
x Cardio-metabolic health Arterial thickness
CIMT was positively associated with BMI increase in adolescents, even in moderately overweight ranges, independent of age, gender, systolic blood pressure and plasma lipid concentrations.
Portugal Moreira, 2013 8-18 years, parent-child dyads
x Psychological/Emotional, Quality of life
Psychological/Emotional, Quality of life
Children were classified as healthy, with diabetes, asthma, epilepsy, and obesity. Children with obesity and epilepsy reported the lowest quality of life and highest levels of psychological problems, and parents of obese children reported the lowest quality of life, of the groups studied.
Portugal Teixera, 2001 Mean age 13.2 years
x Cardio-metabolic health Metabolic syndrome Both direct and indirect measures of adiposity were associated with serum cardiovascular risk factors in boys and girls.
Portugal Pedrosa, 2010 7-9 years x Cardio-metabolic health Metabolic syndrome
Presence of metabolic syndrome (MS), i.e. three or more of abdominal obesity, high fasting triglycerides, low HDL, high BP, and high fasting glucose was significantly associated with higher BMI.
Portugal Lopes, 2011 6-14 years x Musculo-skeletal/Motor Musculo-skeletal/Motor
Motor co-ordination was inversely associated with BMI. The strength of the association increased during childhood but decreased into early adolescence. Regardless of age, OW and OB children had significantly lower motor co-ordination than HW children.
Portugal Ribeiro, 2003 8-16 years, all at risk of obesity
x Cardio-metabolic health Blood pressure Systolic and diastolic Blood pressures were significantly and positively related to BMI.
Country First author, year Age group/Sample Health impacts
Other impacts
Area Sub-area Key findings
Romania Morea, 2013 2-19 years x Cardio-metabolic health Metabolic syndrome
Prevalence of metabolic syndrome (3 or more of 5 criteria present based on IDF criteria for adults) was higher in obese children than in overweight and healthy weight children (1.2% of healthy weight, 16.3% OW, 18.3% OB). There were no significant differences of MS prevalence between sexes or age groups
Romania Dumbrava, 2012 9-18 years, all oeverweight/obese
x Cardio-metabolic health Diabetes/Glucose profile 41.4% had prediabetes (PD) - higher in OB (50.7%) than OW (10.0%) and higher in pre-teens than adolescents (44.8% vs 34.5%).
Romania Pelin, 2012 7-18 years, all obese
x Cardio-metabolic health Metabolic syndrome
Metabolic syndrome (2009 IDF criteria) was present in 55.8% of the 120 obese children (3 or more of 5 risk factors), and between 72% and 100% had any one of these five.
Romania Serban, 2015 0-17 years x Cardio-metabolic health Diabetes/Glucose profile
Paediatric diabetes registry and medical centre records 1996-2005: A total of 3196 new cases, aged below 18 years, were found by both the sources. There were significant differences between the groups (p=0.012), the mean incidence being highest in the age group 10-14 years (9.6/100,000/year, 95% CI 9-10.1) and lowest in children aged from 0 to 4 years (4.8/100,000/year, 95% CI 4.4-5.3). Boys were slightly more frequently affected than girls (p=0.038). The age and gender adjusted incidence of type 1 diabetes mellitus increased significantly (p<0.001) from 6.2/100,000/year (95% CI 5.5-6.9) in 2002 to 9.3/100,000/year (95% CI 8.4-10.3) in 2011. The raise in incidence was noticed in all age groups except for 15-17 years. Note that this trend is not empirically linked with trends in prevalence of overweight or obesity in the article.
Romania AnaMaria, 2015 Mean age 9.1 years x Dental health Dental health Higher incidence of caries was associated with underweight, rather than overweight children.
Romania Casariu, 2011 6-12 years, 50 obese and 50 healthy weight
x Cardio-metabolic health Metabolic syndrome
Obese children and adolescents had enhanced concentrations of all markers of future cardiovascular disease, and an increased CIMT, in agreement with their degree of obesity. IMT was more strongly associated with WC than BMI.
117
Country First author, year Age group/Sample Health impacts
Other impacts
Area Sub-area Key findings
Romania Popescu, 2013 10-16 years, 30 obese and 30 healthy weight
x Cardio-metabolic health Metabolic syndrome
Intervention study examining effects of an omega-3 fatty acid diet. Before treatment, OB children had significantly higher values on a range of biomarkers including insulin resistance, glucose and blood lipids.
Romania Gherlan, 2012 Mean age 13.5 years, 38 obese, 24 healthy weight
x Cardio-metabolic health Metabolic syndrome
Biomarkers of an increased risk of adverse CV outcomes were significantly altered in obese children and adolescents compared with the healthy weight group (plasmatic levels of HDL-cholesterol, triglycerides and insulin-resistance biomarkers).
Romania Valean, 2010 10-16 years, all overweight/obese
x Cardio-metabolic health Metabolic syndrome
29% had metabolic syndrome and one or more risk factors was present in all children. Girls had a higher average number of risk factors than boys. Beside abdominal obesity, the most prevalent features of the metabolic syndrome were high blood pressure and low HDL cholesterol.
Romania Chesaru, 2013 Mean age 13.0 years, all overweight/obese
x Cardio-metabolic health Metabolic syndrome
37.4% exhibited one MS diagnosis criterion, 27.6%had two, 20.9% combined three criteria, and 8.36% had four or five of the criteria. The most common cardiometabolic risk factors were abdominal obesity (75.5%) and high blood pressure (41.1%), followed by low HDL-cholesterol (35%), increased fasting blood glucose (23.3%) and hypertriglyceridemia (17.8%).
Romania Chirita-Emandi, 2013
7-18 years x Cardio-metabolic health Blood pressure 21.1% of obese, 12.8% of overweight, and 7.1% of healthy weight children presented hypertension
Romania Brumariu, 2007 5-18 years, all obese
x Cardio-metabolic health Metabolic syndrome Metabolic syndrome was present in 52% of participants.
Romania Marginean, 2010
Not stated; all children admitted to a paediatric hospital 2004-2009
x x Cardio-metabolic health; Psychological/Emotional
Diabetes/Glucose profile; Psychological/Emotional
57% obese, with obesity more prevalent in boys. Insulin resistance was present in 46% of teenagers and 32% of children. All obese teenagers had depression, social isolation and low performance in school.
Romania Mihai, 2011
Mean age 13.1 years, all obese, inpatient metabolic/nutrition unit
x Cardio-metabolic health Diabetes/Glucose profile Patients with abnormal blood glucose profiles had higher BMI than children with normal glucose profiles.
Country First author, year Age group/Sample Health impacts
Other impacts
Area Sub-area Key findings
Slovenia Matejek, 2014 Mean age 7.3 years x Musculo-skeletal/Motor; Pulmonary/Aerobic
Musculo-skeletal/Motor; Pulmonary/Aerobic
Differences in all physical fitness tests administered (explosive power, balance, coordination, speed and endurance) between non-overweight, overweight and obese children were statistically significant, with poorest fitness in the obese group.
Slovenia Leskosek, 2007 7-18 years x Musculo-skeletal/Motor; Pulmonary/Aerobic
Musculo-skeletal/Motor; Pulmonary/Aerobic
The performance in almost all the fitness tests measured in the present study was substantially hindered by obesity – regardless of the age or sex of the children. The greatest influence of obesity was found in tests requiring movement of the whole body.
Slovenia Mocnik, 2015
14-20 years, 50 healthy, 31 hypertensive, 85 OW/OB
x Cardio-metabolic health Arterial thickness
In overweight and hypertensive children and adolescents PWV (pulse wave velocity) positively correlated with BMI and CMAP (central mean arterial pressure), indicating an association between less compliant arteries and childhood obesity and hypertension.
119
CHAPTER 4: EVIDENCE: ADULT IMPACTS OF CHILDHOOD
OVERWEIGHT AND OBESITY
4.1. Introduction McCarthy et al. (2016b) conducted a systematic review of the international literature of the effects
of childhood overweight and obesity on risk of adult overweight and obesity and risk of chronic
disease, disability, reduced quality of life and mortality in adult life. This section summarises the
findings of their review. No local materials from JANPA participants covering this topic were
retrieved, so the evidence concerning impacts of child overweight/obesity in adulthood is based on
the international review only.
McCarthy et al. (2016b) note that the extent to which childhood overweight or obesity contributes
to adult morbidities and other outcomes is difficult to establish for two main reasons. First, there is a
shortage of longitudinal data to study the effects of childhood obesity on adult co-morbidities.
Second, methodologically, it is difficult to determine whether childhood BMI status is a risk factor
independent of adult BMI status.
McCarthy et al. (2016b) conducted their search in three strands:
1. During November-December 2015 database searches in PubMed, EMBASE and CINAHL were
conducted.
2. The database search was supplemented by a search for grey literature in Google Scholar in
December 2015.
3. Based on the advice of the national steering committee45, subsequent searching was
conducted for longitudinal studies which examined the link between childhood BMI and
adult outcomes which were not reported in the 13 review articles. This was done to include
as many relevant conditions as possible.
In all, 366 articles were retrieved from the database search after removal of duplicates. Of these, 18
full texts were retrieved and 12 were deemed eligible. One additional review was retrieved from
reference checking and 15 further sources (individual studies rather than reviews) were identified
for inclusion to cover additional comorbidities/impacts as noted above.
Of the 13 reviews identified, 12 were conducted systematically, and three also included a meta-
analysis. A majority of these reviews were based mainly on longitudinal studies and measured BMI.
However, the type of effect estimates reported varied across studies (relative risks, hazard ratios and
odds ratios) and also whether the effect estimates in adulthood were based on 1SD or 1-unit
increase in BMI, BMI z-score quartiles or BMI categories of overweight and obese. A description of
the details (including main conclusions) of each of these 13 publications is shown in the Appendix 2
(Table A11).
45
Membership of the steering committee for the safefood/JANPA WP4 studies is shown in the Contributors section at the beginning of this document.
Fifteen additional studies, all primary analyses, were also included in this review. Publication dates
ranged from 1993-2015. All studies except one were based on large, nationally representative
longitudinal surveys. Main findings of each of these studies are shown in the Appendix 2 (Table A12).
McCarthy et al. (2016b) extracted the best estimates of effects for each of the conditions and
outcomes covered in their review (i.e. giving preference to pooled effect estimates on the basis of
meta-analyses, if available; if not, the most recent effect estimates from studies with larger sample
sizes). These are shown in the Appendix 2 (Table A13).
A summary of each of the areas considered in these sources, as discussed by McCarthy et al.
(2016b), follows.
4.2. Child or adolescent overweight and obesity and adult morbidities
4.2.1. Type 2 diabetes
The evidence strongly supports a link between childhood overweight/obesity and risk of type 2
diabetes in adulthood in both sexes.
In the systematic review and meta-analysis by Llewellyn et al. (2016), statistically significant positive
relationships between childhood obesity and adult diabetes were found among children aged 6
years and under (pooled OR per SD of BMI 1.23), aged 7-11 years (pooled OR per SD of BMI 1.78)
and aged 12 and over (pooled OR per SD of BMI 1.70). The systematic review by Juonala et al. (2011)
reported a significant pooled relative risk of 5.4 (adjusted for age, sex, height, length of follow-up
and cohort) comparing individuals who were consistently overweight or obese from childhood to
adulthood on the basis of IOTF cut-points to those with normal BMI. In 9 of 10 studies identified in
the review by Park et al. (2012) a significantly increased risk of type 2 diabetes was found (with odds
ratios per one SD increase in BMI ranging from 1.22 to 2.04 in these nine studies).
4.2.2. Coronary heart disease (CHD) and ischaemic heart disease (IHD)
There is some evidence for a link between childhood overweight/obesity and CHD in adulthood,
though the evidence is not as consistent as that for type 2 diabetes, and the results suggest that
higher BMI in later childhood rather than early childhood poses a greater risk.
Llewellyn et al. (2016) reported that childhood BMI was significantly associated with CHD among
children aged 12 years and over (pooled OR per SD of BMI 1.30) and among children aged 7-11 years
(pooled OR per SD of BMI 1.14), but there was no statistically significant association between
childhood obesity and CHD among children aged 6 years and under (pooled OR per SD BMI = 0.97).
Of 15 studies identified by Park et al. (2012), 10 reported a significant relationship. Statistically
significant hazard ratios summarised in Park et al. ranged from 1.53 for the association between CHD
mortality and high BMI to 5.43 for the association between incident CHD and high BMI. A review by
Owen et al. (2009) found no association between BMI in children aged 2-6 years and later CHD risk
(on the basis of three studies), but reported a significant positive association between BMI at ages 7-
18 years and later CHD risk (on the basis of seven studies; pooled RR = 1.09).
Regarding IHD, there is evidence for a weak positive association between childhood BMI and risk of
IHD in adulthood, but this evidence is based on one study discussed in Lawlor et al. (2006). The
study, however, was of high quality and drew on data from three UK cohort samples. Findings
indicated a pooled hazard ratio for IHD per 1 SD of BMI of 1.09, adjusting for family social class.
121
4.2.3. Stroke
There is not strong evidence for an association between childhood BMI and stroke in adulthood.
The meta-analysis by Llewellyn et al. (2016) reported pooled odds rations for studies that included
children aged 6 years and under, and aged 7-11 years, that were not statistically significant.
However, pooled odds ratios for studies including children aged 12-18 years were weak, though
statistically significant (pooled OR 1.06). The review by Park et al. (2012) included eight studies
examining the association between childhood/adolescent BMI and adult stroke. Of these, four
reported statistically significant associations between childhood BMI and adult stroke incidence or
mortality, while three did not find any statistically significant association.
4.2.4. Cancers
McCarthy et al. (2016b) note that the association between childhood BMI and cancer in adulthood
varies, depending on the type of cancer; some studies examined incidence, while others examined
mortality; furthermore, some studies differentiated between high and very high BMI, while others
did not.
The review by Park et al. (2012) included one study which indicated that being in the highest BMI
quartile during childhood was associated with a 40% increase in risk of all cancers (adjusting for a
range of variables including socio-economic status), but two further studies in this review reported
no association between childhood BMI and all-cancer mortality. The meta-analysis by Llewellyn et al.
(2016) reported a modest, though significant, odds ratio per SD of BMI at age 7-11 years (1.14) and
incidence of all types of cancer in adulthood (based on findings from the Body Orr cohort). This
meta-analysis also reported small, though statistically significant associations between childhood
BMI and incident hepatocellular carcinoma, liver cancer, colon cancer, ovarian cancer, renal cell
carcinoma (male only sample), and urothelial cancer. On the other hand, the review reported no
association between childhood BMI and incident breast cancer. Similarly, the review by Park et al.
(2012) included four studies concerning breast cancer: three found no or mixed associations, while
one found that childhood BMI was a significant risk factor. Park et al.’s review suggests that there
may be gender differences in the links between childhood BMI and risk of certain cancers: colorectal
cancer mortality risk was higher among males than females in one study they reviewed; in another
study, though, incident colorectal cancer risk was significant and similar in both males and females
(with relative risks of around 2.0 in both genders). In another study reviewed by Park et al., very high
BMI (> 85th percentile) was significantly associated with incident renal cell carcinoma in males, but
not in females.
McCarthy et al. (2016b) included four papers that drew on the Copenhagen School Records Register
(Aarestrup et al., 2014, 2016; Kitahara et al., 2014a, 2014b) to examine the associations between
childhood (measured) BMI and various cancers during adulthood. Note that adult BMI was not
possible to include in these studies.
Aaresrup et al. (2014) used this data source to examine whether BMI in boys (aged 7 to 13) was
associated with increased prostate cancer risk at age 40. They found that childhood BMI was
associated with a marginally significant increased risk of prostate cancer among the youngest age-
group studied (7-8 years), and no significant associations among the other groups. Hazard ratios
became attenuated and non-significant after adjusting for children’s height. They also reported that
changes in boys’ BMI over time (at ages 7-13) was not associated with risk of prostate cancer.
Aarestrup et al. (2016) investigated associations between childhood BMI in girls and risk of
endometrial cancer. Women were followed up until a diagnosis of endometrial cancer or
hysterectomy (or death, emigration, loss to follow-up, or end of the study on December 31, 2012).
There was a non-linear association between childhood BMI and endometrial cancers, oestrogen-
dependent cancers, and the sub-type of endometrioid adenocarcinoma. At all childhood ages (from
7-13 years), girls with a BMI z-score higher than 0 had a greater risk of all endometrial cancers,
oestrogen-dependent cancers and endometrioid adenocarcinoma compared with girls with a BMI z-
score of 0. Adjusting for childhood height resulted in an attenuation of the association, but it
remained statistically significant.
Kitahara et al.’s (2014a) study on the association between childhood BMI and adult thyroid cancer,
also using the Copenhagen School Records Register data, reported that BMI at each age was
positively associated with thyroid cancer risk. However, the hazard ratios were larger for papillary
than follicular thyroid cancer, and the strongest associations were observed for papillary thyroid
cancer in men. Finally, Kitahara et al. (2014b) also explored child BMI and adult glioma, and found no
evidence of a significant association in either sex.
4.2.5. Metabolic syndrome
A review by Lloyd et al. (2012) included three studies examining the association between childhood
overweight/obesity and risk of metabolic syndrome in adulthood46. Two found significant, positive
associations: for example, one of these two reported that for every 1 SD change in objectively
measured childhood BMI among children aged 8-17 years, the odds of having four criterion risk
variables for metabolic syndrome in adulthood was 2.03, adjusting for age at baseline sex and race;
the third study found no association.
4.2.6. Components of metabolic syndrome
4.2.6.1. Total cholesterol, LDL and HDL cholesterol, and triglycerides
The evidence for an association between childhood BMI and total cholesterol in adulthood is mixed.
The review by Lloyd et al. (2012) included four studies that examined association between childhood
BMI and total cholesterol in adulthood. One reported a weak significant correlation between
childhood BMI (ages 5-17) and total adult cholesterol (r = .10). A second study in this review
reported significant positive correlations between increase in BMI at age 8-18 and total cholesterol
in adulthood which was stronger in males (r = .20-.45) than females (r = .10-.26). In contrast, the two
remaining studies reported no, or weak negative associations, between child BMI and adult
cholesterol. Two of the four studies adjusted for adult BMI: one resulted in an inversion of the
positive child BMI-adult cholesterol association, while the other only resulted in slight changes in this
relationship.
46
The International Diabetes Federation (IDF) clinical and diagnostic definition of the metabolic syndrome is based on the presence of central obesity, along with any two of: raised triglycerides > 1.7mmol/l), reduced HDL cholesterol (> 1.03 mmol/l in females and > 1.29 mmol/l in males), raised blood pressure (systolic > 130, or diasystolic > 85mm Hg), and raised fasting plasma glucose (> 5.6 mmol/l). Metabolic syndrome is present in individuals at high risk of type 2 diabetes and cardiovascular disease. Additional metabolic measurements are related to this syndrome: abnormal body fat distribution such as liver fat content, atherogenic dyslipidaemia (such as small LDL particles), dysglycaemia, insulin resistance, vascular dysregulation, proinflammatory state, prothrombotic state, and hormonal factors (Alberti et al., 2006).
123
Similarly, the evidence for an association between childhood weight status and LDL/HDL cholesterol
levels in adulthood is mixed.
In one of the studies reviewed (a primary analysis of four longitudinal cohorts; Juonala et al., 2011),
participants were classified as being consistently overweight or not from childhood to adulthood
(IOTF cut-points); those consistently overweight had a significantly higher risk of elevated LDL
cholesterol (RR =1.8) and for lowered HDL cholesterol (RR = 2.1) (adjusting for age, sex, height,
length of follow-up, and cohort membership).
In studies reviewed by Lloyd et al. (2012), two examined associations between HDL and LDL
cholesterol levels in adulthood and childhood BMI. One of the two reported significant weak
correlations between BMI at ages 5-17 and adult LDL (r = .11) and HDL ( r = -.14) cholesterol, which
inversed after adjustments for adult BMI status. A second study failed to find significant associations
between childhood weight status and adult HDL or LDL cholesterol levels, with or without
adjustments for adult BMI.
There is some evidence to support an association between BMI in childhood and triglyceride levels in
adulthood. However, this depends on whether adjustments are made for adult weight status in
analyses.
Juonala et al.’s (2011) analyses of four longitudinal cohorts reported a significantly higher risk of
elevated triglyceride levels in adults who had been consistently overweight since childhood (RR =
3.0). Lloyd et al.’s (2012) review included two studies which reported significant positive associations
between child weight status and adult triglyceride levels. However, one of these studies also
reported the association after adjustment for adult BMI, which resulted in an inversion of the
relationship between child BMI and adult triglyceride levels from weak positive to weak negative.
4.2.6.2. Insulin resistance
There is evidence to support an association between BMI in childhood and markers of insulin
resistance in adulthood. However, again, the relationship depends on whether adjustments are
made for adult weight status in analyses.
Lloyd et al.’s (2012) review included six studies that examined relationships between childhood BMI
and measures of insulin concentrations or insulin resistance in adulthood. Three of the six reported a
moderate, positive association. For example, one of the studies reported a correlation of .32
between BMI measured at ages 5-17 years and acute insulin response at a mean age of 25 years
(adjusting for sex, adult percentage body fat, and age in childhood and at follow-up). However, the
remaining three studies reported no association, or weak negative associations, after adjustments
were made for adult weight status.
4.2.6.3. Carotid artery atherosclerosis
The evidence supports a positive association between BMI in childhood and carotid artery
atherosclerosis in adults, but again, the nature of these associations tends to be attenuated or
inversed if adjustments are made for adult BMI status in analyses.
Juonala et al.’s (2011) analyses of four longitudinal cohorts reported a significantly higher risk of
carotid intima-media thickness in adults who had been consistently overweight since childhood (RR
= 1.7). The review by Lloyd et al. (2010) included six studies that examined this association and in
five of the six, there was a significant positive relationship. However, after adjustments for adult BMI
were made in analyses, only one of the five studies reported a significant association.
4.2.6.4. Hypertension
There is strong and consistent evidence for an association between childhood BMI and hypertension
in adulthood, though adjustments for adult BMI status attenuate this association.
On the basis of two cohort studies, Llewellyn et al. (2016) reported a significant pooled odds ratio
per SD of BMI of 1.29. Similarly, all five studies included in the review by Park et al. (2012) reported a
significant association between child BMI and risk of hypertension in adulthood, and all three articles
reviewed by Reilly and Kelly (2011) that examined hypertension found significant associations. The
studies reviewed by Park et al. suggest that the risk increases with children’s age. For example, one
study reported odds ratios of 1.35 (7 years old), 1.65 (11 years old), and 1.96 (16 years old). Another
study reviewed by Park et al. provided odds ratios for adolescents aged 16.5-19 years of 1.75 and
3.75 for overweight and obese, respectively.
A review by Lloyd et al. (2010) included four studies that examined the associations between
childhood BMI and adult hypertension with adjustments for adult BMI. Two of these studies found
that the positive association reversed with the adjustment for adult BMI, while the other two studies
reported that the positive association between child BMI and hypertension remained after adjusting
for adult BMI.
4.2.6.5. Non-alcoholic adult fatty liver disease (NAFLD)
NAFLD was not included as an adult outcome in the 13 reviews identified by McCarthy et al. (2016b);
however, they included a primary study from Denmark (Zimmerman et al., 2015) which examined
the association between NAFLD in adults and childhood BMI. BMI was assessed on multiple
occasions at ages 7-13 years, and follow-up began at age 18 years. Results indicated that an increase
in BMI from age 7-13 years rather than absolute BMI value was associated with adult NAFLD risk.
4.2.7. Asthma
McCarthy et al. (2016b) note that evidence for an association between childhood overweight or
obesity and asthma in adulthood was available in just three studies in the sources that they
reviewed. The evidence is therefore limited.
Park et al.’s (2012) review included two studies that examined this association. One of the two
reported a significant association in females but not males (between childhood BMI measured at 7
years and self-reported asthma at >21 years); the other found no association between BMI
measured at age 10 years and self-reported asthma at age 26 years. The third study is reviewed by
Reilly and Kelly (2011): the study found that obesity at age 14 years (BMI > 95th percentile) was
associated with a higher probability of doctor-diagnosed asthma relative to normal BMI (BMI < 85th
percentile) (OR = 2.09). A range of variables were adjusted for in this analysis (mother’s age and pre-
pregnancy BMI, smoking during pregnancy, parity, social class, parental allergies, month of birth,
sex, gestational age, smoking status, physical activity, and professional training). However, BMI was
based on self-reports rather than objective measurements.
125
4.2.8. Musculo-skeletal problems
Lower back pain was not included as an adult outcome in the 13 reviews identified by McCarthy et
al. (2016b). They identified a primary study from the UK (Power et al., 2001) which examined the
association between the presence of lower back pain in adults and childhood BMI. The analyses
were based on the 1958 British Birth Cohort. In multivariate analyses, there was no association
between lower back pain in adulthood (ages 32-33 years) and child BMI (age 7 years).
McCarthy et al. (2016b) identified two primary studies that looked at osteoarthritis, while the review
paper by Park et al. (2012) included one study examining arthritis. All three reported significant
associations, but results varied by gender and specific symptoms considered.
Antony et al. (2015) analysed results from the Australian Schools Health and Fitness Survey (1985-
2010). Participants were aged 7-15 years at baseline and 31-41 years at follow-up (sample size at
follow-up, at 449, is small, but is reasonably representative). Knee pain was assessed by the WOMAC
(Western Ontario and McMaster Universities Osteoarthritis) index. No significant associations
between childhood overweight measures and total WOMAC knee pain, stiffness and dysfunction
scores in adulthood. However, childhood overweight measures were associated with knee pain,
stiffness and dysfunction among men. Associations remained unchanged after adjustment for adult
overweight. Long-term overweight status was also associated with knee pain, with subjects who
were overweight in both childhood and adulthood having the greatest prevalence and risk of knee
pain.
MacFarlane et al. (2011) analysed data from the 1958 British Birth Cohort. They reported that BMI
was significantly associated with knee pain, but that the strength of this association increased with
age. Persons with a BMI of >30 kg/m2 at age 23, 33 or 45 years experienced approximately a
doubling in the risk of knee pain at 45 years. There was a significant association with knee pain at the
age of 45 years with high BMI from as early as age 11 years, but the association was stronger at the
age of 16 years.
Park et al. (2012) reported on one study conducted in the US, where it was found that the risk of
arthritis in older adults (in their 70s) was significantly associated with overweight in adolescence
(without any adjustments, RR = 2.0).
4.2.9. Reproductive health
Just two studies included by McCarthy et al. (2016b) looked at reproductive health, and both of
these concerned females only.
One was a primary analysis by Lake et al. (1997) which used data from the 1958 British Birth Cohort.
They found that other than menstrual problems, childhood BMI had little impact on the
reproductive health of women; however, adult BMI was associated with measures of reproductive
health. Obesity at 23 years and obesity at 7 years both independently increased the risk of menstrual
problems by age 33 after adjusting for other confounding factors. Obesity at 23 years increased the
risk of hypertension in pregnancy, after adjusting for confounders. Obesity at 7 years also increased
the risk of hypertension in pregnancy (unadjusted OR 2.14) but the risk did not persist after
adjustment for BMI at 23 years and other confounders. Obese women at 23 years were less likely to
conceive within 12 months of unprotected intercourse after adjustment for confounders.
Reilly and Kelly’s (2011) review included one study that examined variations in prevalence of
polycystic ovarian syndrome (PCOS) by child weight status. The study reported a significant positive
association between obesity at age 14 years and PCOS at age 31 after adjusting for social class (OR =
1.61). However, family history of PCOS was not adjusted for in this study, and PCOS is associated
with insulin resistance (Schwartz & Chadha, 2008).
4.3. Adult overweight and obesity There is strong and consistent evidence for a positive association between overweight/obesity in
childhood and adulthood, and the association is even stronger between overweight/obesity in
adolescence and adulthood.
In the meta-analysis by Simmonds et al. (2016), the pooled results of 15 high-quality studies
indicated that children who were obese at ages 7-11 years were 4.86 times more likely to be obese
as adults than non-obese children. Adolescents aged 12-18 years who were obese were 5.45 times
more likely to be obese in adulthood. Simmonds et al. also estimated that around 55% of obese
children remain obese as adults and 80% of adolescents will remain obese in adulthood. However,
they also estimated that 70% of obese adults were not obese in childhood or adolescence.
Singh et al.’s (2008) systematic review on this topic included 18 studies (or 25 articles). All found a
positive association between obesity in childhood or adolescence and adulthood obesity. Ten of the
studies demonstrated that the persistence of obesity increased with age. When only five high-quality
studies were looked at in this review, risk of overweight in adulthood was estimated to be at least
twice as high among children who were overweight compared with children who were never
overweight.
4.4. Adult mortality There is limited evidence to confirm an association between childhood weight status and all-cause
mortality in adulthood.
Five of the studies included in the review by Park et al. (2012) indicate that the risk of all-cause
mortality is increased by 40-60% in people with high BMI at ages 2-19 years. Adami et al.’s (2008)
review on this topic located eight studies, and although some of these showed an association
between childhood BMI and mortality rates, they concluded that there was not sufficient
longitudinal evidence to support a clear association between childhood/adolescent
overweight/obesity and adult mortality. McCarthy et al. (2016b) further noted that the majority of
studies that examined the link between childhood weight status and mortality did not adjust for
socio-economic status, which may be a confounding factor.
Twig et al. (2016), however, recently reported on the association between adolescent (measured)
BMI and deaths from cardiovascular causes including CHD, stroke and sudden death, and their
analysis did adjust for socio-economic status. They used data from almost 2.3 million adolescents
(males and females) enrolled for Israeli military service at age 17 with a 40-year follow-up. US-CDC
percentiles were used to split the cohort into seven groups based on BMI in adolescence. After
adjusting for age, birth year, sex, education level, socio-economic status and country of origin, Twig
et al. (2016) reported significant positive associations between adolescent obesity (≥95th percentile)
for death from CHD (HR 4.9; 95% CI 3.9, 6.1), death from stroke (HR 2.6; 95% CI 1.7 to 4.1), sudden
death (HR 2.1; 95% CI 1.5, 2.9) and death from total cardiovascular causes (HR 3.5; 95% CI 2.9, 4.1)
127
compared to the reference group (5th-24th BMI percentile). They estimated projected population-
attributable fractions of 28% for death from total cardiovascular causes and 36% for death from
coronary heart disease. The study provides quite strong evidence that overweight and obesity in
adolescence are associated with increased cardiovascular mortality in adulthood.
4.5. Other adult outcomes
4.5.1. Sick leave
One primary study in McCarthy et al.’s (2016b) review examined associations between adolescent
BMI and adult sick leave using longitudinal data, in a male-only sample in Sweden (mean age at
baseline = 18.7 years). Neovius et al. (2012a) found that overweight and obesity were associated
with increased risk for sick-leave compared to healthy weight, especially for sick-leave episodes of
longer duration. Results were adjusted for smoking, socio-economic index and muscular strength.
Overweight was associated with a hazard ratio of 1.20 and obesity a hazard ratio of 1.35 for sick
leave episodes ranging from 8 to 30 days. Overweight was associated with a hazard ratio of 1.19 and
obesity a hazard ratio of 1.34 for episodes lasting more than 30 days.
4.5.2. Disability pension
Only two studies identified in McCarthy et al.’s (2016b) review examined longitudinal associations
between child or adolescent weight status and disability pension in adulthood. These are described
in Reilly and Kelly (2011). Both are based on male-only Swedish conscription samples, with measures
of BMI taken in late adolescence and indicate a higher likelihood of disability pension associated
with obese than with overweight.
One study reported hazard ratios of 1.36, 1.87 and 3.04 for males who were overweight and
moderately and severely obese, respectively, at age 18, and later disability pension (with
adjustments for muscular strength, age, region, socio-economic status and year). This study also
reported hazard ratios for specific kinds of disability (circulatory, musculoskeletal,
psychiatric/nervous system, injuries and tumours). In all cases, overweight and obesity were
associated with significantly higher hazard ratios, and higher hazard ratios for obese compared with
overweight. The median number of work years lost at age 65 were 0.2 years higher among
overweight compared with healthy weight, 1.5 years higher among obese compared with healthy
weight, and 3.6 years higher among severely obese compared with healthy weight. The other study
reported generally consistent findings.
4.5.3. Lifetime productivity losses
Two primary studies in McCarthy et al.’s (2016b) review examined associations between child or
adolescent BMI and adult productivity losses using longitudinal data, one in a male-only sample and
the other in a mixed sample. In a group of male conscripts in Sweden (mean age at baseline = 18.7
years), Neovius et al. (2012b) found that obesity was associated with almost twice as high
productivity losses to society than healthy weight over a lifetime. Using the human capital approach,
the lifetime productivity losses were estimated at 55.6 × €1000 for under/healthy weight, 72.6 × €
1000 for overweight and 95.4 × € 1000 for obesity. Results were adjusted for socio-economic status,
smoking and muscular strength.
Viner et al. (2005) analysed data from the 1970 British Birth Cohort and found that obesity in
childhood only was not associated with adult social class, income, years of schooling, educational
attainment, relationships, or psychological morbidity in either sex after adjustment for confounding
factors. Persistent obesity was not associated with any of the studied adult outcomes in men,
though it was associated with a higher risk of never having been gainfully employed among women
(OR = 1.9).
4.5.4. Educational attainment
Three primary studies reviewed by McCarthy et al. (2016b) examined longitudinal associations
between earlier BMI and adult educational attainment. Amis et al. (2014) used data from the
National Longitudinal Study of Adolescent Health which followed children from grades 7-12 (ages 12-
18 years) for 13 years. They found that, after adjusting for demographics, family environment, prior
academic achievement, behavioural variables, community environment, and general and mental
health, adults who had been obese (<95th percentile) were 8.9% less likely to graduate from college
This effect was stronger for females (-12.2%) than males (-5.0%). Gortmaker et al. (1993) reported,
on the basis of the US National Longitudinal Survey of Labor Market Experience, no significant
association between BMI at ages 17-18 years in men, but an estimated 0.3 years less of formal
education in women who had been overweight at age 17-18. Results were adjusted for various
baseline characteristics including household income, parental education, presence of a chronic
health condition, and test scores. Sargent and Blancflower (1994) examined data from the National
Child Development Study (birth cohort in England, Scotland and Wales). Adults who had been obese
at age 16 had completed fewer months of schooling after the age of 16: 3.1 fewer months in males,
and 4.3 fewer months in females. Adjusted results were not reported by Sargent and Blanchflower.
4.5.5. Income
Four primary longitudinal studies inform the evidence on associations between income/poverty in
adults and earlier weight status. Evidence is mixed but suggests that a stronger link for females than
males.
In the same study described in Section 4.5.4, after adjusting for a range of background
characteristics and potential confounders, Amis et al. (2014) estimated that adults who had been
obese at ages 12-18 years earned 7.5% less than their non-obese counterparts, and that this loss of
earnings was greater for females (-8.7%) than males (-6.0%).
Sargent and Blanchflower (1994) found that women (in England, Scotland and Wales) who had been
in the top 10% of the BMI distribution at age 16 earned 7.4% less, and women in the top 1% earner
11.4% less, than their normal-weight counterparts at age 23, after adjusting for parental social class
and test scores at baseline. Moreover, this association persisted with adjustments for BMI status at
age 23. In contrast, there was no significant relationship between BMI status at age 16 and earnings
at age 23 in males.
In analyses of the 1970 British Birth Cohort, when participants were aged 10 to 30 years, Viner et al.
(2005) found that females who were obese in childhood (at 10 years) and persistently obese into
adulthood had a significantly lower mean annual net income compared with those that were not
obese in either period. There was no statistically significant relationship between childhood obesity
and adult income in males.
Gortmaker et al. (1993) reported that, independent of baseline socio-economic status and test
score, US women who had been overweight between the ages of 16 and 24 had significantly lower
129
household income and higher rates of household poverty than the women who had not been
overweight, independent of base-line socioeconomic status and aptitude-test scores. These
associations were not statistically significant among men.
4.5.6. Psychological health
McCarthy et al. (2016b) highlight the scarcity of high-quality longitudinal studies that have examined
the associations between child or adolescent weight status and psychological health in adulthood.
One review that they identified (Sikorski et al., 2015) examined associations between adolescent
BMI and indicators of psychological wellbeing, though it did not meet the original search criteria of
following children or adolescents into adulthood. However, of the 25 studies identified in Sikorski et
al.’s review, six were longitudinal, with follow-up periods of 1-4 years. Broadly speaking, the findings
of these four studies support weak associations between overweight or obesity and self-esteem and
social supports/loneliness, at least in the short term (from adolescence to young adulthood).
Luppino et al. (2010) conducted a systematic review and meta-analysis of the longitudinal
associations between adult overweight/obesity and depression. This was not included in McCarthy
et al.’s (2016b) review since the study falls outside its scope – it examined adults only. However it is
discussed here because it provides useful evidence in the nature of the relationship between
depression and weight status. Luppino et al. (2010) note that existing cross-sectional evidence does
not provide information on the mechanisms that link depression and obesity. Their analysis included
15 studies, all of which expressed weight status in terms of BMI and classified overweight as 25-
29.99 and obesity as 30 or higher.
Results indicated that obesity and overweight at baseline both predicted depression at follow-up
(OR [obesity] = 1.55; OR [overweight] = 1.27) and the association was stronger among older adults
and not significant among younger adults (aged <20 years). Depression predicted obesity (OR = 1.58)
but not overweight. The results confirm a reciprocal association which may be reinforced and
strengthened over time. Luppino et al. (2010) suggest that the link between overweight/obesity and
depression could be mediated by inflammation, insulin resistance, and/or psychological distress
arising from societal norms and values regarding weight. In turn, the link between depression and
obesity may be accounted for by neuroendocrine disturbances, adoption of unhealthy lifestyle
behaviours, and/or some antidepressant medications.
Tables
Table T4.1. Summary of main findings of 13 review papers included in McCarthy et al.’s (2016b) systematic review on the impacts of childhood
overweight/obesity in adulthood
First author, year
Type Studies reviewed Main findings and conclusions
Llewellyn, 2016 Systematic review and
meta-analysis 37 studies
Examined the associations between childhood BMI and type 2 diabetes, CHD, cancers, and stroke. Childhood BMI was associated with type 2 diabetes, CHD, and some cancers but not, or inconsistently, with stroke and breast cancer. Authors conclude that childhood BMI is not a good predictor of the incidence of adult morbidities because the majority of adult obesity-related morbidity occurs in adults who were of healthy weight in childhood. Targeting obesity reduction solely at children may not substantially reduce the overall burden of obesity-related disease in adulthood.
Park, 2012 Systematic
review 39 studies
Childhood BMI was associated with type 2 diabetes, hypertension, CHD, and all-cause mortality. Few studies examined associations independent of adult BMI and these tended to show that effect sizes were attenuated after adjustment for adult BMI in standard regression analyses. This approach of adjusting, however, is subject to over-adjustment and consequent problems with interpretation.
Juonala, 2011 Primary analysis 4 longitudinal cohort studies
The study classified participants using both child and adult BMI, comparing four groups depending on normal/overweight status as child/adult. Overweight/obese children who were obese as adults had increased risks of type 2 diabetes, hypertension, dyslipidemia, and carotid-artery atherosclerosis. The risks of these outcomes among overweight or obese children who became nonobese by adulthood were similar to those among persons who were never obese.
Owen, 2009 Systematic
review 15 studies (17 estimates)
BMI is positively related to CHD risk from childhood onwards; the associations in young adults are consistent with those observed in middle age. There was considerable variation between studies, however.
Simmonds, 2016
Systematic review and
meta-analysis 37 studies (22 cohorts)
Examined associations between childhood BMI and type 2 diabetes, hypertension and CHD (associations tended to be significant) and stroke and breast cancer (inconsistent or not significant). Childhood BMI is not a good predictor of adult obesity or adult disease; a majority of obese adults were not obese as children; most obesity-related adult morbidity occurs in adults who had a healthy childhood weight. However, obesity (as measured using BMI) was found to persist from childhood to adulthood.
131
First author, year
Type Studies reviewed Main findings and conclusions
Reilly, 2009 Systematic
review 28 studies (8 mortality, 11 cardiometabolic
morbidity, 9 other morbidity)
Childhood BMI was associated with: risk of premature mortality (4/5 studies included in the review), increased risk of cardiometabilic morbidities (11/11 studies), increased risk of later disability pension, asthma, and polycystic ovary syndrome (4/4 studies). Findings for cancer morbidity were inconsistent (5 studies). Overall, there is consistent evidence to demonstrate that overweight/obesity in childhood/adolescence has adverse consequences on premature mortality and physical morbidity in adulthood
Lloyd, 2010 Systematic
review 16 studies
Little evidence was found to suggest that childhood obesity is an independent risk factor for CVD. Results suggest that relationships observed are dependent on the tracking of BMI from childhood to adulthood. Evidence suggests, contrary to expectation, that risk of raised blood pressure is highest in those who are at the lower end of the BMI scale in childhood and overweight in adulthood.
Lloyd, 2012 Systematic
review 11 studies
Little evidence was found to support the view that childhood obesity is an independent risk factor for adult blood lipid status, insulin levels, metabolic syndrome or type 2 diabetes. The majority of studies failed to adjust for adult BMI.
Nadeau, 2011 Narrative
review N/A
Evidence supports the concept that precursors of adult CVD begin in childhood, and that pediatric obesity has an important influence on overall CVD risk. Lifestyle patterns also begin early and impact CVD risk. Whether childhood obesity causes adult CVD directly, or does so by persisting as adult obesity, or both, is less clear.
Simmonds, 2016
Systematic review and
meta-analysis 15 studies
Around 55% of obese children go on to be obese in adolescence, around 80% of obese adolescents will still be obese in adulthood and around 70% will be obese over age 30. However, 70% of obese adults were not obese in childhood or adolescence.
Singh, 2008 Systematic
review 25 studies
All studies consistently reported an increased risk of overweight and obese youth becoming overweight adults, suggesting that the likelihood of persistence of overweight into adulthood is moderate. However, predictive values varied considerably across studies included.
Adami, 2008 Systematic
review 8 studies
There is a scarcity of studies determining the risk of mortality in adulthood in relation to overweight/obesity in childhood /adolescence. There is a need for studies that control for variables such as physical activity, smoking, maintenance of stable weight, adult BMI, and onset of puberty. There is no clear evidence on the association between childhood/adolescent overweight /obesity and adult mortality.
Sikorski, 2015 Systematic
review 46 studies
Adults with obesity had higher psychological risk factors; they may be a mediator between weight discrimination and pathopsychological outcomes.
Table T4.2. Summary of main findings of 15 individual studies included in McCarthy et al.’s (2016b) systematic review on the impacts of childhood
overweight/obesity in adulthood
First author, year
Type Sample Main findings and conclusions
Zimmermann, 2015
Primary analysis 244,464 boys and girls, born between 1930
and 1989 age 7-13 years, Copenhagen, followed up at age 18 years
Increase in BMI from age 7-13 years rather than absolute BMI value was associated with adult non-alcoholic fatty liver disease (NAFLD).
Power, 2001 Primary analysis
1958 British Birth Cohort, individuals who experienced onset of low back pain at 32 to
33 years of age (n= 571) and individuals who were pain free (n = 5210).
In multivariate analyses, there was no association between low back pain in adulthood and child BMI.
Antony, 2015 Primary analysis n=449, ages 31-41, BMI data gathered 25
years prior to this, Australia
Height, weight and knee injury were recorded, knee pain assessed by WOMAC (Western Ontario and McMaster Universities Osteoarthritis) index. No significant associations between childhood overweight measures and total WOMAC knee pain, stiffness and dysfunction scores in adulthood. However, childhood overweight measures were associated with knee pain, stiffness and dysfunction among men. Associations remained unchanged after adjustment for the corresponding adult overweight measure. Change in overweight status from childhood to adulthood was also associated with knee pain, with subjects who were overweight in both childhood and adulthood having the greatest prevalence and risk of knee pain.
MacFarlane, 2011
Primary analysis 1958 British Birth Cohort Study, followed
up at 45 years of age (n = 8579); 1636 with knee pain.
BMI was associated with knee pain: persons with a BMI of >30 kg/m2 at 23, 33 or 45 years experienced
approximately a doubling in the risk of knee pain at 45 years. There was a significant association with knee pain at the age of 45 years with high BMI from as early as age 11 years, but the association was stronger at the age of 16 years.
Lake, 1997 Primary analysis
1958 British Birth Cohort Study, 5799 females with at height, weight and
reproductive data at 7, 11, 16, 23 and 33 years of age.
Other than menstrual problems, childhood body mass index had little impact on the reproductive health of women. Obesity at 23 years and obesity at 7 years both independently increased the risk of menstrual problems by age 33 after adjusting for other confounding factors. Obesity at 23 years increased the risk of hypertension in pregnancy, after adjusting for confounders. Obesity at 7 years also increased the risk of hypertension in pregnancy (unadjusted OR 2.14) but the risk did not persist after adjustment for BMI at 23 years and other confounders. Obese women at 23 years were less likely to conceive within 12 months of unprotected intercourse after adjustment for confounders.
Viner, 2005 Primary analysis 1970 British Birth Cohort Study, 8490
participants with BMI data at 10 and 30 years
Obesity in childhood only was not associated with adult social class, income, years of schooling, educational attainment, relationships, or psychological morbidity in either sex after adjustment for confounding factors. Persistent obesity was not associated with any adverse adult outcomes in men, though it was associated among women with a higher risk of never having been gainfully employed.
133
First author, year
Type Sample Main findings and conclusions
Gortmaker, 1993
Primary analysis Nationally representative sample, 10,039, aged 16 to 24 years old in 1981, follow up
in 1981 for 65-79% of original cohort.
Women who had been overweight had completed 0.3 fewer years of school, were 20% less likely to be married, had lower household income and higher rates of household poverty than the women who had not been overweight, independent of base-line socioeconomic status and aptitude-test scores. Men who had been overweight were 11% less likely to be married. There was no evidence of an effect of overweight on self-esteem in either gender.
Sargent, 1994 Primary analysis
National Child Development Study, 1958 (England, Scotland, Wales), 12,537
respondents at age 23 years
Regression analyses indicated no relationship between obesity at any age and earnings at age 23 years in males. There was a statistically significant inverse relation between obesity and earnings in females, independent of parental social class and ability test scores in childhood. Females who had been in the top 10% of the BMI at age 16 years earned 7.4% less and females in the top 1% earned 11.4% less.
Neovius, 2012 Primary analysis Military conscripts (all male), Sweden, 1969-1970, n=43,989, mean age 18.7, followed up between 1986 and 2005
Overweight and obesity were associated with increased risk for sick-leave compared to healthy weight, especially for sick-leave episodes of longer duration. Results were adjusted for smoking, socio-economic index and muscular strength. Overweight was associated with 20% and obesity with >30% risk elevation for episodes ranging from 8 to 30 days as well as for episodes >30 days.
Amis, 2014 Primary analysis
Participants in the US National Longitudinal Study of Adolescent Health (n=11,308), measured BMI 1994-1995 (ages 12-18
years), followed up in 1996, 2001-2002, 2007-2008
Adults who had been obese (> 95th
percentile) at age 12-18 years were 8.9% to obtain a college degree and earned 7.5% less annually 13 years later, than adults who had not been obese. In contrast they were no more or less likely to graduate high school on time or attend graduate school than the non-obese group. Effects were stronger for females than males for both college degree (-12.2% vs -5.0%) and earnings (-8.7% vs -6.0%). There were some differences by ethnic/racial group. Results were adjusted for demographics, family environment, prior academic achievement, behavioural variables, community environment and general and mental health.
Aarestrup, 2014 Primary analysis
Copenhagen School Health Records Register (born 1930–1969) (n=125,208
males). BMI was measured objectively at 7-13 years and participants were followed
from the age of 40 years.
Childhood BMI was positively associated with prostate cancer in adulthood at age 7 (unadjusted HR 1.06; 95% CI 1.01, 1.10 per BMI z-score), and age 13 (unadjusted HR 1.05; 95% CI 1.01, 1.10 per BMI z-score). After adjusting for childhood height, most associations became insignificant.
Aarestrup 2016 Primary analysis
Copenhagen School Health Records Register (born 1930–1969) (n=155,505
females). BMI was measured objectively at 7-13 years and participants were followed
from age 18 to end of 2012
There was a non-linear association between childhood BMI and all endometrial cancers, oestrogen-dependent cancers and the sub-type of endometrioid adenocarcinoma. At all childhood ages (from 7-13 years), girls with a BMI z-score higher than 0 had a greater risk of all endometrial cancers, oestrogen-dependent cancers and endometrioid adenocarcinoma compared with girls with a BMI z-score of 0. In contrast, BMI was not associated with non-oestrogen-dependent cancers except in the oldest childhood age-groups.
Kitahara 2014a Primary analysis
Copenhagen School Health Records Register (born 1930–1969) (n=321,085).
BMI was measured objectively at 7-13 years and participants were followed to end of
2010.
BMI at each age was positively associated with thyroid cancer risk. However, the hazard ratios were larger for papillary than follicular thyroid cancer, and the strongest associations were observed for papillary thyroid cancer in men.
First author, year
Type Sample Main findings and conclusions
Kitahara 2014b Primary analysis
Copenhagen School Health Records Register (born 1930–1969) (n=326,466).
BMI was measured objectively at 7-13 years and participants were followed to end of
2010.
The authors found no significant association between childhood BMI and adult glioma.
Twig 2016 Primary analysis (n=2,298,130) Israeli adolescents aged
approx. 17 years with measured BMI with follow up 40 years later.
After adjustment for sex, age, birth year, socio-economic status and country of birth, characteristics, there was a positive association between adolescent obesity (US-CDC ≥95th percentile) for death from CHD (HR 4.9; 95% CI 3.9, 6.1), death from stroke (HR 2.6; 95% CI 1.7 to 4.1), sudden death (HR 2.1; 95% CI 1.5, 2.9) and death from total cardiovascular causes (HR 3.5; 95% CI 2.9, 4.1) compared to the reference group in the 5th-24th percentiles. Overall the study provides strong evidence for a link between childhood obesity and adult cardiovascular mortality risk.
135
Table T4.3. Effect estimates of the association between childhood BMI and adult morbidities and outcomes (from McCarthy et al., 2016b)
Morbidity or Outcome
Cohort Country Dates Publication Childhood age at baseline
Age at adult follow-up
Gender N at baseline
N at follow-up
Effect estimates Study Quality
<12 years ≥12 years
Diabetes Meta-analysis of:
Aberdeen
British birth cohort 1958
Bogalusa/Young Finns
Israel SPEC
National Growth and Health study/ Princeton follow-up study Norway
Scotland
England, Wales, Scotland
Finland and USA Israel
USA
Norway
1950–2000
1958–2000
1984–2007
1976-NR
1973–2003
1963–2005
Review: Llewellyn (2015)
Lawlor (2006)
Hypponen (2003)
Magnussen (2010)
Tirosh (2011)
Morrison (2010)
Bjorge (2008)
Mean 4.9
7, 11, 16
12–19
Median 7.2–14.5 0–11
14–19
46–50
41
19–26
NR NR
Mean 52
NR
52% male
51.1% male
100% male
52% male
51% male
12,150
16,751
20,745
161,063
NR
227,000
5,793
10,683
10,439
117,415
12,439
226,682
Age 6 and under:
Pooled OR per SD of BMI 1.23 (1.10-1.37)
Age 7-11:
Pooled OR per SD of BMI 1.78 (1.51-2.10)
Age 12-18:
Pooled OR per SD of BMI 1.70 (1.30-2.22)
High (CASP 10/10)
Coronary Heart Disease
Meta-analysis of:
Aberdeen
Boyd Orr
Copenhagen Health Records Register (born 1930–76)
Helsinki 1924
Helsinki 1934 – males
Helsinki 1934 – females
Israel SPEC
Scotland
England, Scotland
Denmark
Finland
Finland
Finland
Israel
1950–2000
1937–1995
1930–2011
1924–1997
1934–2003
1934–2003
1976-NR
Review: Llewellyn (2015)
Lawlor (2005)
Gunnell (1998)
Baker (2007)
NR
Eriksson (2001)
Forsen (2004)
Tirosh (2011)
Mean 4.9
2–14
7–13
NR (approx. 6–16)
0–12
0–11
Median 7.2–14.5
48–54
Up to 73
≥25
NR (approx. 31–73)
27–63
27–64
NR
NR
49% male
51% male
NR
100% male
0% male
100% male
12,150
NR
276,835
NR
5,502
5,486
161,063
11,106
2,399
NR
NR
3,544
3,003
117,415
Age 6 and under:
Pooled OR per SD of BMI 0.97 (0.85-1.10)
Age 7-11:
Pooled OR per SD of BMI 1.14 (1.08-1.21)
Age 12-18:
Pooled OR per SD of BMI 1.30 (1.16-1.47)
High (CASP 10/10)
Morbidity or Outcome
Cohort Country Dates Publication Childhood age at baseline
Age at adult follow-up
Gender N at baseline
N at follow-up
Effect estimates Study Quality
<12 years ≥12 years
Stroke
(meta-analysis from Llewellyn et al., 2015)
Meta-analysis of:
Aberdeen
Boyd Orr
Helsinki 1934
Copenhagen Health Records Register (born 1930-76)
Scotland
England, Scotland
Finland
Denmark
1950–2000
1937–1995
1934–2003
1930–2011
Review: Llewellyn (2015)
Lawlor (2005)
Gunnell (1998)
Osmond (2007)
Baker (2007)
Mean 4.9
2–14
2–14
7–13
48–54
Up to 73
26–33
NR
NR
49% male
58% male
51% male
12,150
NR
8,181
276,835
11,106
2,399
1,492
NR
Age 6 and under: Pooled OR per SD of BMI 0.94 (0.75-1.19)
Age 7-11: Pooled OR per SD of BMI 1.02 (0.94-1.10)
Age 12-18: Pooled OR per SD of BMI 1.06 (1.04-1.09)
High (CASP 10/10)
Hyper-tension
(meta-analysis from Llewellyn et al., 2015)
Meta-analysis of:
British birth cohort 1958
Beijing Child and Adolescent Metabolic Syndrome study
National Longitudinal Study of Adolescent Health (National Growth and Health study (NGHS)/ Princeton follow-up study (PFS)
England, Wales, Scotland
China
USA
1958–2000
2004–2010
1995–2001
Review: Llewellyn (2015)
Li (2007)
Cheng (2011)
Merten (2010)
7–16
6–16
PFS: Mean 12.4,
NGHS: NR
45
Mean 16 (SD 1.8)
PFS: 32–44,
NGHS: 19
52% male
54% male
PFS: 47% male, NGHS: 0% male
13,294
2,189
1,889 (PFS: 822, NGHS: 1,067)
9,285
1,184
Up to 1,058 (NGHS)
Age 7-11: Pooled OR per SD of BMI 1.67 (0.89-3.13)
Age 12-18: Pooled OR per SD of BMI 1.29 (1.19-1.40)
High (CASP 10/10)
Breast Cancer
(meta-analysis from Llewellyn et al., 2015)
Meta-analysis of:
Medical Research Council National Survey of Health and Development
Helsinki 1924 Copenhagen Health Records Register (born 1930-76)
UK
Finland
Denmark
1946–1999
1924–1997
1930–2011
Review: Llewellyn (2015)
De Stavola (2004)
Hilakivi-Clarke (2001)
Ahlgren (2004)
2–15
7 and 15
7–14
47–53
Min 38 (76% > 50 y)
NR
0% male
0% male
0% male
2,547
NR
161,063
2,187
3,447
117,415
Age 6 and under: Pooled OR per SD of BMI 0.88 (0.67-1.16)
Age 7-11: Pooled OR per SD of BMI 0.90 (0.77-1.05)
Age 12-18: Pooled OR per SD of BMI 0.92 (0.82-1.03)
High (CASP 10/10)
137
Morbidity or Outcome
Cohort Country Dates Publication Childhood age at
baseline
Age at adult follow-
up
Gender N at
baseline
N at follow-up Effect estimates Study
Quality
<12 years ≥12 years
All Cancer Boyd Orr cohort England, Scotland 1937–1995 Jeffreys (2004) 2-14 Up to 66 49% male 2,997 2,374 OR per SD of BMI
1.14 (1.00-1.29)
High (CASP
10/10)
Renal Cell Carcinoma
Israeli army cohort Israel 1967–2006 Leiba (2013) 16-19 Mean=44 100% male NR 1,110,835 OR per SD of BMI
1.19 (1.04-1.37)
High (CASP
9/10)
Rectal Cancer Israeli army cohort Israel 1967–2006 Levi (2011) 16-19 19-57 100% male NR 1,109,864 OR per SD of BMI
0.96 (0.88-1.10)
High (CASP
9/10)
Pancreatic Cancer
Israeli army cohort Israel 1967–2006 Levi (2012) 16-19 29-56 100% male NR 720,927 OR per SD of BMI
1.17 (0.96-1.52)
High (CASP
9/10)
Ovarian Cancer Norway cohort Norway 1963–2005 Engeland (2003) 14-19 Mean=41 0% male NR 111,883 OR per SD of BMI
1.22 (1.01-1.49)
High (CASP
10/10)
Urothelial Cancer
Israeli army cohort Israel 1967–2006 Leiba (2012) 16-19 Mean=35 100% male NR 1,110,835 OR per SD of BMI 1.21 (1.06-1.38)
High (CASP 9/10)
Colon Cancer Norway cohort Israel 1967–2006 Bjorge (2008) 16-19 19-57 51% male NR 1,109,864 OR per SD of BMI 1.21 (1.07-1.38)
High (CASP 9/10)
Lung or related cancer
Norway cohort Norway 1963–2005 Bjorge (2008) 14-19 Mean=52 51% male 227,000 226,682 OR per SD of BMI 1.10 (0.84-1.44)
High (CASP 10/10)
Heptocellular Carcinoma (boys only)
Copenhagen Health Records Register (born 1930-76)
Denmark 1930–2011 Berentzen (2013) 7-13 31-80 51% male 372, 636 285,884 (for males and females
OR per SD of BMI 1.31 (1.12-1.53)
OR per SD of BMI 1.36 (1.17-1.60)
High (CASP 10/10)
Heptocellular Carcinoma (girls only)
Copenhagen Health Records Register (born 1930-76)
Denmark 1930–2011 Berentzen (2013) 7-13 31-80 51% male 372, 636 285,884 (for males and females
OR per SD of BMI 1.05 (0.78-1.40)
OR per SD of BMI 1.23 (0.93-1.65)
High (CASP 10/10)
Morbidity or Outcome
Cohort Country Dates Publication Childhood age at baseline
Age at adult follow-up Gender N at
baseline
N at follow-up
Effect estimates Study Quality
<12 years ≥12 years
Liver Cancer (boys only)
Copenhagen Health Records Register (born 1930-76)
Denmark 1930–2011 Berentzen (2013) 7-13 31-80 51% male 372, 636 285,884 (for males and females
OR per SD of BMI 1.27 (1.11-1.45)
OR per SD of BMI 1.30 (1.14-1.48)
High (CASP 10/10)
Liver Cancer (girls only)
Copenhagen Health Records Register (born 1930-76)
Denmark 1930–2011 Berentzen (2013) 7-13 31-80 51% male 372, 636 285,884 (for males and females
OR per SD of BMI 1.20 (0.97-1.49)
OR per SD of BMI 1.32 (1.07-1.64)
High (CASP 10/10)
Cancer Mortality
Norway cohort Norway 1963–2005 Bjorge (2008) 14-19 Mean=52 51% male 227,000 226,682 OR per SD of BMI 1.10 (0.95-1.23)
High (CASP 10/10)
Colon Cancer Death
Norway cohort Norway 1963–2005 Bjorge (2008) 14-19 Mean=52 51% male 227,000 226,682 OR per SD of BMI 1.48 (1.05-2.11)
High (CASP 10/10)
Type 2 Diabetes
(meta-analysis from Juonala et al., 2011)
Bogalusa (n=632)
Muscatine (n=722)
Childhood Determinants of Adult Health (n=2331)
Young Finns Study (n=2640)
USA
USA
Australia, Finland
and USA
Finland
1973-1996
1971-1981
1985-2006
1980-2002
Berenson (1998), Davis
(2001), Magnussen
(2008) , and Raitakari
(2003)
Range: 3-19 Range 23-46
(Length of follow-up:
Ranges from 19.9±0.6 to
26.0±2.3 years)
41% male
48% male
49% male
46% male
NR 6328 Among age range 3-19:
Who were overweight or obese in
childhood but non-obese in adulthood
(n=274): Pooled RR 1.3 (95% CI 0.4-4.1)
Who were overweight or obese in
childhood and obese in adulthood
(n=500): Pooled RR 5.4 (95% CI 3.4-8.5)
Who were normal BMI in childhood but
obese as adults (n=812):
Pooled RR 4.5 (95% CI 2.9-6.8)
(Reference category: those with normal
BMI in childhood and non-obese in
adulthood).
High (CASP
10/10)
139
Morbidity or Outcome
Cohort Country Dates Publication Childhood age at baseline
Age at adult follow-up Gender N at
baseline
N at follow-up
Effect estimates Study Quality
Hypertension
(meta-analysis from Juonala et al., 2011)
Bogalusa (n=632)
Muscatine (n=722)
Childhood Determinants of Adult Health (n=2331)
Young Finns Study (n=2640)
USA
USA
Australia, Finland
and USA
Finland
1973-1996
1971-1981
1985-2006
1980-2002
Berenson (1998), Davis
(2001), Magnussen
(2008) , and Raitakari
(2003)
Range: 3-19 Range 23-46
(Length of follow-up:
Ranges from 19.9±0.6 to
26.0±2.3 years)
41% male
48% male
49% male
46% male
NR 6328 Among age range 3-19:
Who were overweight or obese in
childhood but non-obese in adulthood
(n=274): Pooled RR 0.9 (95% CI 0.6-1.4)
Who were overweight or obese in
childhood and obese in adulthood
(n=500): Pooled RR 2.7 (95% CI 2.2-3.3)
Who were normal BMI in childhood but
obese as adults (n=812):
Pooled RR 2.1 (95% CI 1.7-2.4)
Adjusted for: age, sex, height, length of
follow-up and cohort.
(Reference category: those with normal
BMI in childhood and non-obese in
adulthood).
High
(CASP
10/10)
High-risk LDL cholesterol (meta-analysis from Juonala et al., 2011)
Bogalusa (n=632)
Muscatine (n=722)
Childhood Determinants of Adult Health (n=2331)
Young Finns Study (n=2640)
USA
USA
Australia, Finland
and USA
Finland
1973-1996
1971-1981
1985-2006
1980-2002
Berenson (1998), Davis
(2001), Magnussen
(2008) , and Raitakari
(2003)
Range: 3-19 Range 23-46
(Length of follow-up:
Ranges from 19.9±0.6 to
26.0±2.3 years)
41% male
48% male
49% male
46% male
NR 6328 Among age range 3-19:
Who were overweight or obese in
childhood but non-obese in adulthood
(n=274): Pooled RR 1.1 (95% CI 0.7-1.6)
Who were overweight or obese in
childhood and obese in adulthood
(n=500): Pooled RR 1.8 (95% CI 1.4-2.3)
Who were normal BMI in childhood but
obese as adults (n=812):
Pooled RR 1.5 (95% CI 1.2-1.9)
Adjusted for: age, sex, height, length of
follow-up and cohort.
(Reference category: those with normal
BMI in childhood and non-obese in
adulthood).
High
(CASP
10/10)
Morbidity or Outcome
Cohort Country Dates Publication Childhood age at baseline
Age at adult follow-up
Gender N at
baseline
N at follow-up
Effect estimates Study Quality
High-risk HDL cholesterol (meta-analysis from Juonala et al., 2011)
Bogalusa (n=632)
Muscatine (n=722)
Childhood Determinants of Adult Health (n=2331)
Young Finns Study (n=2640)
USA
USA
Australia, Finland and
USA
Finland
1973-1996
1971-1981
1985-2006
1980-2002
Berenson (1998), Davis
(2001), Magnussen (2008) ,
and Raitakari (2003)
Range: 3-19 Range 23-46
(Length of follow-
up: Ranges from
19.9±0.6 to
26.0±2.3 years)
41% male
48% male
49% male
46% male
NR 6328 Among age range 3-19:
Who were overweight or obese in
childhood but non-obese in adulthood
(n=274): Pooled RR 1.0 (95% CI 0.7-
1.3)
Who were overweight or obese in
childhood and obese in adulthood
(n=500): Pooled RR 2.1 (95% CI 1.8-
2.5)
Who were normal BMI in childhood
but obese as adults (n=812):
Pooled RR 2.2 (95% CI 1.9-2.6)
(Reference category: those with
normal BMI in childhood and non-
obese in adulthood).
High
(CASP
10/10)
High-risk triglycerides cholesterol (meta-analysis from Juonala et al., 2011)
Bogalusa (n=632)
Muscatine (n=722)
Childhood Determinants of Adult Health (n=2331)
Young Finns Study (n=2640)
USA
USA
Australia, Finland and
USA
Finland
1973-1996
1971-1981
1985-2006
1980-2002
Berenson (1998), Davis
(2001), Magnussen (2008) ,
and Raitakari (2003)
Range: 3-19 Range 23-46
(Length of follow-
up: Ranges from
19.9±0.6 to
26.0±2.3 years)
41% male
48% male
49% male
46% male
NR 6328 Among age range 3-19:
Who were overweight or obese in
childhood but non-obese in adulthood
(n=274): Pooled RR 0.7 (95% CI 0.4-
1.2)
Who were overweight or obese in
childhood and obese in adulthood
(n=500): Pooled RR 3.0 (95% CI 2.4-
3.8)
Who were normal BMI in childhood
but obese as adults (n=812):
Pooled RR 3.2 (95% CI 2.7-3.8)
Adjusted for: age, sex, height, length
of follow-up and cohort.
(Reference category: those with
normal BMI in childhood and non-
obese in adulthood).
High
(CASP
10/10)
141
Morbidity or Outcome
Cohort Country Dates Publication Childhood age at baseline
Age at adult follow-up
Gender N at
baseline
N at follow-up
Effect estimates Study Quality
High-risk carotid-artery intima-media thickness (meta-analysis from Juonala et al., 2011)
Bogalusa (n=632)
Muscatine (n=722)
Childhood Determinants of Adult Health (n=2331)
Young Finns Study (n=2640)
USA
USA
Australia, Finland and
USA
Finland
1973-1996
1971-1981
1985-2006
1980-2002
Berenson (1998), Davis
(2001), Magnussen (2008)
and Raitakari (2003)
Range: 3-19 Range 23-46
(Length of follow-
up: Ranges from
19.9±0.6 to
26.0±2.3 years)
41% male
48% male
49% male
46% male
NR 6328 Among age range 3-19:
Who were overweight or obese in
childhood but non-obese in adulthood
(n=274): Pooled RR 0.9 (95% CI 0.6-
1.3)
Who were overweight or obese in
childhood and obese in adulthood
(n=500): Pooled RR 1.7 (95% CI 1.4-
2.2)
Who were normal BMI in childhood
but obese as adults (n=812):
Pooled RR 1.5 (95% CI 1.3-1.8)
Adjusted for: age, sex, height, length
of follow-up and cohort.
(Reference category: those with
normal BMI in childhood and non-
obese in adulthood).
High (CASP
10/10)
Polycystic Ovarian Syndrome
Longitudinal, population-based study of a cohort of women born in 1966 in northern Finland.
Finland 1966-1983 Laitinen (2003) 14 years 31 years 0% male 1836
(note: BMI
data was
self-
reported
at baseline
(age 14)
with some
missing
data)
2007
(note: BMI
data was
measured
at age 31)
Overweight at age 14 years:
RR 1.12 (95% CI 0.87–1.43)
Obese at age 14 years:
RR 1.61 (95% CI 1.24-2.08)
(Reference category: those with
healthy weight at age 14y)
Note on adult risk:
Those who were healthy weight at
14y and overweight or obese at 31y:
RR 1.27 (95% CI 1.07–1.52). Those
who were overweight or obese both
at 14 and 31y: RR 1.44 (1.16–1.78).
(Reference category: Healthy weight
at 31 y)
Morbidity or Outcome
Cohort Country Dates Publication Childhood age at
baseline
Age at adult follow-up
Gender N at baseline
N at follow-up
Effect estimates Study Quality
<12 years ≥12 years
Asthma 1970 British Cohort Study UK 1970-1996 Shaheen (1999) 10 26 45% male 8960 6420 Age 10 years: BMI at age 10y was not associated with asthma at age 26y.
High (CASP 9/10)
Non-alcoholic fatty liver disease (NAFLD)
Copenhagen School Health Records Register (CSHRR)
Denmark 1930-2010 Zimmerman (2015) 7-13 18-80 50% male 372,636 244,464 Age 7 years:
In both sexes, childhood BMI z-score was not consistently associated with adult NAFLD.
However, change in BMI z-score between 7 and 13 years of age was positively associated with NAFLD in both sexes:
Males: HR 1.15 (95% CI 1.05 to 1.26) per 1-unit gain in BMI z-score between ages 7 and 13 years
Females: HR 1.12 (95% CI 1.02 to 1.23) per 1-unit gain in BMI z-score after adjusting for BMI z-score between ages 7 and 13 years
High (CASP 10/10)
143
Morbidity or Outcome
Cohort Country Dates Publication Childhood age at
baseline
Age at adult follow-up
Gender N at baseline N at follow-up
Effect estimates Study Quality
<12 years ≥12 years
Low back pain
1958 British birth cohort UK 1958-1991 Power (2001) 7 32-33 49% male 11407 5781 Age 7 years: No association between low back pain at age 32-33y and BMI at 70
th-85
th percentile
(unadjusted OR 1.05; 95% CI 0.79, 1.42). No association between low back pain and BMI >85
th
percentile (unadjusted OR 1.24; 95% CI 0.93, 1.66)
Reference group: BMI 30th–70
th percentile
High (CASP 10/10)
Osteoarthritis:
(Knee pain, Knee stiffness, Physical dysfunction)
1958 British Birth Cohort
UK 1958-1991 MacFarlane (2011) 1, 11, 16 45 49% male 18,558
8579
Age 11 years:
Knee pain at age 45 was not significantly associated with overweight (BMI 25-30 kg/m2) (RR 1.27; 95% CI 0.91, 1.77) compared to those with BMI<20 kg/m2 at age 11 after adjustment for confounding factors (socioeconomic status, smoking status, knee injury (after 33 years), marital status, gender and psychological distress).
Age 16 years:
Knee pain at age 45 was significantly associated with overweight (BMI 25-30 kg/m2) (RR 1.31; 95% CI 1.07, 1.61) and obesity (BMI >30 kg/m2) (RR 1.59; 95% CI 1.05, 2.40) compared to those with BMI<20 kg/m2 at age 16 after adjustment for confounding factors
High (CASP 10/10)
Morbidity or Outcome
Cohort Country Dates Publication Childhood age at
baseline
Age at adult follow-up
Gender N at baseline
N at follow-up
Effect estimates Study Quality
Osteoarthritis
1958 British Birth Cohort
Australian Schools Health and Fitness Survey
Australia 1985-2010 Antony (2015) 7-15 31-41 52% male 794 449 Among age range 7-15 (mean 11y): Knee Pain: Overweight (BMI >25 kg/m2) at age 7-15y was associated with knee pain at age 31-41y (RR 1.72; 95% CI 1.11, 2.69) in multivariable model for males but not significant for females. Walking knee pain: Childhood overweight (BMI >25 kg/m2) was associated with walking knee pain in adulthood among both males and females (RR 2.64; 95% CI 1.29, 5.40). Knee Stiffness: Childhood BMI (per kg/m2) at age 7-15y was associated with knee stiffness at age 31-41 (RR 1.11; 95% CI 1.05, 1.19) for males but not significant for females. Physical dysfunction: Childhood overweight (BMI >25 kg/m2) at age 7-15y was associated with knee stiffness at age 31-41 (RR 1.61; 95% CI 1.07, 2.43) for males but not significant for females. Multivariate model included age, sex, height (for weight), duration of follow-up, child and adult knee injury, smoking status and socioeconomic position).
High (CASP 10/10)
Gout Third Harvard Growth Study
USA 1922-1988 Must (1992) 13-18 early 70s 45% male 508 342 (83 lost to follow-up and 83 did not respond)
Age 13-18 years: Males: Gout was significant associated with overweight (≥75th percentile) in adolescence (unadjusted RR 3.1; 95% CI 1.1, 9.3) compared to those who were not overweight (25th and 50th percentile) in adolescence. This result was attenuated when adjusted for adult BMI (adjusted RR 2.2; 95% CI 0.7, 6.9). Overall (among both genders), there was no statistically significant association between adolescent overweight and gout (unadjusted RR 2.7; 95% CI 0.9, 7.7) compared to those who were not overweight in adolescence.
High (CASP 9/10)
145
Morbidity or Outcome
Cohort Country Dates Publication Childhood age at
baseline
Age at adult follow-up
Gender N at baseline N at follow-up
Effect estimates Study Quality
Subfertility (time taken to conceive from cessation of contraception)
1958 British birth cohort UK 1958-1991 Lake (1997) 7, 11, 16 33 0% male NR 5799 Age 7 years: No association between childhood overweight or obese at 7 y old and fertility (achieving a pregnancy within 12 months) at age 33 y.
(Reference group: Normal BMI)
High (CASP 10/10)
Adult obesity
(meta-analysis from Simmonds et al. 2015)
Meta-analysis of:
Bogalusa Heart Study
National Heart, Lung, and Blood Institute Growth and Health Study (NGHS)
Australian Schools Health and Fitness Survey (ASHFS)
1958 British birth cohort
USA
USA
Australia
UK
1973-1994
1986-2001
1985-2006
1958-1991
Freedman (2005)
Thompson (2007)
Venn (2007)
Power (2007)
10
11
11
11
27
22
29
33
45% male
0% male
48% male
52% male
2392
2379
6839
18,558
2057
2054
5170
11,212
Age 7-11 years:
Children who
were obese at the
age of 7-11y were
more likely to be
obese as adults
than non-obese
children (RR 4.86;
95% CI 4.29, 5.51)
High (CASP
10/10)
Morbidity or Outcome
Cohort Country Dates Publication Childhood age at
baseline
Age at adult follow-up
Gender N at baseline N at follow-up
Effect estimates Study Quality
Adult obesity
(meta-analysis from Simmonds et al. 2015)
Meta-analysis of:
Young Finns Study
Bogalusa Heart Study
National Heart, Lung, and Blood Institute Growth and Health Study (NGHS)
1958 British birth cohort
Bogalusa Heart Study
National Longitudinal Study of Youth 1979
Finland
USA
USA
UK
USA
USA
1980- 2001
1973-1994
1986-2001
1958-1991
1973-1994
1979-2002
Juonala (2006)
Gordon-Larsen (2004)
Thompson (2007)
Power (2007)
Freedman (2005)
Wang (2008)
15
16
16
16
16
16
31
23
22
33
27
37
NR
49% male
0% male
52% male
45% male
52% male
3596
2392
2379
18,558
2392
2513
2373
NR
2054
11,212
2057
1309
Age 12 years and over:
Children who were obese
at the age of 12y and
over were more likely to
be obese as adults than
non-obese children (RR
5.45; 95% CI 4.34-6.85)
High (CASP
10/10)
All-cause mortality
Norway cohort Norway 1963-2001 Engeland (2003) 14–19 45-50
(followed for
an average of
31.5 years)
51% males 227,003 226,958 Age 14-19 years:
Very high BMI (≥85th
centile) at 14-19y was
associated with an
increased risk of all-cause
mortality among males
(RR 1.4; 95% CI 1.3, 1.6)
and females (RR 1.4; 95%
CI 1.2, 1.5)
(Reference category:
Those in the medium
category (25-74th
centile).
High (CASP
10/10)
147
Morbidity or Outcome
Cohort Country Dates Publication Childhood age at baseline
Age at adult follow-up
Gender N at baseline
N at follow-up
Effect estimates Study
Quality
Sick Leave Data from the Military Service Conscription Registry (MSCR)
Sweden 1986-2005 Neovius (2012) 18 NR (250 493
person-years
of follow-up)
100% male 43,989 NR Age 18 years:
Overweight was associated sick leave ranging
from 8 to 30 days (HR 1.20; 95% CI 1.15–
1.24) and long-term sick leave >30 days (HR
1.19; 95% CI 1.15–1.23).
Obesity was associated sick leave ranging
from 8 to 30 days (HR 1.35; 95% CI 1.24–
1.47) and long-term sick leave >30 days (HR
1.34; 95% CI 1.24–1.47).
Adjusting for smoking, socio-economic index
and muscular strength.
High
(CASP
9/10)
Lifetime productivity losses
1970 British birth cohort England, Scotland,
and Wales
1970- 2000 Viner (2005) 10 30 48% male 12,160 8490 Age 10 years:
Females: Persistent obesity from childhood
to adulthood was significantly associated
with a higher risk of never having been
gainfully employed (AOR 1.9, 95% CI 1.1 to
3.3) in the multivariable model.
Males: Persistent obesity from childhood to
adulthood was not significantly associated
with a higher risk of never having been
gainfully employed (AOR 1.4, 95% CI 0.9 to
2.3) in the multivariable model.
Adjusting for: maternal education, social
class in childhood and adulthood, maternal
and paternal BMI, and height at 10 and 30
years.
High
(CASP
8/10)
Morbidity or Outcome
Cohort Country Dates Publication Childhood age at baseline
Age at adult follow-up
Gender N at baseline
N at follow-up
Effect estimates Study
Quality
Education Status (Years of schooling)
National Longitudinal Survey of Labour Market Experience, Youth Cohort (NLSY)
USA 1979-1988 Gortmaker (1993) 17-18 24-25 51% male 10,039 7931 Age 17-18 years:
Females: Overweight at age 17-18 years was
significantly associated with fewer years
completed at school (0.3 year less; 95% CI
0.1 to 0.6, p=0.009) compared to non-
overweight.
Males: Overweight at age 17-18 years was
not significantly associated with fewer years
completed at school (0.2 year less; 95% CI
0.5 to 0.0, p=0.08)
Adjusting for: base-line characteristics,
including household income, the mother's
and father's educational level, the score on
the AFQT, the presence of a chronic physical
health condition, height, self-esteem, age,
and race or ethnic group.
High
(CASP
8/10)
*Effect estimates reported as OR per SD of BMI are applicable to both childhood overweight and obesity.
Source: McCarthy et al
149
STUDY PROTOCOLS
Chapters 5 - 11
CHAPTER 5: OUTLINE THE STUDY PROTOCOLS
5.1 Development
The development of the Study Protocols proceeded in the following steps:
Systematic review of the International literature (Hamiltion et al (2016) in preparation)
conducted by the Irish National Team with significant additional funding from safefood
Collection of details of addtional (international and local) sources in the “Data Sources
Survey”
These were summarised into a draft Study Protocols
Feedback from JANPA WP4 countries, consultations with expert groups and ISAC
Testing in the Irish study
Finalising of the the Study Protocols
5.2 Chapters relevant to the Study Protocols
The Study Protocols has several parts:
Chapter 6 gives an overview of existing studies that estimate the lifetime costs of childhood overweight and obesity.
Chapter 7 gives a summary of the JANPA WP4 methodology
Chapter 8 describes the model inputs, outputs metrics used in to describe the lifetime impacts and costs of childhood obesity and overweight
Chapter 9 describes the steps of the modelling
Chapter 10 describes how the model outputs are used to calculate model metrics
Chapter 11 gives an overview of activities to explore the validity of findings and generalisability of the JANPA WP4 methodology.
151
CHAPTER 6: EXISTING STUDIES OF LIFETIME COST OF CHILDHOOD
OVERWEIGHT AND OBESITY
6.1. Approaches used to estimate costs
Types of studies
Broadly speaking, health forecasting models can be grouped into three categories (Astolfi et al.,
2012):
Microsimulation models simulate entire populations and are flexible, allowing a range of
scenarios to be tested. They allow examination of forecasted results by different
characteristics included in the model, such as by diseases, age-groups, or treatments. Life-
course or disease events can be represented in the lives of the simulated individuals and in
dynamic models, certain characteristics and behaviours evolve over the life course, with
attributions based on risks or probabilities. Individual life trajectories are usually simulated
until death. To estimate the potential impacts of an intervention or other change, the model
is run twice, once as the ‘base case’ and then again with a ‘variant’ scenario; comparisons
between the base model and variants provide information on the potential impacts of
interventions. This family of models require large amounts of data to construct a sample
that adequately represents the population of interest. Data are usually gathered from a
variety of sources and, depending on the particular research question(s), data on disease
progression after initial diagnosis, and on degrees of response that individuals may have
changes in an external variable (elasticities), may be required.
Component-based models forecast health expenditure by component, such as by financing
agents or providers of care. An important sub-set of this family is the cohort model, where
individuals are grouped into cells according to several key attributes. The unit of analysis is
therefore driven by the combination of attributes under examination. Typically age is the
principal criterion. Further refinements are obtained by sub-dividing the cohorts according
to other common attributes. Each cell in the model is associated with an average cost of
health goods and services. Actuarial projections allow predicting the likely evolution of the
population and therefore the future number of individuals included in each cell. Future
health expenditure is determined by multiplying the average costs by the projected number
of individuals included in each cell. More advanced cohort-based models also account for
factors influencing epidemiologic trends such as exposure to risks factors (such as smoking).
The popularity of component-based models is likely due to the fact that they are relatively
simple and inexpensive to implement.
Macro models focus on total health expenditure. They include analysis of time-series and
cross-sections of aggregate indicators. This group of models includes computable general
equilibrium models (CGE) which link health expenditure growth to its impact on the overall
economy. Macro models are best suited to short-term projections in the presence of clear
trends.
The three classes of models are illustrated in Figure 5.1. Combinations of these three classes of
models are also used. The models reviewed here fall into the first two classes described above.
All classes of models rely on sets of assumptions, and it is important that these assumptions are
transparent to users of the results. Astolfi et al. (2012) therefore recommend that results from these
models should be accompanied by measures of uncertainty associated with projections. Sensitivity
analyses which systematically vary and compare values of the model input parameters are also
commonly used to test the robustness of findings. Regarding simulation models, there should be a
clear statement of model structure (i.e. assumptions, equations and algorithms), data used, and
results of validation exercises (Levy et al., 2011).
Figure 6.1. Families of health forecasting models
Source: Astolfi et al., 2012, Figure 3
Costs
It is common in health forecasting models to distinguish between two types of costs: direct and
indirect. Direct costs cover those related to health-care services and provision and include in-patient
and out-patient hospital care and treatment, primary (General Practitioner) care, and
drug/pharmaceutical costs. Indirect costs to broader societal costs that relate to losses in
productivity, including work absenteeism and presenteeism, lower income, disability and sick leave,
and premature mortality.
The costs associated with overweight and obesity are estimated using one of two approaches (Perry
et al., 2012). Top-down approaches usually draw on country-specific data on the population
prevalence of overweight and obesity, along with data on the prevalence of various conditions that
have been shown to be associated with raised BMI. Frequently, relative risks (RRs) for these
conditions on the basis of BMI category are derived from observational studies. These are combined
153
with prevalence of overweight and obesity to derive population attributable fractions (PAFs), that is,
the proportion of cases of a given condition or disease that is attributable to overweight or obesity.
The bottom-up approach uses individual-level data, most frequently collected through cross-
sectional surveys with information on BMI and health care utilisation and/or measures of
productivity loss. The excess service utilisation (and/or loss in productivity) is then estimated using
multivariate regression analysis and then monetised using country-specific cost data. Bottom-up
approaches may also use longitudinal data which provide information on disease occurrence and
health-care utilisation, depending on availability of suitable longitudinal data (Trogdon et al., 2008).
A key area of debate in studies of this kind is whether, and by how much, to discount future costs.
Severens and Milne (2004, p. 399) comment that “neither theoretical nor empirical arguments are
adequate to determine an optimal solution regarding which discounting method and/or discount
rate should be used.” The most commonly used method, uniform discounting using a constant non-
zero discount rate, tends to prioritise immediate treatment at the expense of prevention, thereby
working against long-term public health measures. To attempt to address this issue, some authors
(e.g. Hollingworth et al., 2012) report both discounted and undiscounted rates.
Comparisons of studies
Comparisons of costing studies are complicated by heterogeneity in scope (for example, obesity
only, vs obesity and overweight; number and kinds of conditions included), costs estimated (for
example, full or partial direct or indirect, or both), presentation of results (for example, absolute vs
per capita excess spending, discount rate applied to projected costs, percentage of total health-care
spending), and variations in national health-care systems (for example, proportion of public vs
private sector health-care funding) (Perry et al., 2012; Tsai et al., 2011).
Estimates of costs are also highly dependent on the design of the costing study: Bierl et al. (2013)
compared different costing methods for obesity and concluded that cost outcomes are largely
affected by study designs, such as population size and age, cost categories (medical vs. total), length
of data collection and BMI cut-points. They observed that modelling (simulation) studies tended to
provide the most conservative estimates and highlighted the importance of decision-makers’
awareness of different purposes, strengths and weaknesses of different studies when interpreting
cost outcomes.
Small number of studies
Systematic reviews on both direct costs (e.g. Tsai et al., 2011) and indirect costs (e.g. Trogdon et al.,
2008) associated with adult overweight and obesity have been published, but these examine current
rather than projected costs. Dee et al.’s (2014) systematic review on this topic retrieved just five
studies that examined direct and/or indirect costs of obesity in adult populations using the PAF
method and analyses of cross-sectional and longitudinal datasets. They noted considerable
heterogeneity across these five studies in methodological approaches and findings. There is very
limited published evidence on lifetime costs associated with childhood overweight and obesity. A
recent systematic review on lifetime direct costs of child/adolescent overweight/obesity in the US
(Finkelstein et al., 2014) retrieved just 6 studies. We are unaware of any systematic reviews on
lifetime indirect costs of child/adolescent overweight/obesity and authors have recommended more
research in this area (Finkelstein et al., 2014).
6.2. International/European reviews This section draws a systematic review conducted by Hamilton et al. (in preparation) on the lifetime
direct and indirect costs associated with child/adolescent overweight/obesity. None of the local
materials from WP4 countries examined this topic, so this section is based on the international
systematic review only.
Finkelstein et al review of direct lifetime medical costs
Prior to considering Hamilton et al.’s (in preparation) findings, a brief summary of Finkelstein et al.’s
(2014) systematic review on direct costs associated with childhood overweight and obesity is
provided. This review is a good starting point, as it highlights differences between study
methodologies which give rise to different results. Five of the six studies reviewed by Finkelstein et
al. (2014) are included in Hamilton et al. (in preparation) (that is, all except Thompson et al., 1999) 47.
First, Finkelstein et al. (2014) note that not all studies incorporated excess costs incurred during
childhood. Four of the six studies in their review incorporated costs during adulthood only (two from
age 20 years; Finkelstein et al., 2008; Tucker et al., 2006, one from age 34 years; Thompson et al.,
1999, and one from age 40 years; Wang et al., 2010). Only two accounted for medical costs in
childhood (one from age 6 years; Ma & Frick, 2011, the other from age 12 years; Trasande, 2010).
Second, although all six studies incorporated higher annual medical cost estimates for obese people,
methods for doing so differed across studies. One study (Thompson et al., 1999) used PAFs
associated with five obesity-related medical conditions, a second (Tucker et al., 2006) applied an
increment in cost per BMI unit, and four studies used multivariate regression to estimate costs as a
function of BMI while controlling for other characteristics. Two of these four studies generated age-
specific estimates (Finkelstein et al., 2008; Wang et al., 2010), while the other two assumed that
incremental cost of adult obesity did not vary by age (Ma & Frick, 2011; Trasande, 2010). This latter
approach risks overestimating lifetime costs, since annual costs tend to start small and increase with
age. Trasande (2010), however, built in a downward adjustment of 38% to account for this.
Third, while five of the six studies adjusted for differential life expectancies by BMI status, the
methods used to make these adjustments differed across studies, with different grouping
(stratifying) variables used, notably how and whether smoking status is included; one study
truncated the upper age limit (Trasande, 2010).
Fourth, three of the studies reported estimates separately by gender and age (Thompson et al.,
1999; Wang et al., 2010; Trasande, 2010), while three studies included race as well (Tucker et al.,
2006; Ma & Frick, 2011; Finkelstein et al., 2008).
Importantly, only two of the six studies accounted for BMI transitions over time, though again, the
methods for doing so differed across the two studies. Tucker et al. (2006) applied a growth curve
47
Tucker et al. (1999) was not included by Hamilton et al. (in preparation) as it was published in 1999, while their search parameters specified 2000-2016.
155
equation derived from previously published work, while Trasande (2010) incorporated transition
probabilities derived from a small longitudinal study.
All 6 studies applied an annual discount rate of 3%.
Finkelstein et al. (2014) derived a pooled estimate of lifetime costs on the basis of the 6 studies,
inflating values to 2012 values from the perspective of a 10-year-old child, filling in data gaps by
using values from the Medical Expenditure Panel Survey (MEPS)48 from previously published work.
They estimated that the lifetime direct medical cost from the perspective of a 10-year-old obese
child ranges from $12,660 to $19,630, after taking weight gain in adulthood into account. This
translates to a total cost in US terms of $14 billion based on the number of obese 10 year-olds in the
US today (assuming a cost of $19,000 per capita). This is likely to be only a small portion of the total
costs, since some studies have suggested that indirect costs exceed direct costs (Dee et al., 2014;
Lightwood et al., 2009).
Hamilton et al review
Hamilton et al. (in preparation) conducted their literature search during December 2015-January
2016 and covered the following sources:
1. Library databases: Cochrane, Pubmed; EBSCO (Medline; Academic Search Complete;
CINAHL; EconLit); Embase; Web of Science.
2. Searches of the reference lists of selected articles.
3. Grey literature searches: Google; publicly available databases; national agency websites.
Note that the criterion for lifetime costs was not applied in advance, since a scoping exercise
indicated that this overly narrowed the search. Also, the age limit of the study samples from the
perspective of an obese or overweight child/ adolescent was extended upwards to age 20, so some
of the selected studies do not cover the 0-18 year-old age-group.
In all, 13 studies were included in the review. As already noted, five of these studies were identified
in the review by Finkelstein et al. (2014).
The next section discusses the results of these studies, grouping them in terms of studies conducted
on US data, and on European data, with studies ordered by year of publication. A majority of studies
(8) were conducted in the US, with just 5 from Europe (two from Germany, two from Sweden, and
one from the Netherlands). Also, most studies (8) examined direct costs only; four examined indirect
costs, and just one examined both direct and indirect costs. Studies were relatively evenly split in
terms of the predominant modelling method used (7 micro-simulation, 6 were cohort-based). See
Table T6.1 provides details of each study.
48
This is a nationally representative survey of the civilian non-institutionalized population, administered by the Agency for Healthcare Research and Quality. MEPS includes data on participants’ health services utilization and corresponding medical costs. The data also include age, race/ethnicity, gender, socioeconomic status, insurance status, education, and self-reported BMI. http://meps.ahrq.gov/mepsweb/
6.2.1. Studies of (direct) healthcare costs based on US data (Hamilton et al (2016) review)
6.2.1.1. Tucker et al. (2006): Direct costs (US)
Tucker et al.’s (2006) study aimed to quantify changes in clinical and cost outcomes associated with
increasing levels of body mass index (BMI). A semi‑Markov model was developed to project and
compare life expectancy (LE), quality‑ adjusted life expectancy (QALE) and direct medical costs
associated with distinct levels of BMI in simulated adult cohorts over a lifetime horizon. Cohort
definitions included age (20–65 years), gender, race, and BMI (24–45). The study is noteworthy in
that it modelled BMI rather than weight status categories (overweight, obese). The study also
modelled changes in BMI over time.
The Markov model simulated user-defined adult cohorts, accounting for age, gender, race, and BMI,
over a lifetime horizon. The cycle length was 1 year and there were two health states: Alive and the
absorbing state, Dead. Probabilities of dying were dependent on age, gender, race, and current BMI
throughout the simulation, and first order Monte Carlo simulation was utilized to randomly progress
subjects through the model.
Probability of death was obtained from published estimates based on Third National Health and
Nutrition Examination Survey (NHANES 3; 1988–1994), the First National Health and Nutrition
Epidemiologic Follow-up Survey (NHANES 1 and 2; 1971–1992), and the NHANES 2 Mortality Study
(1976–1992). The model incorporated changes in BMI over time, a step that had not been included
in analyses published prior to Tucker et al. (2006)49. This step was added since in younger people
who were overweight or had moderate obesity, BMI tended to increase over time, but for older
people BMI tended to decrease, regardless of the BMI level. Furthermore, females tended to gain
more and lose less weight over time compared to men. Tucker et al. (2006) assumed that BMI
progression was independent of race. Changes in health utility over time according to age, gender,
and BMI were also applied in the model50.
Simulations were run on four hypothetical cohorts consisting exclusively of either Caucasian or
African-American males or females. Simulated subjects were assigned an integer-defined age
between 20 and 65 years, and a BMI between 24 and 45.
Direct medical costs (global, rather than individual costs for specific conditions, including six ICD-9
categories: neoplasms, mental disorders, circulatory disease, digestive disorders, respiratory
conditions, and musculoskeletal conditions) were obtained from published sources and inflated to
2004. Total direct costs were increased by 2.3% for each BMI unit above a BMI 25 and by 1.3% for
each year increase in age. Males were assumed to consume 21% less in costs than females. A third
party reimbursement perspective was taken and an annual discount rate of 3% was applied.
Results indicated that total excess lifetime costs from age 20 to death were as follows: males, BMI
34 vs 24, $8,704; BMI 44 vs 24, $14,910; females, BMI 34 vs 24, $12,001; BMI 44 vs 24, $22,634.
49
The predicted BMI (Y) at time (t), from baseline, was calculated using the following formula: Y(t) = 0.266 + 0.0014 Age –
0.036 Sex + 0.985 Baseline BMI + (0.759 – 0.0051 Age – 0.026 Sex – 0.016 Baseline BMI)t + (–0.0037 – 0.00011 Age +
0.00033 Sex + 0.00016 Baseline BMI)t2. See Heo et al. (2003). 50
Specifically: Utility for next cycle = 0.5910 + (0.3602 × current utility) – (0.0268 × sex) + ε; see Hakim et al. (2002).
157
Further simulations were run where BMI did not progress over time. Sensitivity analyses were not
performed beyond this.
Tucker et al. (2006) note that a limitation of the health state utilities was that they did not fully
reflect the disability associated with overweight and obesity in the wider population, such as that
related to cardiovascular disease or cancer morbidity; however, they also note the lack of data in
this area.
6.2.1.2. Finkelstein et al. (2008): Direct costs (US)
Finkelstein et al. (2008) estimated age-specific and lifetime costs for overweight (BMI: 25–29.9),
obese I (BMI: 30–34.9), and obese II/III (BMI: >35) adults separately by race/gender. They note that
previous studies on the lifetime costs of obesity used PAF approach, which includes only a limited
number of diseases, and fails to account for confounding and effect modification fully. Also, other
studies estimated costs without accounting for differences in race or grades of overweight and
obesity, even though both medical costs and mortality vary systematically by race and BMI class.
Finkelstein and colleagues (2008) attempted to overcome these shortcomings by using a common
econometric approach combined with age, gender, race/ethnicity, and obesity-specific life tables
generated from nationally representative samples. This approach allows the model to estimate the
annual increase in medical costs without having to identify a comprehensive list of obesity-
attributable diseases. It also allows for uniquely quantifying these costs for each age and BMI class,
as well as by race and gender.
Finkelstein et al. (2008) used data from MEPS gathered 2001-2004 and conducted their analysis in
three steps. First, they used regression analysis to estimate annual age-specific obesity-attributable
medical expenditures for each demographic group using the MEPS data. All medical expenditures
were inflated to 2007 dollars. They applied these expenditure estimates to race-, gender-, age-, and
BMI-specific survival estimates, and used a discount rate of 3%. They presented lifetime medical
expenditure estimates from the perspective of a 20- and a 65-year-old adult in each BMI class. The
difference in the present value of lifetime costs between normal-weight adults and those in each
BMI class provided an estimate of the lifetime costs for someone who remains in that BMI class from
the starting age throughout their adult life.
Sensitivity analyses were conducted to assess uncertainty of the parameter estimates for medical
expenditures, adjustments for differential mortality by BMI class, and potential correlation in
medical expenditures over time.
At baseline, Finkelstein et al. (2008) noted that there was substantial variation in BMI class as well as
in medical expenditure by race and gender: black men had the lowest costs, while white women had
the highest costs; prevalence of (self-reported) class I obesity ranged from 13.6% (white women) to
22.1% (black women). Obese II/III ranged from 7.4% (white men) to 19.3% (black women).
Figure 5.1 shows Finkelstein et al.’s (2008) estimates of annual health care costs stratified by sex,
race and BMI class. For white men and women, regardless of BMI, costs increase annually until
individuals reach their early 60s and then begin to decrease dramatically, primarily due to
considerable increases in mortality. Costs for those who are obese I are slightly higher for young
adults. Beyond this level, costs tend to converge across BMI groups. Results are generally similar for
blacks. However, because of greater mortality rates and lower annual expenditures, the height of
the expenditure curve peaks at younger ages and at lower expenditure levels. Only for overweight
black men and women do the expected costs ever drop below the costs for healthy weight. That is,
even after adjusting for differential survival (with the possible exception of overweight), there are no
savings associated with excess weight at any age.
Figure 6.1. Survival adjusted annual health care costs from the perspective of a 20 year-old
(Finkelstein et al., 2008)
Source: Finkelstein et al., 2008, Figure 1. BMI is self-reported. All estimates are presented in 2007 dollars. Normal is
defined as a BMI of 21–25. Overweight is 25–29.9. Obese I is 30–34.9. Obese II/III is >35.
Table 6.1 shows Finkelstein et al.’s (2008) lifetime excess cost estimates per individual by race and
sex. They note that, with the exception of white women, the lifetime costs of overweight are not
significantly different to zero. The excess costs per person are $21,550 and $29,460 for white
women in obese classes I, and II/III respectively, while they are $16,490 and $126,720 for white men
in obese classes I, and II/III respectively.
Table 6.1. Excess per-capita lifetime direct health care costs attributable to overweight and
obesity (Finkelstein et al., 2008)
Race and sex
Overweight Obese I Obese II/III
Cost % after age
65 Cost % after age
65 Cost % after age
65
White men 630 N/A 16490 10 16720 9
Black men -1150 N/A 12290 28 14580 21
White women 8120 11 21550 16 29460 13
Black women -180 N/A 5340 16 23750 3
Source: Finkelstein et al., 2008, Table 2. Present value is discounted at 3% per annum. All figures represent 2007 dollars.
Overweight = BMI 25–29.9, obese I = BMI 30–34.9, obese II/III = BMI >35. “% After 65” = % of costs that occur for a 20 year
old after the age of 65.
Comparing these results to previous studies, Finkelstein et al. (2008) note that there are two main
reasons for differences: first, because of the inverse relationship between survival and excess BMI,
the difference in costs between those who are obese I and obese II/III is much less than cross-
159
sectional differences reported in earlier studies; and second, a large amount of the difference
between their estimates and studies that used Medicare claims data are likely to be due to their use
of charges, as opposed to payments for quantifying costs (as used by Finkelstein et al.). The results
of the study provide evidence against claims that the net lifetime medical costs of obesity are
negative due to reduced survival (see van Baal et al., 2008, discussed below) and provide strong
evidence for differences in costs depending on both race and gender, though Finkelstein et al. (2008)
note that the reasons for these differences are unclear.
Finkelstein et al. (2008) note a number of limitations to their study. First, cost projections assume
that a person who is obese at age 20 remains obese until death. However, healthy weight individuals
tend to gain weight over time. As a result, their estimates may overstate the actual lifetime costs
attributable to obesity at age 20. Second, estimates are based on current medical technology;
introduction of new technologies may affect both costs and survival. Third, due to data limitations,
the analysis did not include information on weight history, and had to rely on self-reported rather
than measured BMI. Finally, some direct and all indirect costs were not included (e.g. nursing home
care, absenteeism, presenteeism, disability, worker’s compensation, decreased quality of life).
6.2.1.3. Lightwood et al. (2009): Direct and indirect costs (US)
Lightwood et al. (2009) used the Coronary Heart Disease (CHD) Policy Model51 to estimate the
increase during 2020-2050 in adult obesity, obesity-attributable CHD, and obesity-attributable
diabetes, associated with increases in adolescent overweight and obesity, using US data. Both direct
and indirect costs were included.
Adolescent overweight was classified in a binary fashion, i.e. BMI above the 95th percentile of the
US-CDC growth charts. Obesity estimates (ages 12-19 years and age 35 years) were based on
measured BMI data from the National Health and Nutrition Examination Survey (NHANES) (1974-
1974, 1976-1980, 1988-1994, 1999-2000). Using a linear time trend function, Lightwood et al. (2009)
predicted the rate at which adolescents became obese as adults 20 years later. After age 35,
transition probabilities were applied in order to incorporate the natural increase in BMI that occurs
with age. In adulthood, binary classification was again used, i.e. obesity with BMI > 30. Lightwood et
al. (2009) did not assign a CHD risk directly to obesity, but rather through obesity’s effects on
biomarkers (diasystolic blood pressure, levels of HDL-C and LDL-C, and diabetes status).
Simulation models were run on the basis of two adult populations aged 35-64 from 2020 to 2050.
These were identical with the exception that one accounted for the increase in prevalence resulting
from adolescent obesity. Four settings were compared. The first assumed that the same treatment
protocols would continue into the future, while subsequent models introduced increasingly
aggressive treatment regimens for CHD and diabetes.
51
The CHD Policy Model is a computer-simulation state-transition (Markov cohort) model of incidence, prevalence, mortality and costs for US residents aged 35 to 84 years. The demographic/epidemiological sub-model predicts the incidence of CHD as well as death from other causes and the data are stratified by age and gender, as well as risk factors (diasystolic blood pressure, smoking status, levels of HDL-C and LDL-C, BMI, and diabetes status). After CHD develops, the bridge sub-model classifies the CHD event (cardiac arrest, myocardial infarction, angina) and its sequelae for 30 days. Next, the disease sub-model predicts number of subsequent CHD events, revascularization procedures, other causes among individuals with CHD, and CHD deaths. These are stratified by age, gender and history of events. The model has been validated using data from randomised controlled trials for the reduction of CHD events with statins and other risk factors.
Costs were reported in 2007 US dollars, discounted at 3% annually.
Direct costs covered excess healthcare costs associated with obesity, diabetes and CHD, using data
from the Medical Expenditure Panel Survey (MEPS). Indirect costs were defined as social value of
lost productivity attributable to mortality and morbidity due to sick and disability leave, early long-
term disability, and other early retirement and lost workdays due to illness. Employee compensation
was used to measure value of lost productivity (i.e. median annual age- and gender-specific
compensation for full-time and part-time including wage and other benefits). Comorbidities were
taken into account in the calculation of indirect costs. Mortality costs were calculated as the
difference in the annual population multiplied by age- and gender-specific population employment
population ratios multiplied by median wage by age and 10-year age category. Morbidity-related
productivity losses were estimated as the reduced probability of employment attributable to the
diseases among working adults with obesity. These were adjusted by socio-demographic
characteristics, health behaviours and diabetes status. Overall indirect cost excluded the effects of
diabetes. In doing this, it was assumed that the average loss of employment was 63%.
Sensitivity analysis was conducted by comparing results with models that adjusted for obesity in the
absence of CHD, and obesity adjusted for no diabetes.
Results indicated that under existing treatment protocols, current adolescent overweight was
predicted to result in 161 million life years complicated by obesity, diabetes or CHD, a loss of 1.48
million life years, and total excess costs of $254 billion: $208 billion indirect (81.9% of total) and $46
billion direct (18.1% of total). Comparing this to the other three settings of increasingly aggressive
treatment regimes, a maximum of a reduction of 0.42 million life years but no reduction in the
number of complicated life years, and an increase of total costs to $261 billion, was estimated; i.e.
an increase in direct costs by between 18% and 31% depending on the setting, and a decrease of just
2% to 4% of indirect costs, again depending on the setting.
Lightwood et al. (2009) note that costs are incurred primarily due to lost productivity arising from a
higher proportion of younger people being disabled or deceased. Many of these costs cannot be
avoided with currently available medical treatments. They conclude that “Prevention of excessive
weight gain in childhood and adolescence may be the only effective way to reduce the prevalence of
serious chronic conditions and the resulting economic costs.” (p. 2234).
Lightwood et al. (2009) note that their results are conservative, as aspects of both direct and indirect
costs were not included. Additional indirect costs include household production losses, unpaid
caregiving, and costs arising from pain and restricted mobility. Additional direct costs arise from
other diseases and complications including liver disease, pregnancy complications, musculo-skeletal
complaints, surgery complications, asthma, kidney disease and chronic obstructive pulmonary
disease, were not included in the model.
6.2.1.4. Fernandes (2010): Direct healthcare costs (US)
Fernandes (2010) developed a Monte Carlo model (with simulations based on 10,000 individuals) to
estimate the direct lifetime costs associated with childhood obesity, on the basis of the conceptual
model shown in Figure 5.2. The model consists of three components. The first provides estimates of
obesity among the school-aged population until death; the second estimates tracking of obesity
161
from childhood to adulthood; and the third estimates age-specific costs due to obesity. Analyses
were stratified by race and gender.
Figure 6.2. Conceptual model for lifetime direct costs of childhood obesity (Fernandes, 2010)
Source: Fernandes, 2010, Figure 5.1 (red lines depict key relationships; black lines indicate relationships for which there is
weaker evidence; dotted black line indicates mixed evidence).
Prevalence of child obesity was estimated from NHANES (2001-2006), defined as >95th percentile on
the US-CDC growth charts. To provide the information on BMI tracking from childhood to adulthood,
estimates were derived from 13 longitudinal studies that provided two measurements of BMI, the
first at ages 6-11 years and the second at ages 20-35 years. Fernandes (2010) estimated the
probabilities of being obese as an adult based on the distribution of the samples52. Healthcare costs
were initially estimated on the basis of 12 studies that reported per-capital annual costs, adjusted to
2008 dollars and discounted at a rate of 3%. These studies differed in terms of the design (cross-
sectional vs prospective) and inclusion/exclusion of prescription medications, so “best guess”
estimates for costs were derived from cross-sectional estimates of the Medical Expenditures Panel
Survey (MEPS).
Fernandes (2010) presented two sets of costs, noting that the difference between the two is largely
due to the discounting of future costs. Assuming excess costs begin at age 30, the per-capita excess
lifetime costs associated with childhood obesity were estimated at $8,399 for males and $9,812 for
girls. Assuming excess costs begin at age 9, the per-capita excess lifetime costs associated with
childhood obesity were estimated at $12,047 for males and $15,639 for females. Fernandes (2010)
also estimated that if there was a 1% reduction in obesity prevalence among children, a saving of $1
billion could be incurred.
Sensitivity analyses were conducted by increasing mortality risk by 2% and 5% at ages 30-65 but this
had little effect on costs (e.g. for a 5% increase, lifetime costs were estimated to decrease by 1%).
Also, lifetime costs were estimated to be 4% lower using IOTF cut-points rather than US-CDC cut-
points for childhood BMI53. In a third sensitivity analysis, Fernandes (2010) allowed for the fact that
some adults who are obese at age 30 become healthy weight later. This resulted in reductions of
52
Probabilities for obesity in adulthood given childhood obesity for males and females were 56% and 57% respectively, and for non-obese children, these were 13% and 15% in males and females, respectively. 53
Fernandes (2010) notes that the IOTF cut-points may be more appropriate in longitudinal studies that track individuals from childhood to adulthood.
lifetime costs of 2-3% in males and 6-7% for females. In a fourth sensitivity analysis, the discount
rate was varied by 0% to 5% and the results show that discount rate applied results in large
differences in lifetime cost estimates, as one would expect.
A limitation noted by Fernandes (2010) is that the relationship between obesity status and health
care costs remain constant over the lifespan, and she suggests that this could be addressed by future
studies.
6.2.1.5. Trasande (2010): Direct healthcare costs (US)
Trasande (2010) used data on obese and overweight US 12 year-olds in 2005 and applied a cost-of-
illness approach, projecting three consequences suggested in the literature: additional health care
expenses during childhood, additional adult health care expenses that can be attributed to
childhood obesity/overweight, and QALYs lost by obese/overweight adults who were
obese/overweight children. He then simulated the 2005 cohort over its lifetime, assuming a one
percentage-point decrease in the prevalence of obesity.
Data were taken from the National Health and Nutrition Examination Survey (NHANES) 2003-2006
and applied to US census data to calculate numbers of obese and overweight children. Data from the
2005 Nationwide Inpatient Sample (NIS) and the 2001-2005 Medical Expenditure Panel Survey
(MEPS) were used to calculate annual per patient medical expenses attributable to childhood
obesity/overweight; this value was multiplied by the number of children in the cohort.
Sensitivity analyses explored the results of different scenarios. He first added two cohorts, aged 6
and 19 years, and calculated the outcome for each when obesity was reduced by one percentage
point, just as for 12 year-olds. Then he varied the outcome of the intervention in each of the three
cohorts to a 1% reduction in overweight. Finally, for each cohort, he assumed a simultaneous
reduction in obesity by 1% and an increase in overweight by 1%. Because the study relied on a
mathematical model, he also performed a sensitivity analysis to assess the impact of varying
uncertain parameters in the model, and applied a range of discount rates (0-5%).
Estimates of adult overweight/obesity attributable to overweight/obesity in childhood were derived
from the Fels longitudinal study (which followed about 350 individuals from birth to age 39). Adult
expenditures attributable to childhood obesity were calculated by multiplying the number of obese
or overweight adults who had been obese or overweight as children by the increase in per patient
medical spending, based on published analyses of MEPS 2006. To estimate QALYs lost in adulthood,
Trasande (2010) multiplied the number of adult cases of obesity and overweight attributable to
childhood obesity by the number of QALYs lost among obese and overweight adults in a nationally
representative sample (again based on published analyses).
Trasande’s (2010) results indicate that during childhood, children who were 12 in 2005 were
estimated to incur $2.77 billion in attributable medical expenses. An additional 325,254 adults were
estimated to be overweight and an additional 252,295 adults obese as a result of elevated BMI in
childhood. These obese adults were expected to incur an estimated $3.47 billion in additional
medical expenditures because they were obese or overweight as children, and 2,102,522 QALYs
were estimated to be lost as a result of elevations in childhood BMI.
163
Comparing the various scenarios for a 1% reduction in excess BMI, Trasande (2010) estimated that
between $0 and $87.7 million could be saved in child healthcare spending, between $66.7 and
$403.0 million could be saved in adult healthcare spending, and between 16,158 and 1563,308
QALYs could be saved, depending on the age cohort and assumptions underlying the 1% reduction.
The pattern of results indicates that reductions later in childhood generally produced higher cost
offsets and QALYs saved, while one-percentage-point reductions in childhood overweight achieved
the most modest economic benefits and gains in QALYs. The intermediate scenario, simultaneously
reducing obesity by 1% and raising overweight by 1%, still produced large economic and QALY
savings, mainly driven by the medical expenses and high QALY losses endured by obese adults. All
scenarios, however, indicated that large investments in preventing obesity would be cost-effective if
they reduced the prevalence of obesity and overweight. For example, a $2.03 billion investment
($1,526 per child with elevated BMI) would produce a cost-effectiveness ratio of $50,000 per QALY if
it reduced obesity by 1%.
It should be noted that Trasande (2010) did not estimate additional costs of a healthy weight adult
who was an overweight or obese child. Another limitation of the study is that the Fels data, on which
the population projections were based, was not representative of the US population. However, he
argues that alternative data sources would not have influenced the results to the same extent as age
of intervention, discounting rate, or elevated BMI category (as reported in his sensitivity analyses).
Trasande (2010) concluded that “prevention is widely agreed to be the wisest approach to reduce
downstream consequences of this [childhood obesity] epidemic. … additional research into
interventions is necessary and … even some costly interventions of uncertain efficacy may be worth
pursuing. … this analysis underscores the need to focus on preventing childhood obesity and
overweight as a cost-effective way to improve the nation’s health” (p. 377).
6.2.1.6. Wang et al. (2010): Direct healthcare costs (US)
Wang et al. (2010) used health-care expenditure data from MEPS 2000 to estimate the direct
lifetime costs associated with adolescent obesity (at age 16-17 years). They incorporated costs into
their model from age 40 onwards, and discounted at an annual rate of 3% to a 17 year-old’s
perspective in 2007. The model incorporated progression in BMI from age 17 to 14 using the most
recently available probability estimates.
Excess costs were estimated as follows: Obese compared with healthy weight: Male: +$10 307;
Female: +$9526. Overweight compared with healthy weight: Male: -$4050; Female: -$354. In males,
obese compared to healthy weight had 0.59 less QALYs (9.12 vs 9.71) and in females, obese
compared to healthy weight had 1.19 less QALYs (9.24 vs 10.43). Despite the fact that overweight
males were estimated to have lower lifetime medical costs than healthy weight males, the authors
demonstrate that a 1% reduction in obesity among 16-17 year-olds would reduce (discounted)
lifetime medical costs from age 40 by $586.3 million. Furthermore, different conclusions may have
been drawn if Wang et al. (2010) had incorporated costs incurred prior to age 40.
6.2.1.7. Ma & Frick (2011): Direct healthcare costs (US)
Similar to some other researchers that have examined lifetime costs associated with childhood
overweight and obesity (e.g. Trasande, 2010), Ma and Frick (2011) investigated the age level(s),
levels of effectiveness, and costs at which an early intervention for obesity be affordable.
They used data from the National Health and Nutrition Examination Survey (NHANES) 2003-2006
which includes measured BMI, MEPS 2006, and conducted a literature review to obtain information
on obesity prevalence over time. Based on this review, they applied persistence in obesity rates of
50% at age 6, 76% age 7-12, and 86% at age 13-18. These persistence rates could then be applied to
the data to estimate the proportion of obese adults who had been obese as children or adolescents.
BMI was treated as binary, i.e. obese/non-obese, on the basis of the 95th percentile of the US-CDC
growth curves.
They used a procedure similar to Finkelstein et al. (2008) in modelling, i.e. a two-part strategy that
first estimated differences in probability of health care utilisation by race and obesity status, and
then estimated differences in expenditure among those using any care. Controls were applied in
both models (i.e. race, poverty status, age group, health insurance, region, marital status and
smoking status). They then calculated age, gender and race-specific medical expenditures
attributable to obesity by combining the two parts of the modelling. The difference in expenditure
between normal-weight and obese groups was taken as the excess cost. Applying a 3% discount
rage, costs were aggregated to quantify lifetime medical expenditures from the perspective of
children and adolescents aged 6, 12 and 18 years. This ‘base’ model was then adjusted to examine
cost savings of both population-based and targeted (age-group specific) hypothetical interventions.
Ma and Frick (2011) report that in 2006, annual excess medical costs per capita attributable to
obesity were $1548 for adults and $264 for children (aged 6-17). The base model results indicate
that lifetime per capita medical expenditure attributable to obesity ranged from $19,114 to $40,874,
depending on sex, race and smoking status, and adjusting for life expectancy. Table 5.2 shows the
lifetime costs estimated by Ma and Frick (2011).
Sensitivity analyses excluded by adjusting life expectancies by 5 years, and these resulted in cost
estimates between 4.8% and 8.9% lower, depending on race and gender. Reducing these further to
age 65 in all groups resulted in cost estimates between 20.1% and 24.2% lower. Ma and Frick (2011)
therefore suggest a lower bound of the cost estimates shown in Table 6.2 could be less than 20-25%.
However, they note that their cost estimates could be at the lower bound in any case: productivity
loss and loss due to restricted activity were not included in costs, and potential spillover effects
associated with interventions were not factored into potential savings.
Table 6.2. Excess per-capita lifetime direct health care costs attributable to overweight and
obesity (Ma & Frick, 2011)
Race and sex Lifetime cost Race and sex Lifetime cost
Non-smoking Smoking
White female 40874 White female 37264
White male 32321 White male 28682
Black female 37032 Black female 33782
Black male 25960 Black male 22594
165
Hispanic female 30765 Hispanic female 27588
Hispanic male 23178 Hispanic male 19114
Source: Ma and Frick, Table 2. Costs are discounted at 3% annually. All dollar amounts are the value of the US dollar in
2006.
Ma and Frick (2011) also demonstrated that a current spend of $1.4-1.7 billion could be cost-saving,
even if just a 1% reduction in the obesity rate among children is achieved. Furthermore, based on
savings associated with a 1% reduction in obesity rates, population-based interventions could spend
$280-339 per child, and targeted interventions (i.e. specifically at children with obesity) could spend
$1648-2735 per child, depending on age group. However, their analysis indicates that interventions
targeted at younger children need to be more effective than those targeted at adolescents in order
to achieve equivalent economic returns.
Ma and Frick (2011) note some of the limitations of their study. First, there were limited data on the
persistence of obesity over time. Second, estimation of lifetime costs was based on aggregation of
data from different birth cohorts, as there were no longitudinal data that permitted direct
estimation. Third, there is a lack of data on obesity-specific survival rates (though this was explored
in their sensitivity analyses). They further note that their cost estimates are higher than those of
Trasande (2010), but this is because Trasande modelled costs up to age 55 years while Ma and Frick
modelled costs up to end of life.
6.2.2. Studies of (direct) healthcare costs based on European data (Hamilton et al (2016)
review)
6.2.2.1. Van Baal et al. (2008): Direct healthcare costs (Netherlands)
Van Baal et al.’s (2008) study aimed to estimate the annual and lifetime health care costs conditional
on the presence of obesity and smoking. They used the National Institute for Public Health and the
Environment’s chronic disease (RIVM-CDM) model. This is a dynamic population model that
describes the life courses of cohorts in terms of transitions between risk factor classes and changes
in disease states over time. Smoking was incorporated in three groups (never, former, current), as
was BMI (normal, overweight, obese). Cohorts based on combinations of these risks (with 500 in
each) were simulated until death. Specifically, the three groups were ‘obese’ (never smoked, BMI
>30), ‘smokers’ (current smokers, BMI 18-25) and ‘healthy living’ (never smoked, BMI 18-25). Note
that limiting comparisons to these three cohorts prevents comparisons being made between obese
smokers and others.
Risk factors were linked to 22 obesity and/or smoking-related diseases and used to model the chain
from risk to disease to death. The diseases included in the model (CHD, COPD, stroke, diabetes,
musculo-skeletal diseases, lung cancer, some other cancers, others) are estimated to account for
60% of total morbidity and mortality, and 15% of health-care costs in the Netherlands. Cost-of-illness
data from 2003 for the Netherlands was used to estimate costs. Discount rates of both 3% and 4%
were applied.
Various sensitivity analyses adjusted parameters associated with disease epidemiologies and health-
care costs.
The results indicated that, due to differences in life expectancy (e.g. healthy weight non-smokers
were estimated to live for an additional 4.5 years), lifetime health expenditure was €31,000 higher
per healthy weight 20 year-old, compared to an obese 20 year-old, but €30,000 lower per obese 20
year-old compared with smoking 20 year-old.
Comparing costs for specific diseases in the model, van Baal et al. (2008) note that costs for
musculo-skeletal conditions and diabetes were highest in the obese cohort while costs associated
with cancers and lung cancer were similar. Since differences in total lifetime costs across cohorts are
due to differences in lifetime expectancies, additional health care costs are due to life-years gained
in the healthy cohort, which is likely to suffer from ‘expensive and non-lethal’ diseases – in other
words, the additional life-years come at a price. Van Baal et al. (2008) note that one advantage of
their methodology is that it allows for an explicit causal link between BMI and diseases. Other
studies, using individual BMI data and healthcare utilisation may be confounded with other factors
such as socio-economic status.
However, they note that their models assume that health-care costs per disease are constant across
BMI levels, which may not be the case. They also assumed that no transitions between risk factor
classes occurred over time, which in reality is often the case (e.g. quitting smoking, losing weight).
Furthermore, although their model included 22 diseases, this only accounts for 15% of total health-
care costs. Finally, it may well be the case that the inclusion of indirect costs associated with obesity
(and smoking) would give rise to different conclusions. It would appear that the application of
differential mortality risk by BMI status is of key importance in these studies, since Finkelstein et al.’s
(2008) analysis indicates cost savings. Van Baal et al.’s (2008) conclude that “sound estimates of
medical costs in life-years gained should be taken into account in cost-effective analysis of
prevention” (p. e30).
6.2.2.2. Sonntag et al. (2015): Direct healthcare costs (Germany)
Sonntag et al. (2015) note that published literature on this topic suffers from the significant
shortcoming that the evidenced epidemiological impact of childhood obesity on the development in
adulthood is not translated into economic calculations. They cite Fernandes (2010; also reviewed
here) as the only paediatric study to date that has considered the long-term economic impact of
childhood obesity. However, because of differences in the prevalence of overweight and obesity and
in healthcare systems, the results of Fernandes’ (2010) study are not easily transferable to Germany.
Theirs is the first European cost-of-illness study that quantifies the lifetime burden of paediatric
overweight and obesity.
Their study on direct costs in Germany can be considered alongside their study on indirect costs
(described below): both use similar methodology. Markov modelling was used to make these
estimates. The model runs in two parts – childhood and adolescence (ages 3-17 years), and
adulthood. Children enter model 1 at age 3, moving between BMI states annually until age 18, when
they move either to model 2a (always healthy weight) or model 2b (overweight or obese at any time
167
in model 1). State transition probabilities were used to model the probability of moving between
BMI states.
Age- and sex-specific BMI percentiles were used to classify children as healthy weight (< 90th
percentile), overweight (<90th percentile-97th percentile) or obese (>97th percentile) of national BMI
reference curves for Germany. To incorporate risk information based on BMI status history, memory
was introduced into the Markov model by distinguishing between healthy weight/healthy weight
after overweight/overweight after healthy weight/overweight after obese. Children who enter
model 2a comprised the comparison group for the analysis and the difference in total costs between
the two models is taken as the excess indirect cost attributable to childhood overweight and obesity.
Sex- and age-dependent population death rates were based on the most recently available German
life table. Excess mortality risk due to overweight and obesity was calculated using a longitudinal
study from the US, weighted using age-specific obesity prevalence data from Germany. Estimates of
overweight and obesity-attributable costs, stratified by gender and age, were calculated from a
published German top-down cost of illness (COI) study, which included a large number of obesity-
related diseases, e.g. type 2 diabetes, hyperlipidaemia, hypertension, coronary heart disease,
digestive diseases and some cancers for 2002. Direct costs were estimated for in- and out-patient
treatment, rehabilitation, health protection, ambulance, administration, research, education,
investments and other facilities.
The estimation of age- and gender-specific costs per person was made in three steps. First, the
gender-specific number of overweight and obese persons was calculated per age category. Second,
age and gender-specific costs for overweight and obesity were divided by the number of persons in
the respective age, gender and weight categories. Third, costs per overweight or obese person were
adjusted to 2010 Euros and discounted at 3%.
Sensitivity analyses varied the parameters for prevalence of BMI categories (+/-20%); the
assumption of no excess mortality in healthy weight adults who were obese during childhood;
reweighting the German excess cost data using a recent systematic review; and discount rates for
cost data (0%, 5%).
Discounted (3%) lifetime excess costs amounted to €4262 for men and €7028 for women. In men, it
was estimated that 44% of these costs occurred after age 60, 13% of total costs were due to
overweight, and 87% due to obesity. In women, it was estimated that 32% of these costs occurred
after age 60, 5.4% of total costs were due to overweight, and 94.6% due to obesity. The expected
lifetime excess costs were higher for women, primarily due to higher life expectancy and higher
healthcare expenditures. Comparing their estimates with those of Fernandes (2010), Sonntag et al.
(2015) commented that they may be due to differences in cost structures between German and the
US healthcare settings, distribution of BMI among obese groups in the two populations, and possibly
due to variability in cost attribution. It may also be noted that while Sonntag et al. (2015) introduced
costs into the modelling only in the adulthood, Fernandes (2010) included costs from age 9 years.
Sonntag et al. (2015) note some limitations of their study: a majority of these stem from lack of
available longitudinal data on transition probabilities/trajectories for BMI. Second, they applied US-
derived mortality risk estimates to a German population, though this is common practice in studies
of this kind. Third, they note that the top-down COI approach may result in more conservative cost
estimates compared with a bottom-up approach: the latter can account for multiple diagnoses and
interactions between obesity-attributable diseases.
6.2.3. Studies of (indirect) societal costs based on US data (Hamilton et al (2016) review)
6.2.3.1. Amis et al (2014) indirect costs based on US
Amis et al., 2014; is described in Chapter 4 (see also Table A14).
6.2.4. Studies of (indirect) societal costs based on European data (Hamilton et al (2016)
review)
6.2.4.1. Lundborg et al. (2014): Indirect costs (Sweden, comparisons with UK and US)
Hamilton et al.’s (in preparation) review included three somewhat similar studies that estimated
losses in income and/or educational attainment arising from overweight/obesity in adolescents and
young adulthood. Two of these studies (Amis et al., 2014; Neovius et al., 2012b) are described in
Chapter 4 (see also Table A14). The third is a study of Swedish male military conscripts, with
extensions to samples in the US and Britain (Lundborg et al., 2014).
Since military enlistment in Sweden is compulsory, this results in highly precise data for the purposes
of the study (Lundborg et al., 2014). Data from army recruits (at age 18 years, n = 145,193, enlisted
1984-1997 and covering 92% of the population) included objectively measured BMI, a standardised
cognitive test (Enlistment Battery 80), and an interview-based measure of non-cognitive skills
(stability, endurance, initiative-taking, responsibility, and social competence). BMI status was
categorised as underweight, healthy weight, overweight and obese using the conventional cut-
points of 18.5, 25 and 30.
These records were matched to annual earnings for 2003 (for 96% of the original sample). Parental
education and income was obtained from Statistics Sweden for 1980. Results indicated that before
adjusting for cognitive and non-cognitive skills and parental education and income, obese men
earned 17.5% less than normal-weight men, and overweight men earned 6.5% less. After accounting
for cognitive and non-cognitive skills at age 18, sibling effects and parental characteristics, the
respective loss in earnings was estimates at 4.7% and 2.5%, respectively. This pattern was replicated
using additional data sets from the United Kingdom and the United States, where the results were
similar. The UK (earnings at age 42) and US (ages 39-49) estimates confirm that the penalty is unique
to those who were overweight or obese early in life (since they included adjustments for BMI over
time).
The findings also confirm negative associations between cognitive and non-cognitive skills and BMI
for values exceeding 21-23, and Lundborg et al. (2014) provide evidence that the mechanism of the
penalty operates through ‘occupational sorting’ whereby people are sorted into lower occupational
categories. However, health status at age 18 was unrelated to the income penalty. Sensitivity
analyses confirm that the results are robust against the exclusion or inclusion of social insurance
169
benefits on earnings. Lundborg et al. (2014) conclude that “…the rapid increase in childhood and
adolescent obesity could have long-lasting effects on the economic growth and productivity of
nations. We believe that the rationale for government intervention for these age groups is strong
because children and adolescents are arguably less able to take future consequences of their actions
into account” (p. 1593).
6.2.4.2. Sonntag et al. (2016): Indirect costs (Germany)
Sonntag et al. (2016) have estimated the lifetime indirect costs of childhood overweight and obesity
in Germany, noting that few studies have attempted to estimate long-term costs of childhood
overweight and obesity comprehensively, and theirs is the first to provide incidence-based estimates
of indirect lifetime costs. Markov modelling was used to make these estimates. The model runs in
two parts – childhood and adolescence (ages 3-17 years), and adulthood. As was the case in their
study of direct costs (Sonntag et al., 2015), children enter model 1 at age 3, moving between BMI
states annually until age 18, when they move either to model 2a (always healthy weight) or model
2b (overweight or obese at any time in model 1). State transition probabilities were used to model
the probability of moving between BMI states.
Also similar to their study on direct costs (Sonntag et al., 2015), age- and sex-specific BMI percentiles
were used to classify children as healthy weight (< 90th percentile), overweight (<90th percentile-97th
percentile) or obese (>97th percentile) of national BMI reference curves for Germany. To incorporate
risk information based on BMI status history, memory was introduced into the Markov model by
distinguishing between healthy weight/healthy weight after overweight/overweight after healthy
weight/overweight after obese. Children who enter model 2a comprised the comparison group for
the analysis and the difference in total costs between the two models is taken as the excess indirect
cost attributable to childhood overweight and obesity.
Mortality risk estimates were based on age- and sex-specific mortality rates for Germany for 2009,
and for adulthood, they were based on data from the European Prospective Investigation into
Cancer and Nutrition. Relative risk estimates for mortality associated with adult overweight and
obesity were taken from two high-quality studies from a systematic review on this topic.
Indirect costs associated with overweight and obesity were derived from a literature search on BMI-
related costs associated with sick leave, early retirement, and premature death. A human capital
approach was taken in estimating indirect costs.
Sensitivity analyses varied the BMI transition probabilities, assumed no excess mortality of adults
who had been obese in childhood, varying discount rates, varying excess costs, and using the WHO
rather than the German cut-points for child overweight and obesity.
The simulated BMI trajectories indicate that at age 51-60, about 20% of the population with
overweight or obesity had initially developed it during childhood or adolescence. The results show
that individuals with a history of childhood overweight or obesity had significantly higher lifetime
costs than individuals who had not been overweight or obese during childhood. The excess indirect
lifetime costs (i.e. the difference between model 2a and model 2b) was estimated at €4,209 for
males and €2,445 for females after applying 3% annual discount rates. These estimates are 3.15
times higher for males and 4.52 times higher for females with a history of childhood obesity
compared to those without a history of childhood obesity. This equates to a total (current) cost of
€393 billion, or a discounted cost of €145 billion for Germany. A majority of indirect costs are due to
lost productivity during working age.
Sonntag et al. (2016) present two scenarios to evaluate the impact that a reduction in the
prevalence of childhood overweight and obesity could have on indirect lifetime costs. With a
reduction of 14% in the prevalence of childhood overweight and obesity, for example, costs could be
reduced by 4% for males and 2% for females.
171
Tables
Table T5.1. Summary of studies on lifetime costs of childhood overweight and obesity (from Hamilton et al., in preparation)
First author, year
Country Focus Cost components Type of model Age range covered
Definition of BMI status
Key findings
Amis, 2014 US Indirect Educational attainment and income
Longitudinal cohort
12-18 years with a 13-year follow-up
Obese (>95th percentile) (probably US-CDC, not stated)
After adjusting for demographics, family environment, academic history, behavioural (including health-related) variables, self-reported general and mental health, neighbourhood characteristics, adults who were obese at age 12-18 years earned 7.5% less. Females: 8.7% less; Males: 6.0% less; Whites: 5.7% less; Blacks: 11.9% less; Hispanics: 2.0% less. No evidence of an effect of adolescent obesity on on-time high school graduation; small negative relationship between adolescent obesity and the likelihood of attending college; however, obese adolescents who attended graduate were 8.9% less likely to graduate compared to their non-obese peers (females: -12.2%; males: -5.0%; Whites: -12.0%; Blacks: -5.3%; Hispanics: -2.3%) . Results suggest that loss of earnings are mediated through obtaining a college degree.
Fernandes, 2010
US Direct
Medical Expenditures Panel Survey (MEPS). Morbidities not explicitly modelled; costs of obesity drawn directly from studies looked at
Micro-simulation
6-11 years, lifetime costs beginning at age 9 and age 30 years
Obese (>95th percentile) (probably US-CDC, not stated); in adults, BMI > 30
Assuming excess costs begin at age 30, per-capita excess lifetime were estimated at $8,399 for males and $9,812 for girls. Assuming excess costs begin at age 9, the per-capita excess lifetime costs were estimated at $12,047 for males and $15,639 for females (discount rate: 3%; 2008 values; adjusted for life expectancy).
First author, year
Country Focus Cost components Type of model Age range covered
Definition of BMI status
Key findings
Finkelstein, 2008
US Direct
Medical Expenditure Panel Survey (MEPS) 2001-2004 - includes data on participants’ health services utilization and corresponding medical costs. Used regression analysis to determine annual age-specific obesity-attributable medical expenditures for each demographic group.
Multiple cross-sectional cohort. No possibility to transition between weight categories through life.
20 years, lifetime costs
Overweight (BMI: 25–29.9), obese I (BMI: 30–34.9), and obese II/III (BMI: >35) adults; self-reported BMI
Lifetime costs from a societal perspective, a private employer or insurer’s perspective, and from a Medicare perspective. Discounted at 3% and adjusted for life expectancy. From age 20 years: Overweight Male $0, Female $8,120; Obese I Male $16,490, Female $21,550; Obese II/III Male $16,720, Female $29,460. Inflated to 2007 values.
Lightwood, 2009
US Direct and Indirect
Direct medical costs (hospital in-patient only) and indirect productivity costs from morbidity and premature mortality.
Micro-simulation: CHD Policy Model
12-19 years, costs cover ages 35-64 years: each year between 2020 and 2050, estimated the excess number of 4 mutually exclusive states: life-years lost because of death, prevalent CHD, prevalent diabetes without CHD, and prevalent obesity without CHD or diabetes
Obese (>95th percentile) (US-CDC); in adults, BMI > 30
Cumulative excess total costs are estimated at $254 billion: $208 billion because of lost productivity from earlier death or morbidity and $46 billion from direct medical costs, 2007 values, 3% annual discount rate.
Lundborg, 2014
Sweden Indirect Income Longitudinal cohort
18 year-olds (all male) followed up at age 28-38.
Overweight (BMI: 25–29.9) and obese (BMI: 30+), measured BMI
After accounting for cognitive and non-cognitive skills at age 18, sibling effects and parental characteristics, the respective loss in earnings was estimates at 4.7% and 2.5% for obese and overweight, respectively. Additional analyses using data sets from the United Kingdom and the United States gave similar results. Study provides evidence that the mechanism of the penalty operates through ‘occupational sorting’.
173
First author, year
Country Focus Cost components Type of model Age range covered
Definition of BMI status
Key findings
Ma, 2011 US Direct
Medical Expenditure Panel Survey (MEPS) 2006 - includes data on participants’ health services utilization and corresponding medical costs.
Averaged data from multiple studies
0-6 year-olds, lifetime costs
Childhood obesity > 95th percentile (US-CDC). Obese adult (BMI>30) Vs Normal (BMI 18.5 - 25).
In 2006 values discount rate 3%, lifetime costs adjusting for life expectancies: Obese (BMI>30) Vs Normal (BMI 18.5 - 25): Male: $32 320 ($28 682); Female: $40 870 ($37 260). Authors estimate that for every 1% reduction in obesity rates, at ages 0-6, 7-12, 13-18, could spend up to $2375, $1648, and $1924, respectively.
Neovius, 2012 Sweden Indirect
Lifetime societal productivity losses (sick leave, disability pension and premature death)
Longitudinal cohort
18 - 56 years. Productivity losses extrapolated to 65 year-olds. Males only.
Normal, overweight (>25) and obese (>30). No BMI tracking over time.
Using 3% discount rate; inflated to year 2010 Swedish Kronor. Human Capital approach: €17000 for overweight adolescent; €39800 for obese adolescent (more than half of these productivity losses were related to premature death (€6700 and €27000 respectively). Friction cost method (absent workers assumed to be replaced after 6 months): Overweight €3,500; Obese €6,000.
Sonntag, 2015 Germany Direct
In- and out-patient treatment, rehabilitation, health protection, ambulance, administration, research, education, investments and other facilities.
Micro-simulation 3 years, lifetime costs beginning at age 18 years
Normal, overweight and obese based on national reference curves.
Discounted (3%) lifetime excess costs amount to €4262 for men and €7028 for women.
Sonntag, 2016 Germany Indirect
Cost of lost productivity due to mortality and morbidity: sick leave, early retirement, and premature death
Micro-simulation 3 years, lifetime costs beginning at age 18 years
Normal, overweight and obese based on national reference curves.
Overweight and obesity during childhood resulted in an excess lifetime cost, discounted at 3%, per person of €4,209 (men) and €2,445 (women), with proportion occurring before age 60: 0.81 for males and 0.79 for females.
First author, year
Country Focus Cost components Type of model Age range covered
Definition of BMI status
Key findings
Trasande, 2010
US Direct
Medical Expenditure Panel Survey (MEPS) 2001-2005 - includes data on participants’ health services utilization and corresponding medical costs. Study assumes $50,000 per QALY lost.
Multiple cross-sectional cohort
12 year-olds, lifetime
Overweight (<85th percentile), obese (>95th percentile) (US-CDC); in adults, BMI > 30
3% discount rate, 2005 values. During childhood (up to age 18), US children who were age twelve in 2005 are estimated to incur $2.77 billion in attributable medical expenses. Obese adults will incur $3.47 billion in additional medical expenditures because they were obese or overweight as children, and 2,102,522 QALYs will be lost as a result of elevations in childhood BMI. Male per capita additional lifetime obesity costs due to childhood obesity: $15,850. Females $15,830.
Tucker, 2006 US Direct
Derived from published sources, inflated to year 2004 values, and covered 6 ICD-9 categories: neoplasms, mental disorders, circulatory disease, digestive disorders, respiratory conditions, and musculoskeletal conditions. Total direct costs were increased by 2.3% for each BMI unit above 25 kg m–2, and by 1.3% for each year increase in age. Male subjects were assumed to consume 21% less in costs compared to female subjects.
Micro-simulation 20 years, lifetime costs
BMI integer values, 24-45
Discounted at 3% per annum. Male: BMI 34 Vs 24, $8,704; BMI 44 Vs 24, $14,910. Female: BMI 34 Vs 24, $12,001; BMI 44 Vs 24, $22,634.
175
First author, year
Country Focus Cost components Type of model Age range covered
Definition of BMI status
Key findings
Van Baal, 2008
Netherlands Direct
No precise definition of cost given; covers costs of diseases directly associated with obesity and those of other diseases that tend to occur as life-years are gained are included. Diseases modelled account for roughly 60% of total morbidity and mortality. Inclusion of other diseases entail an increase in the lifetime costs of healthy weight cohort.
Micro-simulation (RIVM-CDM model)
20 years, lifetime costs
Obese (BMI > 30)
Due to differences in life expectancy (e.g. healthy weight non-smokers were estimated to live for an additional 4.5 years), lifetime health expenditure was €31,000 higher per healthy weight 20 year-old, compared to an obese 20 year-old, but €30,000 lower per obese 20 year-old compared with smoking 20 year-old. Discount rates of 3% and 4% applied, 2003 values.
Wang, 2010 US Direct
Medical Expenditure Panel Survey (MEPS) 2001-2004 - includes data on participants’ health services utilization and corresponding medical costs.
Cohort model, incorporating projections of child and adolescent BMI to age 40
16-17 years, lifetime costs beginning at age 40
Adolescents: non-overweight (BMI < 85th percentile), overweight (85th BMI < 95th percentile), and obese (BMI > 95th percentile). Adults: BMI cut-points of 25 and 30 applied.
Medical costs from age 40 upwards (in 2007 values discounted at 3% to age 17 years). Obese compared with healthy weight: Male: +$10 307; Female: +$9526. Overweight compared with healthy weight: Male: -$4050; Female: -$354. In males, obese compared to healthy weight had 0.59 less QUALYs (9.12 vs 9.71) and in females, obese compared to healthy weight had 1.19 less QALYs (9.24 vs 10.43). Despite the fact that overweight males are estimated to have lower lifetime medical costs than healthy weight males, the authors demonstrate that a 1% reduction in obesity among 16-17 year-olds would reduce (discounted) lifetime medical costs from age 40 by $586.3 million.
Source: adapted from Hamilton et al. (in preparation).
CHAPTER 7: OVERVIEW OF MODELLING METHODOLOGY
7.1 EU countries participating in JANPA WP4
Seven European countries are participating in JANPA WP4: Croatia, Italy and Portugal chose to
participate in basic studies (n = 3) while Greece, Ireland, Romania and Slovenia chose to participate
in advanced studies (n = 4).
Figure7.1 European countries participating in EU JANPA WP4
Differences between advanced studies and basic studies are given in Table 7.1 below.
Table 7.1. Differences between basic and advanced studies
Basic studies Advanced studies
Health impacts only
Also include other impacts
Major clinical conditions only
Wider range of clinical conditions
Focus on core pan-European data Also include country-specific data
Focus on “top-down” approaches using international inputs (and possibly) local inputs
More emphasis on “bottom-up” approaches using local inputs
Inputting of (adjusted) data from proxy countries
More complex data imputation methods involving country-specific data
Greater involvement in validation studies
7.2 Governance
7.2.1 Expert International Scientific Advisory Committee (ISAC)
177
An expert International Scientific Advisory Committee (ISAC) guides the scientific aspects of JANPA
WP4.
The ISAC:
Gives scientific advice to WP4 Lead Team.
Reviews background materials and draft reports
Attends ace-to-face meetings
Participates other telecalls (if needed)
Members of the ISAC are:
Associate Prof Jennifer Baker, Institute of Preventive Medicine in Denmark and the
University of Copenhagen. Denmark
Dr Margherita Caroli, Nutrition Unit, Department of Prevention, Azienda Sanitaria Locale
Brindisi. Italy
Dr Anne Dee, Health Service Executive. Ireland
Dr Tony Fitzgerald, Department of Statistics and & Department of Epidemiology & Public
Health. University College Cork. Ireland
Prof David Madden, School Of Economics, University College Dublin. Ireland
Dr Martin O’Flaherty, University of Liverpool. England
Dr Pepijn Vemer, Department of Pharmacoepidemiology & Pharmacoeconomy, University of
Groningen. Netherlands
Prof Kevin Balanda, Institute of Public Health in Ireland. Ireland
7.2.2 Study principles
The seven principles that will underpin the design, implementation and reporting of JANPA WP4 are
outlined in the Table 7.2 below.
Table 7.2: Principles underpinning JANPA WP4
1. Relevance to JANPA WP4 countries and EU
2. Societal economic perspective that, in addition to health impacts and healthcare costs, includes important aspects of public health and impacts and costs experienced by society and its communities
3. Transparency that explains strengths but recognises assumptions and limitations
4. Capacity building in JANPA WP4 countries and EU (research and information)
5. Identifying gaps in research and information
6. Stimulating further development of research and information
7. Health equity
7.3 JANPA WP4 aims and objectives
The aim of JANPA WP4 is to “contribute to the evidence-based economic rationale for action on
childhood obesity”.
Its modelling objectives are, in the seven EU countries participating in JANPA WP4, to:
1.a Describe the current prevalence and trends in childhood overweight and obesity. 1.b Estimate the lifetime impacts and costs of current childhood overweight and obesity. 1.c Breakdown these impacts and costs according to the year they occur 2. Assess the effect of reducing childhood obesity by 1% and 5% on these impacts and costs
3. Explore the feasibility of generalising the JANPA WP4 modelling methodology to other EU
countries.
JANPA WP4 is essentially a modelling project. However, during development of its Study Protocols it
became clear that several important issues could not be incorporated into the modelling. A number
of non-modelling projects that were not initially part of the work package are being developed (see
Appendix 1).
7.4 Challenges
The evidence (see Chapters 1-5) highlights the challenges of estimating lifetime impacts and costs:
Childhood obesity has many impacts and costs occurring in later life54
These later life impacts and costs are usually heavily discounted55
Childhood obesity linked to adult obesity with many shared impacts
Many obese adults were not obese children and so not all adult obesity-related disease is
associated with childhood obesity56
Evidence directly linking childhood obesity and adult obesity and adult diseases is underdeveloped
Many childhood obesity-related diseases are acute conditions unlike chronic diseases that are the focus of most existing studies
54 Sonttag et al found that, amongst males, nearly two-thirds of lifetime excess costs occurred after age 60 years. Amongst
females, this figure was one third.
55 Annual discounting at a rate of 3% pa halves values every 23 years
56 Llewellyn et al. indicated that 70% of adult OW/OB develops in adulthood, while the remaining 30% is a continuation of
OW/OB in childhood
179
• Obese adults who were also obese as children, however, are at greater risk of adult obesity-
related diseases (Juanola et al)57
• Few existing studies incorporate societal impacts and costs although adult productivity loss
is an exception
• Psycho-social impacts and their consequences on mental health, school attendance and educational outcomes may be particularly important in childhood. These are not considered in many existing studies.
7.5 General issues
7.5.1 Incorporating children
Extending UKHF’s existing modelling software involves extending their age ranges, incorporating
impacts and costs that occur in childhood and adulthood, and including acute conditions.
7.5.1.1 Childhood health impacts and direct healthcare costs
Health impacts and direct healthcare costs that occur in childhood are incorporated into UKHF’s
existing modelling software by adding relevant obesity-related diseases (and associated costs) as
impacts and copsts that occur at an earlier age.
The development of the list of childhood is described in Section 8.4.1.
7.5.1.2 Adult health impacts and direct healthcare costs
Broadly speaking, impacts and costs of childhood obesity occur in adulthood for a three possible
reasons:
Obese or overweight children are more likely to be obese or overweight
Being an obese or overweight child may increase the risk of obesity-related diseases associated with adult obesity
The risk of adult disease may be increased even if obese or overweight children do not become obese or overweight adults.
Two approaches to the development of the list of the adult diseases that were to be included in the
modelling were available. The first approach was to use the adult diseases that are significantly
associated with childhood obesity; the other was to use the adult diseases that are significantly
associated with adult obesity.
A list of adult diseases that are significantly associated with childhood obesity was derived from:
The systematic review of the international literature that was undertaken by the Irish
national team with significant additional funding from safefood
Local materials gathered from participating countries during the Local Materials Survey.
57
As described by a meta-analysis by Simmonds et al. (2016), “around 55% of obese children go on
to be obese in adolescence, around 80% of obese adolescents will still be obese in adulthood and
around 70% will be obese over age 30.”
The use of this list (see Table 11.1) was the preferred approach but was rejected because:
Incorporating these diseases into the modelling required the inclusion of lag-times between childhood obesity and the development of adult diseases. While it is technically possible, reliable estimates of these lag-times are not available for most of the diseases.
Several adult diseases known to be associated with adult obesity (which, in turn, is associated with childhood obesity and overweight) were not included because of the lack of existing research and some of the challenges of establishing links between childhood obesity and its adult consequences
The agreed approach, then, was to:
Use an updated version of the list of adult diseases used in the 2012 Irish adult obesity study and WHO (Europe) 53 county study.
Use the list of adult diseases that are significantly associated with childhood obesity to validate model outputs (see Section 11.1.1).
This approach is methodologically simpler as the 4-state semi-Markov process that is used to
simulate disease and deaths requires disease risks in adulthood.
See Section 8.4.2 for details.
7.5.1.3 Acute conditions
UKHF’s existing modelling software implements chronic disease models. To incorporate acute
conditions we first split childhood diseases into two groups - chronic conditions and acute conditions
- and assume that acute conditions last one year.
7.5.2 Incorporating societal impacts
Because of the general lack of research and data in this area, these issues may be explored in
possible non-modelling projects (see Appendix 1).
Modelling will focus on adult productivity losses and lifetime income losses. We will use the list of adult societal impacts significantly associated with childhood obesity to validate model oututs (see Section 11.1.1)
7.5.2.1 Adult productivity losses
Adult productivity losses will include those due to premature mortality as well as those due to work
absenteeism. In each JANPA WP4 country, an estimate of the (per case) annual costs of adult
productivity losses is required58. For this type of calculation, economists use either a Human-Capital
approach or a Friction-Cost approach although it is unclear which is the most appropriate application
of the Friction-Cost approach when an accumulation over many years is used. Therefore, like the
majority of cost studies, we will use the Human-Capital approach.
Loss due to absenteeism
58 Ideally, the annual per disease cost would be be brokendown by disease. These are not avaailable in this project.
181
The use of QOL–productivity curves (combined with disease-specific utility weights) appears
promising but their development is at a formative stage (UK NICE Report). Instead we will directly
incorporate adult productivity losses and associated costs into the adopted the modelling software
as another chronic impact that commences when an individual develops an obesity-related disease
and ends on their expected retirement age. If a working individual dies prematurely from an obesity-
related disease, adult productivity is calculated from age-of-death until their death.
Loss due to premature death
7.5.2.2 Lifetime Income Losses
Significant lifetime income loss is associated with being overweight or obese at age 18 years (see
Section 4.5). We will directly incorporate lifetime income losses into the modelling software as
another (chronic) impact that commences if an individual is obese or overweight when they turn 18
years of age and ends on expected retirement age.
7.6 Model inputs, outputs and metrics
Model outputs of JANPA WP4 will be expressed in terms of excess metrics and effect metrics.
7.6.1 Impact-cost indicators
Excess and effect metrics are constructed from impact-cost indicators which describe various aspects of children’s lifetime experiences such as:
Number of new disease cases
Direct healthcare costs
Adult productivity losses
Number of deaths
Number of premature deaths
Potential Years of Life Lost (PYLLs)
Quality Adjusted Life Years (QALYs)
Disability Adjusted Life Years (DALYs)
These indicators are outputted by the modelling software.
7.6.2 Excess Metrics
Excess metrics describe excesses in these impact-cost indicators that are associated with current childhood obesity. They are differences between the value of an indicator amongst individuals who were overweight or obese as children and its value amongst individuals who were of healthy weight as children:
Excess metric = indicator (individuals who were OW/OB as children) - indicator (individuals who were HW as children)
An example of an excess metric is the number of diabetes cases that are associated with current
childhood obesity.
Excess metrics can be expressed as total differences or per person differences.
7.6.3 Effect Metrics
Corresponding to each excess metric there is an effect metric that describes the effect of a reduction in childhood obesity on the excess. Effect metrics are differences between the value of an excess metric in one of the reduction scenarios and its current value:
Effect metric = excess metric (current childhood obesity) – excess metric (reduced childhood obesity)
The current value of an excess metric thus serves as the base case for the assessment of the effect of a reduction in childhood obesity. A positive effect represents an improvement (ie a reduction in an excess associated with current childhood obesity). An example of an effect metric is the effect of a 1% reduction in current childhood obesity rates on the excess number of diabetes cases that are associated with childhood obesity. Another would be the effect that a 5% reduction in current childhood obesity rates on the lifetime excess direct healthcare costs that are associated with childhood obesity.
Effect metrics can be expressed as either total changes or per person changes.
7.7 Modelling
7.7.1 Cohort simulation studies
Ideally, lifetime costing studies of childhood obesity would be based on comprehensive longitudinal
studies with long term follow-up. Such studies are rare (see Chapter 6).
Instead, existing lifetime costing studies link childhood obesity to adult consequences through its link
to adult obesity within a simulation model that uses modelled individual lifetime BMI trajectories.
The relatively small number of existing lifetime costing studies (see Chapter 6). These studies use
cohort simulation models to estimate the lifetime impacts and costs of childhood obesity. They
compare impacts and costs amongst individuals who were overweight or obese in childhood to
impacts and costs amongst individuals who were of healthy weight in childhood.
Cohort simulation models are described in Figure 7.2.
Figure 7.2. Modelling approach to lifetime costing studies of childhood obesity
183
This approach allows us to use existing adult obesity-related impact and cost research and data.
7.7.2 Modelling steps
7.7.2.1 Modelling lifetime BMI trajectories
The modelling software first uses multivariate regression (age, sex, year) analysis of historical BMI
data to forecast age-sex population BMI distributions in future years. The BMI of each virtual
individual is initialised in the modelling start-year. In future years, it is assumed that everyone stays
at the same BMI percentile within their age-sex peer group as they age (the “constant BMI
percentile assumption”). The BMI distribution of their age-sex peer group is assumed to follow the
forecasted age-sex population BMI distributions as they age. More details are given in Section 9.4.
7.7.2.2 Simulating health impacts
As virtual individuals are aged one year at a time, their BMI values follow their modelled lifetime BMI
trajectories. Their BMI category may change as the BMI distribution of their age-sex peer group
changes as they age.
A 4-state semi-Markov process is used to simulate the occurrence of obesity-related disease and
disability, other diseases and death. The childhood and adult obesity-related diseases are listed in
Section 8.4. It is assumed that these diseases are chronic conditions that last the rest of life.
5This presentation is part of the Joint Action JANPA (Grant agreement n°677063) which has received funding from the European Union’s Health Programme (2014-2020)
Life mecostofchildhoodobesitystudies
Obesity-relatedtreatment&deaths
Obesity-relateddiseases
RRs
AdjustedQOLmeasures&costs
Deathsfromothercauses
Adultproduc vitylosses
Costsofadultproduc vitylosses
For each obesity-related disease, an independent semi-Markov process simulates each virtual
individual’s disease status (including death) in all future years.
The disease processes associated with the different obesity-related diseases are assumed to operate
independently; the modelling does not take into account conditional relative risks arising from multi-
morbidities, but individuals can still get multiple conditions in the simulation.
More details are given in Section 9.5
7.7.2.3 Modelling societal impacts
The modelling incorporates adult productivity losses due to premature death and absenteeism.
Productivity losses due to reduced productivity at work (presenteeism) are not included in the
modelling. Adult productivity losses commence when an individual develops an obesity-related
disease and are incurred through their whole lifetime.
Lifetime income losses are also incorporated into the modelling. This is based on the childhood
obesity status and are also assumed to last over the working life.
More details are given in Section 9.5
7.7.3. Adaptation of UKHF’s modelling software
UKHF has been contracted to undertake the modelling for EU JANPA WP4.
Substantial adaptations to the UKHF’s modelling software will be necessary to accommodate:
• Use of cohort simulation models rather than population simulation models
• Incorporation of children with shorter term impacts into the models
• Use of a societal economic perspective rather than an exclusively health services perspective
• Use of more complex metrics and reporting associated with lifetime costing studies
7.8 Reporting
To manage budget, IPH IRL will undertake a number of the routine data collation and reporting tasks
including the calculation of model metrics and production of graphical outputs.
The virtual individuals’ simulated BMI trajectories and their impact and cost experiences will be
summarised by UKHF in Model Output Tables that will be used by IPH IRL to calculate relevant
excess and effect metrics.
More details are given in Chapter 10
185
7.9 Validity and generalisability
7.9.1 Validity
Validation studies will address the validity of country-specific findings as well as the effect of
research data and modelling assumptions on model outputs.
Examples of findings-based validation studies include comparisons of model-based estimates of the
RRs of adult diseases and societal impacts associated with childhood obesity to the existing
estimates in the research literature
Examples of methods-based validation studies will address the methods used to model lifetime BMI
trajectories and the independent disease processes assumption
Sensitivity analyses will, inter alia, consider imputation methods.
Further details can be found in Chapter 11.
7.9.2 Generalisability
There is interest in knowing if the JANPA WP4 modelling methodology can be extended to other EU
member countries. Amongst other things, we will compare basic models and advanced models in
JANPA WP4 participating countries in advanced studies and develop a toolbox of modelling
resources for undertaking the modelling in other EU countries.
Further details can be found in Chapter 10.
CHAPTER 8: MODEL INPUTS, OUTPUTS AND METRICS
8.1 Research and data domains
Simulation modelling is research-intensive and data-intensive and requires inputs from seven
domains:
1. Population 2. BMI 3. Health impacts 4. Direct Healthcare costs 5. Societal impacts 6. Societal costs 7. Other
Table 8.1. Model inputs
1. Population
Current population estimates
2. BMI
Current distribution and historical trends in BMI
3. Health impacts
Relative risks / odds ratios of obesity-related diseases (and risks in healthy weight individuals)
Current incidence/ prevalence rates for
each obesity-related diseases
all other causes (combined)
Current annual mortality rates for
each obesity-related disease
all other causes (combined)
all causes
4. Direct healthcare costs
Annual direct healthcare costs for each obesity-related disease (either total or per case costs)
5. Societal impacts
Annual adult productivity loss for each obesity-related disease
187
Lifetime income losses
6. Societal costs
Annual cost of adult productivity losses (either total or per case costs) for each obesity-related disease
National income data
7. Other
Age used to define a premature death59
National BMI cutoff-points and references curves (if they exist and are different that IOTF cut-off points).
Discounting rate (for future disease and disability and costs) used in national studies (if relevant)
National utility weights (if relevant)60
Figure 8.1 explains how research and data from these domains is used in the UKHF’s modelling
software.
Figure 8.1. Use of research and data in the modelling software
59 For this, the national life expectancy of each of the one-year age groups at their year of birth will be used 60 EuroQOL’s EQ5D will be used in the utility weights for QALYs and GBoD disability weights will be used for DALYs . If national data is not available in a particular JANPA WP4 country, agreed proxy data will be used.
8.2 Population estimates
Age-sex specific population projections are not needed in cohort simulation models which focus on
the lifetime experiences of the initial cohort; they not only relevant to population simulation models
which focus on the experiences of whole populations in future years.
Five year age bands (0-4, 5-9,….70-74, 75+ years) will be used in the simulations.
8.3 BMI
8.3.1 Current BMI distribution and trends
Self-reported height and weight will not be used to calculate BMI in historical data used to forecast
BMI distributions unless absolutely necessary.
Current childhood obesity rates will be based on the age-sex population BMI distributions from the
most recent cross-sectional population BMI data, projected to 2016 if necessary.
The modelling of lifetime BMI trajectories will be based on current (2016) and the historical trends in
BMI or all age-sex categories.
Current obesity trends in children and adults will be based on the forecasted age-sex population BMI
distributions derived by fitting regression models (involving SEX, AGE and CALENDAR YEAR as
independent variables) to the available historical cross-sectional population BMI data (see Section
7.7.2).
Some studies have found that recent falls in childhood obesity have been limited to youngest age
group and not adolescents (Irish study team – first systematic review). In this study, childhood
obesity has been broken down into three age categories61 62 63 :
Younger children: 0-6 years
Older children: 7-11 years
Adolescents: 12 – 17 years
These age categories will also be used in the Model Output Tables.
61
Splitting 12-17 year olds up might be useful in order to capture / adjust for the effects of puberty
62 There are more drivers of obesity if a child is going to school. So it might be better to find out the ages
for school rather than just use absolute age range.
63
It is important to not group the 20’s and 30s together as the effects of pregnancy might confound differences
189
8.3.2 Reductions in childhood obesity
1% and 5% reductions in childhood obesity rates will be expressed in terms of changes in the mean
age-sex specific BMIs. The variance of the BMI distribution will assumed to be unchanged.
We assume that the 1% and 5% reductions in childhood obesity rates occur instantaneously in 2016 and not gradually over a period of time.
Finally; we assume that the current trends in childhood obesity are unaffected. Of course, a virtual individual’s BMI category (and therefore their annual risk of an obesity-related disease) may change as they age.
8.4 Health impacts
To be incorporated into the modelling for a particular JANPA WP4 country, a disease (or a societal
impact) must satisfy two criteria:
It must be significantly related to childhood obesity in the evidence base
The necessary data must be available in each WP4 country
The (childhood and adult) diseases and societal impacts that are included in a county’s model are
therefore determined in a two stage process:
1. Firstly, a global list of diseases and societal impacts that are significantly related to
(childhood) obesity and overweight are identified from a review of international and local
materials.
2. Secondly; diseases and societal impacts for which inadequate local data or acceptable proxy
data or a variable are removed.
8.4.1 Childhood disease risks
Table 8.2 lists the childhood diseases that are significantly related to childhood overweight and
obesity in the literature. The list was derived from:
A systematic review of the international literature that was undertaken by the Irish National
Team with significant additional funding from safefood
Local materials gathered from JANPA WP4 countries during the Local Materials Survey.
Consultations with clinical experts and JANPA WP4 countries.
See Chapter 3 for further details.
In each JANPA WP4 country, this global list was reviewed in light of available local data or acceptable
proxy data and the disease list used in that country’s modelling was finalised.
Table 8.2. Global list of childhood diseases that are significantly related to childhood overweight
and obesity
Disease Odds Ratio (95% Confidence Interval)
(compared to children with healthy weight)
Quality of evidence
Overweight Obese
Asthma Adjusted:
1.23 (1.17, 1.29)
Unadjusted:
1.43 (1.33, 1.54)
Moderate with conflicting
findings in terms of
association
Wheezing disorders Unadjusted:
1.23 (1.17–1.29)
Adjusted risk:
1.30 (1.19–1.42)
Unadjusted risk:
1.46 (1.36–1.57)
Adjusted risk:
1.60 (1.42–1.81)
Moderate with conflicting
findings in terms of
association
Metabolic
syndrome (MetS):
Study 1: For every one unit increase in zBMI the odds ratio of meeting criteria for metabolic syndrome is 2.4 (1.21 – 4.63).
Good but often defined by different criteria but let’s break down to HB, type 2 diabetes, type 1 diabetes and hyperlipidemia? Study 2:
67.33, (21.32–212.61) Study 2: 249.99, (79.51–785.98)
High blood pressure Study 1: Males: 4.11(3.89–4.34) Females: 5.56 (5.09–6.07)
Good
Study 2: SBP > 140: 2.24; (1.46 – 3.45) DBP 2.10: (1.063–4.17)
Type 2 diabetes Msles: 5.56; (5.09–6.07) Females: 4.42 ( 3.90 – 5.00)
Moderate
Hyperlipidemia Males: 16.07, ( 8.29 – 31.15) Females: 9.00 ( 4.36–18.6)
Moderate
Others
Depression Aged 6–13 years: 3.38, (1.13– 10.1)
Moderate
Musculoskeletal
pain
1.26; ( 1.09-1.45). Good
Obstructive sleep Aged 12+ years: 3.55, (1.30–9.71) Not among younger children
Moderate
191
apnoea
Non-Alcoholic Fatty
Acid Disease
(NAFLD)
13.36 (9.09 - 18.02) 13.74 (9.92 to 19.03) Moderate
Those conditions in Table 8.2 which are less strongly associated with childhood obesity may be
explored in possible non-modelling projects (see Appendix 1).
None of the diseases in Table 8.2 are included in UKHF’s existing modelling software.
8.4.2 Adult disease risks
The international evidence linking adult diseases with adult obesity was initially summarized in the
Irish 2012 adult obesity study (Perry et al (2012)
Table 8.3. Initial list of adult diseases that were included in the Irish cost of obesity study (2012) & WHO (Europe) 53 county study64
* *
*Considered to be acute conditions in the modelling software
This initial list was updated with a review of the literature and consultation with clinical experts and
JANPA WP4 countries.
The global list of adult obesity related diseases that are significantly related to adult overweight and
pobesity is given in Table 8.4 below.
Table 8.4. Global list of adult diseases that are significantly related to adult overweight and
obesity
Condition Overweight male
Overweight female
Obese male Obese female Source
Cancer-Breast, post-menopausal*
1.08 (1.03–1.14) 1.13 (1.05–1.22) Guh et al. (2009)
Colorectal Cancer*
1.51 (1.37–1.67) 1.45 (1.30–1.62) 1.95 (1.59–2.39) 1.66 (1.52–1.81) Guh et al. (2009)
Endometrial Cancer*
1.53 (1.45–1.61) 3.22 (2.91–3.56) Guh et al. (2009)
Oesophageal Cancer*
1.13 (1.02–1.26) 1.15 (0.97–1.36) 1.21 (0.97–1.52) 1.20 (0.95–1.53) Guh et al. (2009)
Kidney Cancer* 1.40 (1.31–1.49) 1.82 (1.68–1.98) 1.82 (1.61–2.05) 2.64 (2.39–2.90) Guh et al. (2009)
Pancreatic cancer*
1.28 (0.94–1.75) 1.24 (0.98–1.56) 2.29 (1.65–3.19) 1.60 (1.17–2.20) Guh et al. (2009)
Gallbladder cancer
1.23 (1.15-1.32) 1.23 (1.15-1.32) 1.15 (1.32-1.74) 1.15 (1.32-1.74) 2007 WCRF/AICR report
Chronic back pain*
1.59 (1.34-1.89) 1.59 (1.34-1.89) 2.81 (2.27-3.48) 2.81 (2.27-3.48) Guh et al. (2009)
Osteoarthritis* 2.76 (2.05-3.70) 1.80 (1.75-1.85) 4.20 (2.76-6.41) 1.96 (1.88-2.04) Guh et al. (2009)
Coronary Artery Disease*
1.29 (1.18-1.41) 1.80 (1.64-1.98) 1.72 (1.51-1.96) 3.10 (2.81- 3.43) Guh et al. (2009)
Stroke* 1.23 (1.13-1.34) 1.15 (1.0-1.32) 1.51 (1.33-1.72) 1.49 (1.27-1.74) Guh et al. (2009)
Hypertension* 1.28 (1.10-1.50) 1.65 (1.24-2.19) 1.84 (1.51-2.24) 2.42 (1.59-3.67) Guh et al. (2009)
DVT ** 1.70 (1.55-1.87) 1.70 (1.55-1.87) 2.44 (2.15-2.78) 2.44 (2.15-2.78) Pomp et al. (2007)
Type II Diabetes* 2.40 (2.12-2.72) 3.92 (3.1-4.97) 6.74 (5.55-8.19) 12.41 (9.03-17.06)
Guh et al. (2009)
Gallbladder Disease*
1.09 (0.87-1.37) 1.44 (1.05-1.98) 1.43 (1.04-1.96) 2.32 (1.17-4.57) Guh et al. (2009)
Asthma* 1.20 (1.08-1.33) 1.25 (1.05-1.49) 1.43 (1.14-1.79) 1.78 (1.36-2.32) Guh et al. (2009)
Gout*** 1.87 (1.29-2.69) 1.67 (1.03-2.72) 3.50 (2.16-5.72) 3.52 (2.16-5.72) Bhole et al. (2010)
Liver Cancer UKHF has RR UKHF has RR UKHF has RR UKHF has RR
Ovarian Cancer* 1.18 (1.12-1.23) 1.28 (1.20-1.36) Guh et al. (2009)
Prostate Cancer* 1.14 (1.00–1.31) 1.05 (0.85–1.30) Guh et al. (2009)
Urothelial Cancer 1.40 (1.31–1.49) 1.82 (1.68–1.98) 1.82 (1.61–2.05) 2.64 (2.39–2.90) UKHF advise using same RR as kidney cancer RR (above)
Thyroid Cancer ○ 1.19 (1.05-1.36) 1.09 (1.00-1.20) 1.50 (1.29-1.75) 1.19 (1.07-1.34) Zhao et al. (2012) http://imr.sagepub.com/content/40/6/2041.full.pdf+html
MS (option 1) ○○ NR BMI 25-27: 1.44 (0.87-2.39) BMI 27-29:
NR 2.25 (1.50–3.37) Munger et al. (2009) http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2777074/
193
1.40 (0.92–2.14)
MS (option 2) ᴓ BMI 25-27: OR 1.2 (0.8-1.9) BMI 27-29: OR 2.4 (1.4-4.3)
BMI 25-27: OR 1.5 (1.0-2.1) BMI 27-29: OR 2.0 (1.4-3.0)
OR 2.1 (1.0-4.3) OR 2.2 (1.5-3.2) Hedström et al. (2012) http://www.ncbi.nlm.nih.gov/pubmed/22328681
Psoriasis (option
1) ᴓ ᴓ
NR 1.21 (1.03, 1.43) NR 1.63 (1.33, 2.00) Kumar et al. (2013) http://onlinelibrary.wiley.com/doi/10.1111/jdv.12001/full
Psoriasis (option
2) ᴓ ᴓ ᴓ
1.17 (1.03-1.33) 1.17 (1.03-1.33) 1.33 (1.08-1.65) 1.33 (1.08-1.65) Lønnberg et al. (2016) http://www.ncbi.nlm.nih.gov/pubmed/27120802
Pulmonary Embolus/ (Pulmonary Embolism)*
1.91 (1.39–2.64) 1.91 (1.39–2.64) 3.51 (2.61–4.73) 3.51 (2.61–4.73) Guh et al. (2009)
Note: PCOS, NAFLD and hypertension in pregnancy likely to be dropped from modelling due to
insufficient disease data
* Meta-analysis of studies. Study-specific unadjusted relative risks were pooled
** Combined OR, adjusted for age and sex
*** Adjusted for age (continuous)
○ Meta-analysis of studies. Adjusted RR reported but unclear what confounders were adjusted for (may be the
study-specific adjustments)
○○ Age (in months), latitude age 15 (north, middle, south), ethnicity (SEuropean, Scandinavian, other Caucasian,
other), smoking (never smoker, 1–9, 10–24, and ≥25pack-years). Note: there are also age-adjusted figures or age-
and smoking-adjusted figures available in this paper.
ᴓ Adjusted for age, residential area (according to study design), ancestry and smoking
ᴓ ᴓ Age, alcohol consumption, smoking status and physical activity. Note: age-adjusted RR also available in paper.
ᴓ ᴓ ᴓ Multivariable adjustment for smoking and adjusts indirectly for sex, age, and childhood environment due to
matching of the twins.
8.4.3 Disease incidence/prevalence Either annual prevalence or annual incidence can be used because the modelling software includes a
modification of the DISMOD software that develops consistent set of prevalence, incidence and
mortality from the data provided.
Forecasts of disease incidence/prevalence are not required; they are based on BMI forecasts and
(assumed fixed) obesity-related incidence/prevalence rates.
8.4.4 Morality
Annual age-sex specific annual mortality rates are required for each obesity-related disease as well
as all other causes (combined)
Forecasts of mortality rates are not required; they are based on BMI forecasts and (assumed fixed)
obesity-related mortality rates.
8.5 Direct healthcare costs Annual total healthcare costs (or per case costs) for each obesity-related disease are required.
Indirect healthcare indirect costs that are incurred by individuals and their patients, are omitted
from the modelling because of the lack of data and research (ref: Australian study). Some aspects of
these indirect costs may be explored in non-modelling projects (see Appendix 1).
A mix of top-down and bottom-up methods will be used in each JANPA WP4 country. The method
will vary with the impact and cost but will use local data or proxy data from a range of sources:
• Hospital: in-patient and out-patient
• Primary care
• Drugs and prescribing
• Ancillary services – e.g. dietician services, etc.
8.6 Societal impacts and costs The societal costs (not associated with the health care system) included in the modelling are guided
by the evidence and consultations with expert social scientists and JANPA WP4 countries
8.6.1 Childhood
Inadequate data is available to include any such impact in the modelling. Rather, aspects may be
explored in possible non-modelling projects (see Appendix 1).
8.6.2 Adulthood
Two societal costs incurred during adulthood are included in the modelling: adult productivity losses
and lifetime income losses. Adult productivity losses commence when a person develops an obesity-
related disease while lifetime income losses commence with person who is obese at age 18 years.
For the calculations of ladult productivity losses (ifetime income loss), annual figures (annual
earnings) will be discounted if premature death occurs as follows:
By 0% if death occurs before national retirement age.
By 70% if death occurs before national retirement age plus 10 years
By 100% if death occurs thereafter
National total costs (or per case disease costs) for adult productivity losses attributable to a given
disease are required.
8.7 Data collation
8.7.1 Data collation workflow
The models require large amounts of research and data from many domains, broken down by sex
and age. The challenges of collating this data are magnified when modelling occurs across seven
195
European countries with an eye to generalisability to the rest of EU. In many research and data
domains, the necessary research or data will be missing in particular JANPA WP4 countries.
Significant data imputation and other methods for dealing with such missing data will be required
(particularly in basic studies).
Figure 8.2: Data collation workflow in JANPA WP4 countries
8.7.2 Use of proxy data
In any JANPA WP4 country, data imputation may be necessary because:
Required data are not available and it is necessary to imput data from proxy countries
Some required data are available but not at the optimal level of detail and it is necessary to
collapse BMI, age, etc. categories.
When the required research and data, from any domain, is unavailable in a JANPA WP4 country, the
most appropriate proxy data will be used. Decisions about data imputation will be made by IPH IRL
in consultation with JANPA WP4 countries and the UKHF (for example, in the Republic of Ireland, it
was agreed that Health Survey for England (2003 – 2014) data be used for children in the 0-4 years
age category). All decisions will be carefully documented.
On other occasions other approached will be adopted to deal with missing data. For example; data
for particular age category may be based on data from a different set of age categories in particular
JANPA WP4 countries.
8.7.3 Top-down and Bottom-up approaches
A mixture of two approaches will be used to calculate impact-related and cost-related model inputs
and outputs:
• Top-down methods that are used to estimate impact-related and cost-related model inputs
and outputs that are based on the application of Population Attributable Fractions (PAFs) to
national disease and healthcare data
• Bottom-up methods that are used to estimate impact-related and cost-related model inputs
and outputs that are based on analysis of disease and healthcare data in cross-sectional
studies or longitudinal studies that also include BMI data
The approach taken in any JANPA WP4 country will depend on the availability of local research or
data, the impact-related and cost-related model inputs and output involved, and whether or not it is
JANPA WP4 in a basic or advanced study.
Ideally, all model inputs and outputs for a JANPA WP4 country would be calculated using a bottom-up approach based to local research and/or data. Failing this, a top-down method applied to local research and/or data is the next preferred approach. The least preferred approach is the use of a top-down method with international inputs. These methods use different amounts of data imputation.
Age-sex specific relative risks (including the risk amongst the HW category) and other morbidity data
will be used to calculate Population Attributable Fractions (PAF) for each obesity-related disease.
These PAFs are then applied to national prevalence data for each obesity-related disease to obtain
Obesity-related national health impact (disease) = PAF (disease) x National prevalence (disease)
Then;
Total obesity-related health impact = sum (over all diseases) of Obesity-related health impact
(disease)
Of course, the PAFs calculated above (by age and sex) can also be applied to cost data.
8.7.4 Data cleaning
In previous modelling projects, UKHF has used a 4-step data cleaning protocol.
In order to manage the budget, IPH IRL will collate data and construct the Model Inputs Table in the agreed format, conduct Stage 1 data cleaning and produce the AIPM files required by UKHF. IPH will check the distributions of the diseases to ensure the data are within expected mean totals based on GBD, EU CVD statistics, etc.
UKHF will then create Stage 2 input files (.dis files) and clean and upload these files into the modelling software.
197
8.8 Impact-cost indicators, excess metrics and effect metrics
8.8.1 Impact-cost indicators
Table 8.5: Full list of impact-cost indicators
Impact-related
Number of new cases of a disease
Number of deaths:
All deaths
Premature deaths
Potential Years of Life Lost (PYLL)
Quality Adjusted Life Years (QALY)
Disability Adjusted Life Years (DALY)
Adult productivity loss65
Cost - related
Direct healthcare costs
Cost of adult productivity losses
Lifetime income losses
8.8.2 Excess metrics
Use excess metrics to express the primary model outputs. Excess metrics are differences between the values of an impact/cost indicator feature amongst individuals who were overweight or obese as children and its value amongst individuals who were of healthy weight as children:
Excess = indicator(individuals who were OW/OB as children) - indicator(individuals who were HW as children)
Sometimes excess metrics are expressed as totals such as the total number of obesity-related
deaths. After division by the number of individuals who were obese or overweight as children, these
excess metrics can also be expressed on a per basis and be interpreted as per impacts or costs
avoided.
65 In terms of the total number of years not working due to premature death plus absenteeism (estimated by dividing costs by average incomes)
To calculate these excess metrics, individuals have to be categorised according to their childhood BMI status; virtual individuals who are underweight are included in the healthy weight group in the simulations and model outputs.
8.8.3 Effect metrics
Corresponding to each excess metric in Table 8.5 there is an effect metric. Effect metrics describe the effect of a (1% or 5%) reduction in childhood obesity on its corresponding excess metric. The value of an excess metric in the current scenario serves as the base case for the assessment of the effect of a reduction in childhood obesity.
Effect metrics are differences between the value of the excess metric in one of the reduction scenarios and its value in the current scenario:
Effect metric = Excess metric (current childhood obesity) - Excess metric (reduced childhood obesity)
A positive effect represents an improvement (i.e. a reduction in the excess associated with current childhood obesity).
An example of an effect metric is the effect of a 1% reduction in current childhood obesity rates on
the excess annual number of diabetes cases that are associated with childhood obesity. Another
would be the effect that a 5% reduction in current on the lifetime excess direct healthcare costs that
are associated with childhood obesity.
In addition to effect metrics for each of the excess metrics in Table 8.1, a further effect metric base
on the BMI distribution will be calculated.
199
Tables
Table T8.1. Summary of other European data sources on child and adolescent overweight and
obesity
Name of study Year(s) conducted
Age(s) of participants
Countries/Regions Sample Measured or reported BMI
References
EU Identification and prevention of Dietary- and lifestyle-induced health Effects in Children and infants (IDEFICS) study
2006-2012 2 to 9 years Belgium, Cyprus, Estonia, Germany, Hungary, Italy, Spain, Sweden
About 2,000 per country, not nationally representative
Measured Ahrens et al., 2006, 2011, 2014
ToyBox study 1998-2011 4 to 7 years Belgium, Bulgaria, Germany, Greece, Poland, Spain
Pooling of various national surveys
Measured
Manios et al., 2012; van Stralen et al., 2012
ENERGY-project: European Energy balance Research to prevent excessive weight Gain among Youth
2010 10 to 12
years
Belgium, Greece, Hungary, the Netherlands, Norway, Slovenia, Spain
About 1,000 per country, not nationally representative
Measured Brug et al., 2010, 2012
Pro Children study 2003 11 years Austria, Belgium, Denmark, Iceland, the Netherlands, Norway, Portugal, Spain, Sweden
About 1,000 per country, nationally representative
Parent-reported
Klepp et al., 2005; Yngve et al., 2008
Healthy Lifestyle in Europe by Nutrition in Adolescence (HELENA) study
2006-2007 13 to 16
years
Athens (Greece); Dortmund (Germany); Ghent (Belgium); Heraklion (Greece); Lille (France); Pécs (Hungary); Rome (Italy); Stockholm (Sweden); Vienna (Austria); and Zaragoza (Spain)
About 200 adolescents from each city, not nationally representative
Measured
Vicente-Rodriguez et al., 2007; Moreno et al., 2008; Martinez-Gomez et al., 2010
WHO Health Behaviour of School-aged Children (HBSC) study
Every four years from 1982, with the most recent published international results for 2013-2014
11, 13 and 15 years
Albania, Armenia, Austria, Belgium (Flemish), Belgium (French), Bulgaria, Canada, Croatia, Czech Republic, Denmark, England, Estonia, Finland, France, FYR Macedonia, Germany, Greece, Greenland, Hungary, Iceland, Ireland, Israel, Italy, Latvia, Lithuania, Luxembourg, FYR Macedonia, Malta, Netherlands, Norway, Poland, Portugal, Rep of Moldova, Romania, Russian Federation, Scotland, Slovakia, Slovenia, Spain, Sweden, Switzerland, Turkey, Ukraine, USA, Wales
Nationally representative school-based sample
Self-reported
Currie et al., 2004, 2008, 2012; Inchley et al., 2016
JANPA WP4 countries are highlighted
201
CHAPTER 9: MODELLING
9.1 Modelling software
9.1.1 Existing modelling software
The UK Health Forum (UKHF) has extensive experience in simulation modelling and has been
involved in a number of international projects that are relevant to EU JANPA WP4 including the WHO
(Europe) 53 country burden of disease study – “WHO Modelling Obesity Project” – and the EConDA
project. UKHF has been sub-contracted to carry out the modelling for EU JANPA WP4.
Figure 9.1 below illustrates the logic of UKHF’s existing modelling software.
Figure 9.1. UKHF’s existing modelling software
The UKHF’s existing modelling software implements population simulation models of chronic
diseases that have been chiefly used to estimate and forecast the population-level impact and cost
of prevalent obesity and overweight n studies such as Foresight Obesity Study, WHO (Europe) 53
country study and the EConDA project.
This presentation is part of the Joint Action JANPA (Grant agreement n°677063) which has received funding from the European Union’s Health Programme (2014-2020)
Example3:ForesightObesityModel
The list of adult chronic diseases in the UKHF’s existing modelling software is presented in Table 8.1 below. Table 9.1. Adult chronic diseases included in UKHF’s existing modelling software
Disease name
Coronary Heart Disease (CHD) Oral cancer Stroke Dementia Diabetes (type 2) Cervical cancer Colorectal cancer Laryngeal cancer Breast cancer Pancreatitis Kidney cancer Bladder cancer Oesophageal cancer Road injuries Endometrial cancer Violence Gall bladder cancer Poisoning Arthritis Depression Hypertension AML Liver cancer CML Pancreatic cancer Liver cirrhosis Lung cancer Gastric cancer COPD Ovarian cancer CKD
9.1.2 Adaptation of UKHF’s modelling software
UKHF has undertaken significant studies that have either:
Described the level of some impact-cost indicator in the total population (currently or in
some future year); these features can include the prevalence or incidence of some disease
Modelled the long-term effect of public health interventions on the values of such impact-
cost indicators. These are based on the differences between the values of indicators in an
intervention group and the values in some comparison group.
To use the UKHF’s existing modelling software to implement a cohort simulation model that
estimates lifetime impacts and costs and incorporates children and societal impacts; substantial
adaptations are first necessary.
These adaptations involve both the modelling algorithms as well as methods of extracting and
presenting model outputs. The adaptations will be necessary to accommodate:
• Use of cohort simulation models rather than population simulation models66
66
Theoretically; we could forecast the total future population impacts and costs for all obesity by applying our age-sex-BMI specific estimates for the surviving children of 2016 to the population forecasts and obesity projections in a future year. But you would not want to do this too far into the future since there wouldn’t be enough surviving children in later year to do this sort of scaling accurately.
203
• Incorporation of children with shorter term impacts into the models
• Use of a societal economic perspective rather than an exclusively health services perspective
• Use of more complex metrics and reporting associated with lifetime costing studies
These adaptations were incorporated into the modelling sub-contract with the UKHF and, to manage
budget, IPH IRL will undertake a number of the routine data collation and reporting tasks including
the calculation of the model metrics and production of graphical outputs.
9.2 Summary of modelling steps
The modelling steps follow the flow of UKHF’s existing modelling software.
Table 9.2. Main steps in the EU JANPA WP4 modelling
Step Output Use
1 Initialising the virtual cohort Sample of children whose sex, age and BMI distributions match the population of children living in 2016
Units of the initial virtual cohort subsequent simulations
2.a Forecasting population BMI distributions
Forecasts of the future population age-sex BMI distributions
Used to constrain the lifetime BMI trajectories of virtual individuals
2b Simulating lifetime BMI trajectories Lifetime BMI trajectories for each member of the initial virtual cohort
Used as the basis for simulating, for each member of the initial virtual cohort, lifetime experience of obesity-related disease, disability, death and societal impacts
3a Simulating health impacts For each member of the initial virtual cohort, lifetime experience of obesity-related disease, disability and death
Basis for the calculation of the health impact indicators
3b Estimating direct healthcare costs For each member of the initial virtual cohort, lifetime experience of direct healthcare costs incurred over the lifetime.
Basis for the calculation of the (healthcare-related) cost related indicators
3c Simulating societal impacts For each member of the initial virtual cohort, lifetime experience of obesity-related societal impacts
Basis for the calculation of the (societal-related) impact excess metrics
3d Estimating societal costs For each member of the initial virtual cohort, lifetime experience of societal costs incurred across the lifetime
Basis for the calculation of the (societal-related) cost indicators
9.3 Step 1: Initialising the virtual cohort
Adults will not be included in the initial virtual cohort.
The UKHF’s existing modelling software takes the initial virtual cohort to be the actual respondents
to the most recent population surveys to be. With this approach, the size of the initial virtual cohort
is different in each country and depends on their childhood health monitoring systems. 67 In this
work package, the initial virtual cohort in each country will comprise between 20 million and 100
million virtual individuals. Some test runs will be necessary to check the run errors to determine the
formal sample size.
In JANPA WP4, the childhood characteristics (age, sex, BMI and disease status) of the most recently
available data will be projected to 2016, so that all JANPA WP4 countries have the same start-year68.
We then stochastically generate an initial virtual cohort of children that have (age, sex, BMI, disease)
distributions that match these projected ones.
A country’s initial virtual cohort will be representative (in terms of their sex, age and BMI
distributions) of all children living in the country in 2016.
At initialization, virtual individuals are assigned an individual BMI value rather than a BMI category.
Age-sex BMI distributions of initial cohort will match those of the current (2016) childhood
population.
When reducing mean population BMI by 1% and 5%, underweight children will be excluded so their
BMI will not be reduced.
Obesity-related (childhood) diseases will be assigned to virtual individuals in year 1 (2016) so that
the age-sex specific prevalence rates of each disease in the initial cohort match those in the current
childhood population. In the start-year, the obesity-related disease processes are assumed to act
independently so that a virtual individual’s initial disease status is independently assigned for each
disease.
9.4 Steps 2a and 2b: Forecasting BMI distributions and simulating lifetime
BMI trajectories
We will use the following procedure to model the lifetime BMI trajectory of virtual individuals:
67 This approach originated from UKHF’s work in projecting short-term effects observed in clinical trials into the future for use in cost effectiveness analyses. 68 Sometimes several population surveys are used to piece together the age ranges of the country’s child
population
205
Firstly, a regression model (with SEX, AGE and CALENDAR YEAR as independent variables) is
fitted to a country’s historical BMI data. This fitted model is then used to produce forecasts
of the country’s population age-sex specific BMI distributions in future years.
In the start-year of the modelling (2016), each virtual child is assigned a BMI value so that, in
total, the initial virtual cohort has the same age-sex specific BMI distributions as those of the
childhood population living in the country in 2016.
As a virtual individual ages with their age-sex peer group, that peer group’s BMI distribution
in any future year is constrained to be the relevant forecasted population age-sex specific
BMI distribution in that year.
The “constant lifetime BMI-percentile” assumption means that a virtual individual’s BMI
percentile in their age-sex peer group remains the same as they age. This percentile is
applied to the relevant forecasted age-sex population BMI distribution to determine their
BMI value and BMI category in a future year.
The lifetime BMI trajectory of every virtual individual is determined by their initial BMI value and the
forecasted population age-sex specific BMI distributions.
9.5 Steps 3a – 3d: Simulating impacts and estimating costs
In all models, virtual individuals are aged one year at a time. Simulations will continue until the last individual in the virtual cohort has died.
Throughout the modelling, there will be no new entries (births and immigration) and individuals will only be lost by death (and not by emigration). This is a characteristic of cohort simulation models.
To calculate excess metrics associated with childhood obesity each virtual individual will be tagged
with their BMI category at age 17 years as they leave childhood and enter adulthood.
9.5.1 Step 3a: Simulating health impacts
This section explains how the occurrence of a single disease is simulated.
The trends in the prevalence of obesity-related disease and deaths are based on the modelled
lifetime trajectories for the virtual cohort members and the risks of obesity-related diseases and
death.
9.5.1.1 Obesity-related diseases
For each obesity-related disease, a 4-state semi-Markov process is used to determine a virtual
individual’s vital/disease status in a future year depending on their status in the previous year. The
four possible states for an obesity-related disease D are:
0. Alive without disease D 1. Alive with disease D 2. Dead from disease D 3. Dead form some other cause
States 2 and 3 are “absorbing” states.
The annual state transition probabilities depend only on an individual’s state and their BMI category
at the beginning of the year.
The UKHF’s existing modelling software assumes that
Individuals can die from other causes
Healthy weight individuals can develop obesity-related diseases and incur healthcare costs
Individuals cannot develop diseases that are not obesity-related (“No diseases not related to
BMI” assumption)
9.5.1.2 Deaths
Obesity-related deaths in any future year in a particular age-sex category are simulated
probabilistically using survival rates that are modelled as exponential distributions (check EConDA
page 14).
9.5.2 Step 3b: Estimating direct healthcare costs
In any future year, the UKHF’s modelling software “scales the total annual disease costs (in the
virtual cohort) by the relative disease prevalence (relative to the start-year)” (EConDA
documentation (page 16)).
Essentially, for each disease, this involves:
Calculating the total annual cost of the disease in the modelling’s start-year
Dividing this figure by the number of prevalent cases in the start-year to obtain the unit cost in the start-year
Apply this unit cost to the number of cases in a future69
The calculations will incorporate appropriate discounting to present-day values.
9.5.3 Step 3c: Simulating societal impacts
9.5.4 Step 3d: Estimating societal costs
9.6 Producing Model Output Tables
The virtual individuals’ simulated BMI, impact and cost of trajectories will be summarised into tables
of model outputs.
69In population simulation models, a final step would involve scaling these up to the entire total population in each simulation year
207
Annual model outputs will be accumulated into five-year reporting periods for presentation in the
model output tables. Five-year reporting periods will provide model outputs for five-year periods
starting with 2016-2019, 2020-2024, 2025-2029, etc . In addition outputs will be provided for the
periods 2016-2019 and subsequent decades to 2050 (as required in the EU JANPA contract).
UKHF will provide tables of the model outputs - the numerators, denominators, etc. that are needed
to calculate the excess metrics and effect metrics (see Table 4a).
• Sex
• Age categories
• BMI categories
• Disease (where appropriate)
• Five year reporting periods
Forecasting population sizes, obesity rates, etc until the last surviving virtual individual dies is
problematic. The sensitivity analysis will examine the last few-reporting periods and assess which, if
any, should be deleted in the WP4 reports (see Chapter 7).
CHAPTER 10: REPORTING
The adaptions to UKHF existing modelling software were incorporated into the modelling sub-
contract with the UKHF and, to manage budget, IPH IRL will undertake a number of the routine data
collection and reporting tasks (including the calculation of model metrics and production of graphical
outputs).
10.1 Reporting work flow
Figure 10.1. Reporting work flow
10.2 Calculating excess metrics , effect metrics and producing graphical
outputs
Using the Model Output Tables produced by UKHF, IPH IRL will calculate all excess metrics and effect
metrics and produce graphical outputs to be reported in the country-specific reports. This will be
done in Excel spreadsheet templates which will be included in the toolbox of modelling resources
that can be used in other EU countries.
Calculation of the excess metrics involves the calculation of the differences in the values of
the impact-cost indicators amongst virtual individuals who were overweight or obese as
children and the values amongst virtual individuals who were of healthy weight as children.
Calculation of the effect metrics involves the calculation of the differences between the
value of excess metrics in current situation and the values in a childhood obesity reduction
scenario.
Modelling by UKHF 1. Output Tables from
UKHF
2. Calculation of excess metrics by IPH IRL
3. Prodeuction of graphical outputs by IPH
IRLe
4. Production of country reports by IPH IRL and participating countries
5. Production of Briefing for EU Ministers by
particpating countries and IPH-IRL
209
CHAPTER 11: ASSESSING VALIDITY AND GENERALISABILITY
IPH IRL will lead validation studies and UKHF will undertake the necessary modelling tasks.
11.1 Validity
11.1.1 Comparison of model-based and research-based relative risks
UKHF will include a selection of adult diseases (from Table 11.1) and adult societal impacts (from
Table 11.2) in the breakdowns of the Model Output Tables. IPH will then calculate occurrence rates
amongst virtual individuals who were obese or overweight in childhood and amongst those who
were of healthy weight. The derived RRs will then be compared to the RRs identified in the
international literature.
Adult diseases
Table 11.1. Global list of adult diseases that are significantly related to childhood obesity and
overweight (McCarthey et al, 2016b)
Condition Effect Estimate Study Type
Adult Obesity Age 7-11: RR 4.86; 95% CI 4.29, 5.51 Age 12-18: RR 5.45; 95% CI 4.34-6.85
Meta-analysis
Type 2 Diabetes Age 6 and under: Pooled OR per SD of BMI 1.23 (1.10-1.37) Age 7-11: Pooled OR per SD of BMI 1.78 (1.51-2.10) Age 12-18: Pooled OR per SD of BMI 1.70 (1.30-2.22)
Meta-analysis
Hypertension Age 7-11: Pooled OR per SD of BMI 1.67 (0.89-3.13) Age 12-18: Pooled OR per SD of BMI 1.29 (1.19-1.40)
Meta-analysis
Coronary Heart Disease Age 6 and under: Pooled OR per SD of BMI 0.97 (0.85-1.10) Age 7-11: Pooled OR per SD of BMI 1.14 (1.08-1.21) Age 12-18: Pooled OR per SD of BMI 1.30 (1.16-1.47)
Meta-analysis
Ischaemic heart disease Age 2-22: Pooled hazard ratio for IHD per 1 SD of BMI 1.09 (95% CI 1.01, 1.10), adjusting for family social class.
Cohort (n=14,561)
Stroke Age 6 and under: Pooled OR per SD of BMI 0.94 (0.75-1.19) Age 7-11: Pooled OR per SD of BMI 1.02 (0.94-1.10) Age 12-18: Pooled OR per SD of BMI 1.06 (1.04-1.09)
High quality meta-analysis (Llewellyn et al., 2015) but association only positive for adolescents
Non-alcoholic fatty liver disease (NAFLD) 70
Age 7-13: Males: HR 1.15 (95% CI 1.05 to 1.26) per 1-unit gain in BMI z-score between ages 7 and 13 years, adjusting for BMI z-score at age 7 years
Cohort (n=244,464)
70 Need to estimate what proportion of liver disease/cirrhosis is non-alcoholic (attributable to NAFLD). This will be based on a diagnosis of cryptogenic cirrhosis which, at a conservative estimate, accounts for 60% of NAFLD.
Females: HR 1.12 (95% CI 1.02 to 1.23) per 1-unit gain in BMI z-score after adjusting for BMI z-score between ages 7 and 13 years, adjusting for BMI z-score at age 7 years
Gout Age 13-18 years: Males Overweight (≥75th percentile) (unadjusted RR 3.1; 95% CI 1.1, 9.3) compared to those who were not overweight (25th and 50th percentile). Overall (males and females): unadjusted RR 2.7 (95% CI 0.9, 7.7)
Third Harvard Growth Study (n=342) – small sample size
Polycystic Ovarian Syndrome
Age 14: Overweight: RR 1.12 (95% CI 0.87–1.43), adjusting for social class Obese: RR 1.61 (95% CI 1.24-2.08), adjusting for social class
Study of weak-moderate quality due to self-reported BMI at age 14 (n=1836)
Hypertension in pregnancy71
Age 7: Overweight: UOR 1.46, p<0.05. Obesity: UOR 2.14, p<0.05.
1958 British birth cohort (n=5799)
Breast cancer Age 6 and under: Pooled OR per SD of BMI 0.88 (0.67-1.16) Age 7-11: Pooled OR per SD of BMI 0.90 (0.77-1.05) Age 12-18: Pooled OR per SD of BMI 0.92 (0.82-1.03)
High quality meta-analysis (Llewellyn et al., 2015) but no association
All cancer Age 7-11: OR per SD of BMI 1.14 (1.00-1.29) High study quality (Boyd Orr cohort) but association is weak (n=2,374)
Renal Cell Carcinoma (males only)
Age 12-18: OR per SD of BMI 1.19 (1.04-1.37) Cohort (n=1,110,835)
Ovarian Cancer (females only)
Age 12-18: OR per SD of BMI 1.22 (1.01-1.49) Cohort (n=111,883)
Endometrial cancer (females only)
Age 7: HR 1.53 (95% CI: 1.29-1.82), i.e. Compared with a 7-year-old girl with a BMI z-score=0, an equally tall girl who was 3.6kg heavier (BMI z-score=1.5) had a HR=1.53 (95% CI: 1.29-1.82).
Cohort (n=155,505)
Urothelial Cancer (males only)
Age 12-18: OR per SD of BMI 1.21 (1.06-1.38) Cohort (n=1,110,835)
Colon Cancer Age 12-18: OR per SD of BMI 1.21 (1.07-1.38) Cohort (n=1,109,864)
Colon Cancer Death Age 12-18: OR per SD of BMI 1.48 (1.05-2.11) Cohort (n=226,682)
Heptocellular Carcinoma (males only)
Age 7-11: OR per SD of BMI 1.31 (1.12-1.53) Age 12-18: OR per SD of BMI 1.36 (1.17-1.60)
Cohort (n=285,884)
Liver Cancer (males only)
Age 7-11: OR per SD of BMI 1.27 (1.11-1.45) Age 12-18: OR per SD of BMI 1.30 (1.14-1.48)
Cohort (n=285,884)
Liver Cancer (females only)
Age 7-11: OR per SD of BMI 1.20 (0.97-1.49) Age 12-18: OR per SD of BMI 1.32 (1.07-1.64)
Cohort (n=285,884)
Thyroid cancer Age 7-13: HR for a 1-unit increase in BMI (approx. 1.5–2 kg/m2) was 1.15 (95% CI, 1.00–1.34).
Cohort (n=321,085)
Prostate cancer (males only)
Age 7: unadjusted HR 1.06; 95% CI 1.01, 1.10 per BMI z-score Age 13: unadjusted HR 1.05; 95% CI 1.01, 1.10 per BMI z-score
Cohort (n=125,208)
71 Little direct cost data. There may be some data available from obstetricians.
211
Cardiovascular mortality Age 17:
Obesity (≥95th percentile): death from total CV causes (HR 3.5; 95% CI 2.9, 4.1) compared to those in the 5th-24th percentiles Obesity (≥95th percentile): death from CHD (HR 4.9; 95% CI 3.9, 6.1) compared to those in the 5th-24th percentiles Obesity (≥95th percentile): death from stroke (HR 2.6; 95% CI 1.7 to 4.1) compared to those in the 5th-24th percentiles Obesity (≥95th percentile): sudden death (HR 2.1; 95% CI 1.5, 2.9) compared to those in the 5th-24th percentiles Adjusting for sex, age, birth year, sociodemographic characteristics, and height.
Cohort (n=2,298,130)
All-cause mortality Age 14-19: BMI ≥85th centile: males (RR 1.4; 95% CI 1.3, 1.6) and females (RR 1.4; 95% CI 1.2, 1.5) compared to those in the 25-74th centile, adjusting for age at measurement and year of birth
Cohort (n=226,958)
Disability pension (males) 72
Age 18: Overweight (BMI 25–25.9; HR 1.36, 95% CI 1.32-1.40), Moderate (BMI 30–34.9; HR 1.87, 95% CI 1.76-1.99), Morbid obesity (BMI ≥35; 3.04, 95% CI 2.72-3.40) compared to normal weight
Cohort (n=1,048,103)
Sick Leave Age 18: Overweight was associated with: Sick leave ranging from 8 to 30 days (HR 1.20; 95% CI 1.15–1.24) Long-term sick leave >30 days (HR 1.19; 95% CI 1.15–1.23). Obesity was associated with: Sick leave ranging from 8 to 30 days (HR 1.35; 95% CI 1.24–1.47) Long-term sick leave >30 days (HR 1.34; 95% CI 1.24–1.47) Adjusting for smoking, socio-economic index and muscular strength.
Cohort (n=43,989)
Lifetime productivity losses - risk of never having been gainfully employed
Age 10: Males: Persistent obesity from childhood to adulthood: AOR 1.4 (95% CI 0.9 to 2.3) in the multivariable model. Females: Persistent obesity from childhood to adulthood: AOR 1.9 (95% CI 1.1 to 3.3) in the multivariable model. Adjusting for: maternal education, social class in childhood and adulthood, maternal and paternal BMI, and height at 10 and 30 years.
Cohort (n=8490)
72 Need estimates of knee surgery that is likely due to osteoarthritis
Education Status (Years of schooling
Age 17-18 years: Males: Overweight (0.2 year less years of schooling; 95% CI 0.5 to 0.0, p=0.08). Females: Overweight (0.3 year less years of schooling; 95% CI 0.1 to 0.6, p=0.009) compared to non-overweight. Adjusting for: base-line characteristics
Cohort (n=7931)
Adult societal impacts
Table 11.2. Global list of adult societal impacts significantly related to childhood obesity and
overweight (McCarthey et al, 2016b)
Disability pension (males)
Age 18: Overweight (BMI 25–25.9; HR 1.36, 95% CI 1.32-1.40), Moderate (BMI 30–34.9; HR 1.87, 95% CI 1.76-1.99), Morbid obesity (BMI ≥35; 3.04, 95% CI 2.72-3.40) compared to normal weight
Cohort (n=1,048,103)
Sick Leave Age 18: Overweight was associated with: Sick leave ranging from 8 to 30 days (HR 1.20; 95% CI 1.15–1.24) Long-term sick leave >30 days (HR 1.19; 95% CI 1.15–1.23). Obesity was associated with: Sick leave ranging from 8 to 30 days (HR 1.35; 95% CI 1.24–1.47) Long-term sick leave >30 days (HR 1.34; 95% CI 1.24–1.47) Adjusting for smoking, socio-economic index and muscular strength.
Cohort (n=43,989)
Lifetime productivity losses - risk of never having been gainfully employed
Age 10: Males: Persistent obesity from childhood to adulthood: AOR 1.4 (95% CI 0.9 to 2.3) in the multivariable model. Females: Persistent obesity from childhood to adulthood: AOR 1.9 (95% CI 1.1 to 3.3) in the multivariable model. Adjusting for: maternal education, social class in childhood and adulthood, maternal and paternal BMI, and height at 10 and 30 years.
Cohort (n=8490)
Education Status (Years of schooling
Age 17-18 years: Males: Overweight (0.2 year less years of schooling; 95% CI 0.5 to 0.0, p=0.08). Females: Overweight (0.3 year less years of schooling; 95% CI 0.1 to 0.6, p=0.009) compared to non-overweight. Adjusting for: base-line characteristics
Cohort (n=7931)
213
11.1.2. Methods of modelling lifetime BMI trajectories
IPH IRL will undertake a review of the different methods of modelling lifetime BMI trajectories and
summarise the advantages and disadvantages of each. These can include:
More advanced statistical methods (e.g. PAC analyses, exponential models)
Estimation of transition probabilities between BMI categories and Markov processes to
model lifetime BMI trajectories
• Use of latent growth curve analyses to identify latent BMI trajectories from longitudinal
studies and sample to derive lifetime BMI trajectories
Probably the most practical alternative would be to model the age-sex specific BMI transition
probabilities from a country’s historical cross-sectional BMI data.
UKHF will rerun the models in Ireland and Slovenia using the lifetime BMI trajectories file
constructed by IPH IRL instead of the file constructed by the method in the modelling software.
In addition, we will compare forecasts of population BMI distributions based on the modelled
lifetime BMI trajectories and prevalence estimates and forecasts from other sources such as WHO
(Europe) country profiles, COSI studies, HBSC surveys and longitudinal studies.
11.1.3 The independent disease processes assumption
UKHF will produce a one page think piece on the effect of the “independent disease processes”
assumption on model outputs. For example; will it result in under-estimates or over-estimates? Does
the size of the effect depend on the prevalence of each disease and on multi-morbidities in the
reference studies? Is the effect the same in all population subgroups?
11.1.4. Sensitivity analysis
As well as standard sensitivity analyses, IPH IRL will look at:
DATA IMPUTATION. In countries participating in an advanced study, where possible we will compare effects of different imputation required with the three possible methods of calculating model inputs and model outputs: bottom-up methods using local inputs, top-down methods using local inputs and top-down methods using international inputs
LATER REPORTING PERIODS. We will examine the last few reporting periods and assess which, if any, should be deleted in the WP4 reports.
11.2 Generalisability
IPH will assess the generalisability to rest of EU and UKHF will undertake modelling tasks for the
following studies.
11.2.1 EConDA online tool
11.2.2 Modelling resources
If generalisable, the resources developed during the work package can be used to apply JANPA WP4
modelling methodology in other EU countries. These resources include the following questionnaires
Local Material Survey to identify local reports and research to supplement the international
evidence-base
Data Sources Survey to identify local and international data sources not captured by initial
data scoping exercise
Global list of child diseases significantly related to childhood overweight and obesity
Global list of adult diseases significantly related to childhood overweight and obesity
Global list of adult diseases significantly related to adult overweight and obesity
Template to filter global lists in each JANPA WP4 country
Final Data Proposal template
Data Request (to collate the agreed data from the country)
Model Inputs Table
Model Output Table
Metrics Spreadsheet to calculate excess and effect metrics and produce graphical outputs
from Model Outputs Table
215
REFERENCES AND
APPENDICIES
217
REFERENCES
UK NICE Report: “Managing overweight and obesity among children. Report on Economic Modelling
and Cost Consequence Analysis” which looked at effects of childhood weight management
programmes.
Simmonds M, Llewellyn A, Owen CG, Woolacott N. Predicting adult obesity from childhood obesity: a
systematic review and meta-analysis. Obesity Reviews. 2016;17(2):95-107.
Aarestrup, J., Gamborg, M., Cook, M.B., Sorensen, T.I.A., & Baker, J.L. (2014). “Childhood body mass index and the risk of prostate cancer in adult men.” British Journal of Cancer 111: 207-212.
Aarestrup, J., Gamborg, M., Ulrich, J.G., Tia, S., & Baker, J.L. (2016). “Childhood body mass index and height and risk of histologic subtypes of endometrial cancer.” International Journal of Obesity doi: 10.1038/ijo.2016.56.
Aasvee, K., M. Rasmussen, Kelly, C., Kurvinen, E., Giacchi, M. V., & Ahluwalia, N. (2015). "Validity of self-reported height and weight for estimating prevalence of overweight among Estonian adolescents: the Health Behaviour in School-aged Children study." BMC Research Notes 8: 606.
Aberle, N., Blekic, M., Ivanis, A., & Pavlovic, I. (2009). “The Comparison of Anthropometrical Parameters of the Four-Year-Old Children in the Urban and Rural Slavonia, Croatia, 1985 and 2005.” Collegium Antropologicum 2: 347-351.
Adami, F., & Vasconcelos, F.A. (2008). “Childhood and adolescent obesity and adult mortality: a systematic review of cohort studies.” Cadernos de Saude Publica 24 Suppl 4:s558-568.
Aeberli, I., Gut-Knabenhans, I., Kusche-Ammann, R.S., Molinari, L., & Zimmermann, M.B. (2011). "Waist circumference and waist-to-height ratio percentiles in a nationally representative sample of 6-13 year old children in Switzerland." Swiss Medical Weekly 141: w13227.
Ahrens, W., Bammann, K., De Henauw, S. Halford, J., Pigeot, I., Siani, A. & Sjostrom, M. (2006). "Understanding and preventing childhood obesity and related disorders-IDEFICS: a European multilevel epidemiological approach." Nutrition, Metabolicm & Cardiovascular Diseases 16(4): 302-308.
Ahrens, W., Bammann, K., Siani, A., Buchecker, K., De Henauw, S., Iacoviello, L., Hebestreit, A., Krogh, V., Lissner, L., Marild, S., Molnar, D., Moreno, L.A., Pitsiladis, Y.P., Reisch, L., Tornaritis, M., Veidebaum, T. Pigeot, I., & IDEFICS Consortium (2011). "The IDEFICS cohort: design, characteristics and participation in the baseline survey." International Journal of Obesity 35 Suppl 1: S3-15.
Ahrens, W., Pigeot, I., Pohlabeln, H., De Henauw, S., Lissner, L., Molnar, D., Moreno, L.A., Tornaritis, M., Veidebaum, T., Siani, A., & IDEFICS Consortium (2014). "Prevalence of overweight and obesity in European children below the age of 10." International Journal of Obesity 38 Suppl 2: S99-107.
Alberti, K.G., Zimmet, P., & Shaw, J. (2006). "Metabolic syndrome—a new world‐wide definition. A consensus statement from the international diabetes federation." Diabetic Medicine 23(5): 469-480.
Amis, J.M., Hussey, A., & Okunade, A.A. (2014). “Adolescent obesity, educational attainment and adult earnings.” Applied Economics Letters 21(13): 945-950.
Anamaria, B., Chibelean, M., Pacurar, M., Esian, D., Monica-Cristina, M., & Bud, E. (2015). “Correlation between BMI, dental caries and salivary buffer capacity in a sample of children from Mures County, Romania.” European Scientific Journal 11: 38-46.
Anderson, E. L., Howe, L.D., Jones, H.E., Higgins, J.P.T., Lawlor, D.A., & Fraser, A. (2015). "The Prevalence of Non-Alcoholic Fatty Liver Disease in Children and Adolescents: A Systematic Review and Meta-Analysis." PLoS One 10(10): e0140908.
Angelopoulos, P. D., Milionis, H.J., Moschonis, G., & Manios, T. (2006). "Relations between obesity and hypertension: preliminary data from a cross-sectional study in primary schoolchildren: the children
study." European Journal of Clinical Nutrition 60(10): 1226-1234.
Angelopoulos, P.D., Milionis, H.J., Grammatikaki, E., Moschonis, G., & Manios, Y. (2009). “Changes in BMI and blood pressure after a school based intervention: The CHILDREN study.” European Journal of Public Health 19(3): 319-325.
Antonogeorgos, G., Papadimitriou, A., Panagiotakos, D.B., Priftis, K.N., & Nicolaidou, P. (2010). Association of extracurricular sports participation with obesity in Greek children. Journal of Sports Medicine and Physical Fitness 51(1): 121-127.
Antony, B., Jones, G., Venn, .A, Cicuttini, F., March, L., Blizzard, L., Dwyer, T., Cross, M., & Ding, C. (2015). “Association between childhood overweight measures and adulthood knee pain, stiffness and dysfunction: a 25-year cohort study.” Annals of the Rheumatic Diseases 74(4): 711-717.
Ardeleanu, I.S., Nanu, M., Moldovanu, F., Bacalearos, C., & Moculescu, C. (2015). “OVERWEIGHT AND OBESITY SCREENING OF 6-7 YEARS OLD AND 13-14 YEARSOLD CHILDREN IN 14 COUNTIES FROM ROMANIA”. Paper presented at IOMC days, December, Bucharest.
Astolfi, R., Lorenzoni, L., & Oderkirk, J. (2012). “A comparative analysis of health forecasting models.” OECD Health Working Papers 59. Paris: OECD.
Bacopoulou, F., Efthymiou, V., Landis, G., Rentoumis, A., & Chrousos, G.P. (2015). "Waist circumference, waist-to-hip ratio and waist-to-height ratio reference percentiles for abdominal obesity among Greek adolescents." BMC Pediatrics 15: 50.
Barba, G., Troiano, E., Russo, P., Strazzullo, P., Siani, A., on behalf of the ARCA Project study group (2006). “Body mass, fat distribution and blood pressure in Southern Italian children: Results of the ARCA project.” Nutrition, Metabolism & Cardiovascular Diseases 16: 239-248.
Barbu, C.G., Teleman, M.D., Albu, A.I., Sirbu, A.E., Martin, S.M., Bancescu, A., & Fica, S.V. (2015). “Obesity and eating behaviours in school children and adolescents – data from a cross sectional study in Bucharest, Romania.” BMC Public Health 15: 206 DOI 10.1186/s12889-015-1569-9
Barron, C., Comiskey, C., & Saris, J. (2009). "Prevalence rates and comparisons of obesity levels in Ireland." British Journal of Nursing 18(13): 799.
Bhole V, de Vera M, Rahman MM, Krishnan E, Choi H. Epidemiology of gout in women: Fifty-two–year followup
of a prospective cohort. Arthritis & Rheumatism. 2010;62(4):1069-76.
Bibiloni del Mar, M., Pons, A., & Tur, J.A. (2013). "Prevalence of overweight and obesity in adolescents: a
systematic review." International Scholarly Research Notes: Obesity 2013: 392747.
Bierl, M., Marsh, T., Webber, L., Brown, M., McPherson, K., & Rtveladze, K. (2013). “Apples and oranges: A comparison of costing methods for obesity.” Obesity Reviews 14(9): 693-706.
Binkin, N., Fontana, G., Lamberti, A., Cattaneo, C., Baglio, G., Perra, A., & Spinelli, A. (2008). “A national survey of the prevalence of childhood overweight and obesity in Italy.” Obesity Reviews 11: 2-10.
Bodzsar, E. B. & Zsakai, A. (2014). "Recent trends in childhood obesity and overweight in the transition countries of Eastern and Central Europe." Annals of Human Biology 41(3): 263-270.
Booth, J.N., Tomporowski, P.D., Boyle, J.M.E., Ness, A.R., Joinson, C., Leary, S.D., & Reilly, J.J. (2014). “Obesity impairs academic attainment in adolescence: findings from ALSPAC, a UK cohort.” International Journal of Obesity 38: 1335-1342.
Bozanic, A., Beic, J., & Mumanovic, D. (2011). “OVERWEIGHT AND OBESITY AS LIMITATION FACTORS OF AGILITY AND STRENGTH DEVELOPMENT.” Proceedings of the 6
th International Scientific Conference on
Kinesiology Zagreb, September, pp. 207-210.
Bralic, I., Tahirovic, H. & Matanic, D. (2011). "Growth and obesity in 7-year-old Croatian children: secular changes from 1991 to 2008." European Journal of Pediatrics 170(12): 1521-1527.
Bralic, I., Tahirovic, H., Matanic, D., Vrdoljak, O., Stojanovic-Spehar, S., Kovacic, V., & Blazekovic-Milakovic, S. (2012). “Association of early menarche age and overweight/obesity.” Journal of Pediatric Endocrinology and Metabolism 25(1-2): 57-62.
Bralic, I., Vrdoljak, J., & Kovacic, V. (2005). “Associations between parental and child overweight and obesity.” Collegium Antropologicum 29: 481-486.
Branca, F., Nikogosian, H., & Lobstein, T. (2007). The challenges of obesity in the WHO European Region and the strategies for response: Summary. Copenhagen, Denmark, World Health Organization.
Brannsether, B., Roelants, M., Bjerknes, R., & Juliusson, P.B. (2011). "Waist circumference and waist-to-height ratio in Norwegian children 4-18 years of age: reference values and cut-off levels." Acta Paediatrica 100(12): 1576-1582.
219
Brufani, C., Ciampalini, P., Grossi, A., Fiori, R., Fintini, D., Tozzi, A., Cappa, M., & Barbetti F. (2010). “Glucose tolerance status in 510 children and adolescents attending an obesity clinic in Central Italy.” Pediatric Diabetes 11: 47–54.
Brug, J., Te Velde, S.J., Chinapaw, M.J., Bere, E., De Bourdeaudhuij, I., Moore, H., Maes, L., Jensen, J., Manios, Y., Lien, N., Klepp, K.I., Lobstein, T., Martens, M., Salmon, J., & Singh, A.S. (2010). "Evidence-based development of school-based and family-involved prevention of overweight across Europe: The ENERGY-project's design and conceptual framework." BMC Public Health 10.
Brug, J., M. van Stralen, M., Te Velde, S.J., Chinapaw, M.J., De Bourdeaudhuij, I., Lien, N., Bere, E., Maskini, V., Singh, A.S., Maes, L. Moreno, L., Jan, N., Kovacs, E., Lobstein, T., & Manios, Y. (2012). "Differences in weight status and energy-balance related behaviors among schoolchildren across Europe: the ENERGY-project." PLoS One 7(4): e34742.
Brumariu, O., Haliţchi, C., Munteanu, M., Baltag, A., & Aursulesei, V. (2007). “A preliminary clinical study of the metabolic syndrome in children.” Revista Medico-chirurgicala a Societatii de Medici si Naturalisti din Iasi 111(1): 44-48.
Bruno, G., Maule, M., Merletti, F., Novelli, G., Falorni, A., Ianilli, A., Iughetti, L., Altobelli, E., d’Annunzio, G., Piffer, S., Pozzilli, P., Iafusco, D., Songini, M., Roncarolo, F., Toni, S., Carle, F., Cherubini, V., & the RIDI Study Group (2010). “Age-Period-Cohort Analysis of 1990–2003 Incidence Time Trends of Childhood Diabetes in Italy: The RIDI Study.” Diabetes 59: 2281-2287.
Butland, B., Jebb, S., Kopelman, P., McPherson, K., Thomas, S., Mardell, J., & Parry, V. (2007). Foresight: Tackling future choices – project report. London: Department of Innovation, Universities and Skills.
Caird, J., Kavanagh, J., O’Mara-Eves, A., Oliver, K., Oliver, S., Stansfield, C., & Thomas, J. (2014). “Does being overweight impede academic attainment? A systematic review.” Health Education Journal 73(5): 497-521.
Calcaterra, V., Klersy, C., Muratori, T., Telli, S., Caramagna, C., Scaglia, F., Cisternino, M., & Larizza, D. (2008). “Prevalence of metabolic syndrome (MS) in children and adolescents with varying degrees of obesity.” Clinical Endocrinology 68: 868-872.
Capizzi, M., Leto, G., Petroni, A., Zampetti, S., Papa, R.E., Osimani, M., Spoletini, M., Lenzi, A., Osborn, J., Mastantuono, M., Vania, A., & Buzzetti, R. (2011). “Wrist Circumference Is a Clinical Marker of Insulin Resistance in Overweight and Obese Children and Adolescents.” Circulation 128: 1757-1762.
Carolan, E., Hogan, A.E., Corrigan, M., Gaotswe, G., O'Connell, J., Foley, N., O'Neill, L.A., Cody, D., & O'Shea, D. (2013). "The impact of childhood obesity on inflammation, innate immune cell frequency, and metabolic microRNA expression." The Journal of Clinical Endocrinology & Metabolism 99(3): E474-E478.
Casariu, E.D., Virgolici, B., Greabu, M., Totan, A., Daniela, M., Mitrea, N., Ion, A., & Mohora, M. (2011). “Associations between carotid intima media thickness and cardiovascular risk markers in obese children.” Farmacia 59(4): 471-482.
Caserta, C.A., Pendino, G.M., Alicante, S., Amante, A., Amato, F., Fiorillo, M., Mesisneo, A., Polito, I., Surace, M., Surace, P., Vacalebre, C., Zuin, M., Cotichini, R., Marcucci, F., Rosmini, F., & Mele, A. (2010). “Body mass index, cardiovascular risk factors, and carotid intima-media thickness in a pediatric populationin Southern Italy.” Journal of Pediatric Gastroenterology & Nutrition 51(2): 216-220.
Cassimos, D., Sidiropoulos, H., Batzios, S., Balodima, V., & Christoforidis, A. (2011). “Sociodemographic and Dietary Risk Factors for Excess Weight in a Greek Pediatric Population Living in Kavala, Northern Greece.” Nutrition in Clinical Practice 26(2): 186-191.
Cattaneo, A., Monasta, L., Stamatakis, E., Lioret, S., Castetbon, K., Frenken, F., Manios, Y., Moschonis, G., Savva, S., Zaborskis, A., Rito, A.I., Nanu, M., Vignerova, J., Caroli, M., Ludvigsson, J., Koch, F.S., Serra-Majem, L., Szponar, L., van Lenthe, F., & Brug, J. (2010). "Overweight and obesity in infants and pre-school children in the European Union: a review of existing data." Obesity Reviews 11(5): 389-398.
Cavallo, F., Giacchi, M., Vieno, A., Galeone, D., Tomba, A., Lamberti, A., Nardone, P., & Andreozzi, S. (2013). Studio HBSC-Italia (Health Behaviour in School-aged Children): rapporto sui dati 2010. Roma: Istituto Superiore di Sanità.
Cepeda-Lopez, A. C., Aeberli, I., & Zimmermann, M. B. (2010). Does obesity increase risk for iron deficiency? A review of the literature and the potential mechanisms. International Journal for Vitamin and Nutrition Research, 80(4): 263.
Celi, F., Bini, V., de Giorgi, G., Moninari, D., Faraoni, F., di Stefano, G., Bacosi, M.L., Berioli, M.G., Contessa, G, & Falorni, A. (2003). “Epidemiology of overweight and obesity among school children and adolescents in
three provinces of central Italy, 1993–2001: study of potential influencing variables.” European Journal of Clinical Nutrition 57: 1045-1051.
Chalkias, C., Papadopoulos, A.G., Kalogeropoulos, K., Tambalis, K., Psarra, G., & Sidossis, L. (2013). "Geographical heterogeneity of the relationship between childhood obesity and socio-environmental status: Empirical evidence from Athens, Greece." Applied Geography 37: 34-43.
Chesaru, B.I., Dobre, M., Murariu, G., & Nechita, A. (2013). “Risk for metabolic syndrome in a group of overweight children from South-East Romania.” Revista Român de Medicin de Laborator 21: 83-92.
Chirita-Emandi, A., Puiu, M., Gafecnu, M., & Pienar, C. (2013). “Arterial hypertension in school-aged children in Western Romania.” Cardiology in the Young 23: 189-196.
Chirita-Emandi, A., Puiu, M., Gafecnu, M., & Pienar, C. (2012). “GROWTH REFERENCES FOR SCHOOL AGED CHILDREN IN WESTERN ROMANIA.” Acta Endocrinologica (Buc) VIII: 133-152.
Coe, D. P., Ode, J.J., Pfeiffer, K.A., & Pivarnik, J.M. (2010). "Accuracy of body mass index to determine overweight in youth." International Journal of Body Composition Research 8(4): 147-154.
Cole, T. J., Bellizzi, M.C., Flegal, K., & Dietz, W.H. (2000). "Establishing a standard definition for child overweight and obesity worldwide: international survey." BMJ 320: 1-6.
Cole, T. J., Flegal, K.M., Nicholls, D., & Jackson, A.A. (2007). "Body mass index cut offs to define thinness in children and adolescents: international survey." BMJ 335(7612): 194.
Collaborative Group on Hormonal Factors in Breast Cancer. (2012). “Menarche, menopause, and breast cancer risk: individual participant meta-analysis, including 118 964 women with breast cancer from 117 epidemiological studies.” The Lancet Oncology, 13(11): 1141–1151.
Cook, S., Auinger, P., & Huang, T.T. (2009). "Growth curves for cardio-metabolic risk factors in children and adolescents." Journal of Pediatrics 155(3): S6 e15-26.
Corbo, G.M., Forastiere, F., de Sario, M., Brunetti, L., Bonci, E., Bugiani, M., Chellini, E., la Grutta, S., Migliore, E., Pistelli, R., Rusconi, F, Russo, A., Simoni, M., Talassi, F., Galassi, C., & the Sidria-2 Collaborative Group (2008). “Wheeze and Asthma in Children: Associations With Body Mass Index, Sports, Television Viewing, and Diet.” Epidemiology 19(5): 747-755.
Cosoveanu, C.S. (2011). Primary obesity in children: Etiopathogenic, clinical and prophylactic aspects. Unpublished Doctoral Thesis, Faculty of Medicine, University of Medicine and Pharmacy, Craiova.
Currie, C., Nic Gabhainn, S., Godeau, E., Roberts, C., Smith, R., Currie, D., Pickett, W., Richter, M., Morgan, A., & Barnekow, V. (2008). Inequalities in young people's health: HBSC international report from the 2005/2006 survey. Copenhagen, Denmark, World Health Organization.
Currie, C., Roberts, C., Morgan, A., Smith, R., Settertobulte, W., Samdal, O., & Barnejow Rasmussen, V. (2004). Young people's health in context: Health Behaviour in School-Aged Children (HBSC) Study: International report from the 2001/2002 survey. Copenhagen, Denmark.
Currie, C., Zanotti, C., Morgan, A., Currie, D., de Looze, M., Roberts, C., Samdal, O., Smith, O.R.F., & Barnekow, V. (2012). Social determinants of health and well-being among young people. Health Behaviour in School-aged Children (HBSC) study: International report from the 2009/2010 survey. Copenhagen, Denmark, World Health Organization.
Daniels, S.R., Khoury, P.R., & Morrison, J.A. (2000). "Utility of Different Measures of Body Fat Distribution in Children and Adolescents." American Journal of Epidemiology 152(12): 1179-1184.
de Moraes, A.C., Fadoni, R.P., Ricardi, L.M., Souza, T.C., Rosaneli, C.F., Nakashima, A.T., & Falcao, M.C. (2011). "Prevalence of abdominal obesity in adolescents: a systematic review." Obesity Reviews 12(2): 69-77.
de Onis, M. (2007). "Development of a WHO growth reference for school-aged children and adolescents." Bulletin of the World Health Organization 85(09): 660-667.
de Onis, M., Blossner, M., & Borghi, E. (2010). "Global prevalence and trends of overweight and obesity among preschool children." American Journal of Clinical Nutrition 92(5): 1257-1264.
de Onis, M. & Lobstein, T. (2010). "Defining obesity risk status in the general childhood population: which cut-offs should we use?" International Journal of Pediatric Obesity 5(6): 458-460.
de Onis, M., Onyango, A., Borghi, E., Siyam, A., Blossner, M., Lutter, C. & WHO MGRS Group (2012). "Worldwide implementation of the WHO Child Growth Standards." Public Health Nutrition 15(9): 1603-1610.
del Mar Biblioni, M., Pons, A., & Tur, J.A. (2013). “Prevalence of overweight and obesity in adolescents: A systematic review.” International Scholarly Research Notes: Obesity Article ID 392747, doi: 10.1155/2013/392747.
Dee, A., Kearns, K., O’Neill, C., Sharp, L., Staines, A., O’Dwyer, V., Fitzgerald, S., & Perry, I.J. (2014). The direct
221
and indirect costs of both overweight and obesity: a systematic review. BMC Research Notes 7: 242.
Delas, N., Tudor, A., Ruzic, L., & Sestan, B. (2008). “Obesity indicators and athletic performance in 11-15 year-old children.” Hrvatski Športskomedicinski Vjesnik 23(1): 35-44.
Demerath, E.W., Schubert, C.M., Maynard, L.M., Sun, S.S., Chumlea, W.C., Pickoff, A., Czerwinski, S.A., Towne, B., & Siervogel, R.M. (2006). "Do changes in body mass index percentile reflect changes in body composition in children? Data from the Fels Longitudinal Study." Pediatrics 117(3): e487-495.
Di Bonito, P., Valerio, G., Grugni, G., Licenziati, M.R., Maffeis, C., Manco, M., Miraglia del Giudice, E., Pacifico, L., Pellegrin, M.C., Tomat, M., & Baroni, M.G. for the CARITALY Study Group (2015). “Comparison of non-HDL-cholesterol versus triglycerides-to-HDL-cholesterol ratio in relation to cardiometabolic risk factors and preclinical organ damage in overweight/obese children: The CARITALY Study.” Nutrition, Metabolism and Cardiovascular Diseases 25: 489-494.
Due, P., Damsgaard, M.T., Rasmussen, M., Holstein, B.E., Wardle, J., Merlo, J., Currie, C., Ahluwalia, N., Sorensen, T.I., Lynch, J., Borraccino, A., Borup, I., Boyce, W., Elgar, F., Gabhainn, S.N., Krolner, R., Svastisalee, C., Matos, M.C., Nansel, T., Al Sabbah, H., Vereecken, C., & Valimaa, R. (2009). "Socioeconomic position, macroeconomic environment and overweight among adolescents in 35 countries." International Journal of Obesity 33(10): 1084-1093.
Dumbrava, L., Popa, A., & Brink, S. (2012). Risk factors for prediabetes in overweight and obese pre-teens and adolescents. Romanian Journal of Diabetes Nutrition & Metabolic Diseases 19(3): 255-263.
Ekblom, O., Oddsson, K., & Ekblom, B. (2004). "Prevalence and regional differences in overweight in 001 and trends in BMI distribution in Swedish children from 1987 to 2001." Scandinavian Journal of Public Health 32(4): 257-263.
Ellert, U., Brettschneider, A,K., Wiegand, S., & Kurth, B.M. (2014). "Applying a correction procedure to the prevalence estimates of overweight and obesity in the German part of the HBSC study." BMC Research Notes 7: 181.
Evans, D. S., Glacken, M., & Goggin, D. (2011). "Childhood obesity: the extent of the problem among 6-year-old Irish national school children." Child: Care, Health & Development 37(3): 352-359.
Fahey, T., Delaney, L., & Gannon, B. (2005). School children and sport in Ireland. Dublin: ESRI.
Farajian, P., Risvas, G., Karasouli, K., Pounis, G.D., Kastorini, C.M., Panagiotakos, D.B., & Zampelas, A. (2011). "Very high childhood obesity prevalence and low adherence rates to the Mediterranean diet in Greek children: the GRECO study." Atherosclerosis 217(2): 525-530.
Farkas, D., Tomak, Z., Petric, D., & Novac, D. (2015). “Anthropometric characteristics and obesity indicators among preschool children in an urban area in Croatia.” Graduate Journal of Sport, Exercide & Physical Education Research 3: 13-27.
Fernandes, M.M. (2010). Evaluating the Impacts of School Nutrition and Physical Activity Policies on Child Health. Pittsburgh: RAND Corporation.
Ferrao, M.M., Gama, A., Marques, V.R., Mendes, L.L., Mourao, I., Nogueira, H., Velasquez-Melendez, G., & Padez, C. (2013). “Association between parental perceptions of residential neighbourhood environments and childhood obesity in Porto, Portugal.” European Journal of Public Health 23: 1027-1031.
Ferreira, R.J., & Marques-Vidal, P.M. (2008). “Prevalence and determinants of obesity in children in public schools of Sintra, Portugal.” Obesity 16(2): 497-500.
Ferreira Felgueiras, M. (2011). “Obesidade na adolescência: a repercussão no auto-conceito e o fenómeno da violência escolar.” Unpublished Doctoral Dissertation, Universidade Católica Portuguesa.
Finkelstein, E. A., Graham, W. C. K., & Malhotra, R. (2014). “Lifetime Direct Medical Costs of Childhood Obesity.” Pediatrics, 133(5): 854-862.
Finkelstein, E.A., Trogdon, J.G., Brown, D.S., Allaire, B.T., Dellea, P.S., & Sachin, J.K.B. (2008). “The Lifetime Medical Cost Burden of Overweight and Obesity: Implications for Obesity Prevention.” Obesity 16(8): 1843-1848.
Finucane, F., Pittock, S., Fallon, M., Hatunic, M., Ong, K., Burns, N., Costigan, C., Murphy, N., & Nolan, J. (2008a). "Elevated blood pressure in overweight and obese Irish children." Irish Journal of Medical Science 177(4): 379-381.
Finucane, F., Teong, L., Pittock, S., Fallon, M., Hatunic, M., Costigan, C., Murphy, N., Crowley, V., & Nolan, J. (2008b). "Adverse metabolic profiles in a cohort of obese Irish children." Annals of Clinical Biochemistry 45(2): 206-209.
Fredriks, A.M., van Buuren, S., Fekkes, M., Verloove-Vanhorick, S.P., & Wit, J.M. (2005). "Are age references for
waist circumference, hip circumference and waist-hip ratio in Dutch children useful in clinical practice?" European Journal of Pediatrics 164(4): 216-222.
Freedman, D.S., Kit, B.K., & Ford, E.S. (2015). "Are the Recent Secular Increases in Waist Circumference among Children and Adolescents Independent of Changes in BMI?" PLoS One 10(10): e0141056.
Freedman, D. S., Serdula, M.K., Scrinivasan, S.R. & Berenson, G.S. (1999). "Relation of circumferences and skinfold thicknesses to lipid and insulin concentrations in children and adolescents: The Bogalusa Heart Study." American Journal of Clinical Nutrition 69.
Friedemann, C., Heneghan, C., Mahtani, K., Thompson, M., Perera, R., & Ward, A.M. (2012). "Cardiovascular disease risk in healthy children and its association with body mass index: systematic review and meta-analysis." BMJ 345: e4759.
Galcheva, S.V., Iotova, V.M., Yotov, Y.T., Grozdeva, K.P., Stratev, V.K., & Tzaneva, V.I. (2009). "Waist circumference percentile curves for Bulgarian children and adolescents aged 6-18 years." International Journal of Pediatric Obesity 4(4): 381-388.
Genovesi, S., Giussani, M., Federico, P., Vigorita, F., Arcovio, C., Cavuto, S., & Stella, A. (2005). “Results of blood pressure screening in a population of school-aged children in the province of Milan: role of overweight.” Journal of Hypertension 23(3): 493-497.
Gherlan, I., Vladoiu, S., Alexiu, F., Giurcaneanu, M., Oros, S., Brehar, A., Procopiuc, C., & Constantin, D. (2012). “Adipocytokine Profile and Insulin Resistance in Childhood Obesity.” Maedica 7(3): 109-117.
Gonzalez-Casanova, I., Sarmiento, O.L., Gazmararian, J.A., Cunningham, S.A., Martorell, R., Pratt, M., & Stein, A.D. (2013). "Comparing three body mass index classification systems to assess overweight and obesity in children and adolescents." Revista Panamericana de Salud Pública 33(5): 349-355.
Gorber, S., Tremblay, M., Moher, D., & Gorber, B. (2007). "A comparison of direct vs. self-report measures for assessing height, weight and body mass index: a systematic review." Obesity Reviews 8(4): 307-326.
Gortmaker, S.L., Must, A., Perrin, J.M., Sobol, A.M., & Dietz, W.H. (1993). “Social and Economic Consequences of Overweight in Adolescence and Young Adulthood.” New England Journal of Medicine 329(14): 1008-1012.
Grammatikopoulou, M. G., Poulimeneas, D., Gounitsioti, I.S., Gerothanasi, K., Tsigga, M., Kiranas, E., & ADONUT Study Group (2014). "Prevalence of simple and abdominal obesity in Greek adolescents: the ADONUT study." Clinical Obesity 4(6): 303-308.
Griffiths, L.J., Parsons, T.J., & Hill, A.J. (2010). "Self-esteem and quality of life in obese children and adolescents: a systematic review." International Journal of Pediatric Obesity 5(4): 282-304.
Grummer-Strawn, L. M., Reinold, C., & Krebs, N.F. (2010). Use of World Health Organization and CDC Growth Charts for Children Aged 0–59 Months in the United States. Morbidity and Mortality Weekly Report, 59, RR-9. Atlanta, USA, Centers for Disease Control and Prevention.
Guh DP, Zhang W, Bansback N, Amarsi Z, Birmingham CL, Anis AH. The incidence of co-morbidities related to
obesity and overweight: A systematic review and meta-analysis. BMC public health. 2009;9(1):1-20.
Haas, G. M., Liepold, E., & Schwandt, P. (2011). "Percentile curves for fat patterning in German adolescents."
World Journal of Pediatrics 7(1): 16-23.
Hakim, Z., Wolf, A., & Garrison, L.P. (2002). “Estimating the effect of changes in body mass index on health state preferences.” Pharmacoeconomics 20: 393-404.
Hamilton, D., Dee, A & Perry, I.J. (in preparation). Lifetime costs of childhood overweight and obesity. Report commissioned by safefood Ireland. Cork: safefood.
Hassapidou, M., Fotiadou, E., Maglara, E.,& Papadopoulou, S.K. (2006). Energy intake, diet composition, energy expenditure, and body fatness of adolescents in northern Greece. Obesity 14(5): 855-862.
Hassapidou, M., Papadopoulou, S.K., Frossinis, A., Kaklamanos, I., & Tzotzas, T. (2009). "Sociodemographic, ethnic and dietary factors associated with childhood obesity in Thessaloniki, Northern Greece." Hormones 8(1): 53-59.
Hatzis, C. M., Papandreou, C., Vardavas, C.I., Athanasopoulos, D., Balomenaki, E., & Kafatos, A.G. (2012). "Atherogenic risk factors among preschool children in Crete, Greece." Indian Journal of Endocrinology & Metabolism 16(5): 809-814.
Hayden, C., Bowler, J.O., Chambers, S., Freeman, R., Humphris, G., Richards, D., & Cecil, J.E. (2013). "Obesity and dental caries in children: a systematic review and meta‐analysis." Community Dentistry & Oral Epidemiology 41(4): 289-308.
223
Hedstrom AK, Olsson T, Alfredsson L. High body mass index before age 20 is associated with increased risk for
multiple sclerosis in both men and women. Multiple sclerosis (Houndmills, Basingstoke, England).
2012;18(9):1334-6
Heinen, M., Murrin, C. Daly, L., O’Brien, J., Heavey, P., Kilroe, J., O’Brien, M., Scully, H., Mulhern, L.M., Lynam,
A., Hayes, C., O’Dwyer, U., Eldin, N., & Kelleher, C. (2014). The Childhood Obesity Surveillance Initiative (COSI)
in the Republic of Ireland: Findings from 2008, 2010 and 2012. Dublin: Health Service Executive and
Department of Health.
Hendrix, C.G., Prins, M.R., & Dekkers, H. (2014). "Developmental coordination disorder and overweight and obesity in children: a systematic review." Obesity Reviews 15(5): 408-423.
Heo M., Faith, M.S., Mott, J.W., Gorman, B.S., Redden, D.T., & Allison, D.B. (2003). “Hierarchical linear models for the development of growth curves: an example with body mass index in overweight/obese adults.” Statistics in Medicine 22: 1911-1942.
Hollingworth, W., Hawkins, J., Lawlor, D.A., Brown, M., Marsh, T., & Kipling, R.R. (2012). “Economic evaluation of lifestyle interventions to treat overweight or obesity in children.” International Journal of Obesity 36: 559-566.
Hooley, M., Skouteris, H., Boganin, C., Satur, J., & Kilpatrick, N. (2012). "Body mass index and dental caries in children and adolescents: a systematic review of literature published 2004 to 2011." Systematic Reviews 1(1): 57.
Horan, M., Gibney, E., Molloy, E., & McAuliffe, F. (2015). "Methodologies to assess paediatric adiposity." Irish Journal of Medical Science 184(1): 53-68.
Ianuzzi, A., Romano, M.L., Licenziati, M.R., Panico, S., Acampora, C., Rubba, P., Salvatore, V., Trevisan, M., & Auriemma, L. (2004). “Increased carotid intima-media thickness and stiffness in obese children.” Diabetes Care 27(10): 2506-2508.
Ille, J., Furic-Cunko, V., Cigrovski, A, Bogdanic, A., Rojnic Putarek, N., Radica, A., & Dumic, M. (2012). “Incidence of glucose and lipid metabolism disorders in overweight children and adolescents in Croatia.” Endocrine Abstracts 29: 1269.
Inchley, J., Currie, D., Young, T., Samdal., O., Torsheim, T., Augustson, L., Mathison, F., Aleman-Diaz, A., Molcho, M., Webber, M., & Barnekow, V. (2016). Growing up unequal: Gender and socioeconomic differences in young people’s health and well-being – Health Behaviour in School-Aged Children (HBSC) Study: International report from the 2013/2014 survey. Copenhagen: WHO Regional Office for Europe.
Invitti, C., Morabito, F., Guzzaloni, G., Veberti, G., & Gilardini, L. (2003). “Prevalence and Concomitants of Glucose Intolerance in European Obese Children and Adolescents.” Diabetes Care 26(1): 118-124.
Invitti, C., Gilardinia, L., Pontiggia, G., Morabito, F., Mazzilli, G., & Viberti, G., (2005). “Period prevalence of abnormal glucose tolerance and cardiovascular risk factors among obese children attending an obesity centre in Italy.” Nutrition, Metabolism & Cardiovascular Diseases 16: 256-262.
Jaworski, M., Kulaga, Z., Pludowski, P., Grajda, A., Gurzkowska, B., Napieralska, E., Swiader, A., Pan, H., Litwin, M., & Olaf Study Group (2012). "Population-based centile curves for triceps, subscapular, and abdominal skinfold thicknesses in Polish children and adolescents-the OLAF study." European Journal of Pediatrics 171(8): 1215-1221.
Jelastopulu, E., Kallianezos, P., Merekoulias, G., Alexopoulos, E.C., & Sapountzi-Krepia, D. (2012). "Prevalence and risk factors of excess weight in school children in West Greece." Nursing & Health Sciences 14(3): 372-380.
Jones-Smith, J. C., Dieckmann, M.G., Gottlieb, L., Chow, J., & Fernald, L.C. (2014). "Socioeconomic status and trajectory of overweight from birth to mid-childhood: the Early Childhood Longitudinal Study-Birth Cohort." PLoS One 9(6): e100181.
Juonala, M., Magnussen, C.G., Berenson, G.S., Venn, A., Burns, T.L., Sabin. M.A., Scrinivasan, S.R., Daniels, S.R., Davis, P.H., Chen, W., Sun, C., Cheung, M., Viikari, J.S.A., Dwyer, T., & Raitakari, O.T. (2011). “Childhood adiposity, adult adiposity, and cardiovascular risk factors.” New England Journal of Medicine 365(20): 1876-1885.
Juresa, V., Musil, V., Majer, M., Ivankovic, D., & Petrovic, D. (2012). “Behavioral Pattern of Overweight and Obese School Children.” Collegium Antropolicum 36 Suppl 1: 139-146.
Katzmarzyk, P. T., Bray, G.A., Greenway, F.L., Johnson, W.D., Newton, Jr., R.L., Ravussin, E., Ryan, D.H., & Bouchard, C. (2011). "Ethnic-specific BMI and waist circumference thresholds." Obesity 19(6): 1272-
1278.
Katzmarzyk, P. T., Scrinivasan, S.R., Chen, W., Malina, R.M., Bouchard, C., & Berenson, G.S. (2004). "Body Mass Index, Waist Circumference, and Clustering of Cardiovascular Disease Risk Factors in a Biracial Sample of Children and Adolescents." Pediatrics 114(2): e198-e205.
Keane, E., Kearney, P.M., Perry, I., Kelleher, C., & Harrington, J. (2014). "Trends and prevalence of overweight and obesity in primary school aged children in the Republic of Ireland from 2002-2012: A systematic review." BMC Public Health 14.
Keane, E., Layte, R., Harrington, J., Kearney, P.M., & Perry, I.J. (2012). "Measured parental weight status and familial socio-economic status correlates with childhood overweight and obesity at age 9." PloS One 7(8): e43503.
Kelishadi, R., Mirmoghtadaee, P., Najafi, H., & Keikha, M. (2015). "Systematic review on the association of abdominal obesity in children and adolescents with cardio-metabolic risk factors." Journal of Research in Medical Sciences: The Official Journal of Isfahan University of Medical Sciences 20(3): 294.
Kendzor, D.E., Caughy, M.O., & Tresch Owen, M. (2012). "Family income trajectory during childhood is associated with adiposity in adolescence: A latent class growth analysis " BMC Public Health 12.
Kitahara, C.M., Gamborg, M., Berrington de Gonzalez, A., Sorensen, T.I.A., & Baker, J.L. (2014a). “Childhood Height and Body Mass Index Were Associated with Risk of Adult Thyroid Cancer in a Large Cohort Study.” Cancer Research 74(1): 235-242.
Kitahara, C.M., Gamborg, M., Rajaraman, P., Sorensen, T.I.A., & Baker, J.L. (2014b). “A Prospective Study of Height and Body Mass Index in Childhood, Birth Weight, and Risk of Adult Glioma Over 40 Years of Follow-up.” American Journal of Epidemiology 180(8): 821-829.
Kleanthous, K., Dermitzaki, E., Papadimitriou, D.T., Papaevangelou, V., & Papadimitriou, A. (2016). "Overweight and obesity decreased in Greek schoolchildren from 2009 to 2012 during the early phase of the economic crisis." Acta Paediatrica 105(2): 200-205.
Klepp, K. I., Perez-Rodrigo, C., De Bourdeaudhuij, I., Due, P.P., Elmadfa, I., Haraldsdottir, J., Konig, J., Sjostrom, M., Thorsdottir, I., Vaz de Almeida, M.D., Yngve, A., & Brug, J. (2005). "Promoting fruit and vegetable consumption among European schoolchildren: rationale, conceptualization and design of the pro children project." Annals of Nutrition & Metabolism 49(4): 212-220.
Knai, C., Lobstein, T., Darmon, N., Rutter, H., & McKee, M. (2012). “Socioeconomic Patterning of Childhood Overweight Status in Europe.” International Journal of Environmental Research & Public Health 9: 1472-1489.
Kolle, E., Steene-Johannessen, J., Holme, I., Andersen, L.B., & Anderssen, S.A. (2009). "Secular trends in adiposity in Norwegian 9-year-olds from 1999-2000 to 2005." BMC Public Health 9: 389.
Kollias, A., Antonodimitrakis, P., Grammatikos, E., Chatziantonakis, N., Grammatikos, E.E., & Stergiou, G.S. (2009). “Trends in high blood pressure in Greek adolescents.” Journal of Human Hyptertension 23: 385-390.
Kollias, A., Skliros, E., Stergiou, G.S., Leotsakos, N., Saridi, M., & Garifallos, D. (2011). “Obesity and associated cardiovascular risk factors among schoolchildren in Greece: a cross-sectional study and review of the literature.” Journal of Pediatric Endocrinology and Metabolism 24: 929-938.
Kollias. A., Psilopatis, I., Karagiaouri, E., Glaraki, M., Grammatikos, E., Grammatikos, E.E., Garoufi, A., & Stergiou, G.S. (2013). “Adiposity, Blood Pressure, and Carotid Intima-Media Thickness in Greek Adolescents.” Obesity 21(5): 1013-1017.
Kontogianni, M. D., Farmaki, A.E., Vidra, N., Sofrona, S., Magkanari, F., & Yannakoulia, M. (2010). "Associations between lifestyle patterns and body mass index in a sample of Greek children and adolescents." Journal of the American Dietetic Association 110(2): 215-221.
Koroni, M., Garagouni-Areou, F., Roussi-Vergou, C.J., Zafiropoulou, M., & Piperakis, S.M. (2009). "The stigmatization of obesity in children. A survey in Greek elementary schools." Appetite 52(1): 241-244.
Kosti, R.I., Panagiotakos, D.B., Mihas, C.C., Alevizos, A., Zampelas, A., Mariolis, A., & Tountas, Y. (2007). “Dietary habits, physical activity and prevalence of overweight/obesity among adolescents in Greece: The Vyronas study.” Medical Science Monitor 13(10): CR437-CR444.
Kotanidou, E.P., Grammatikopoulou, M.G., Spiliotis, B.E., Kanaka-Gantenbein, C., Tsigga, M., & Galli-Tsinopoulou, A. (2013). "Ten-Year obesity and overweight prevalence in Greek children: A systematic review and meta-analysis of 2001-2010 data." Hormones 12(4): 537-549.
Kovac, M., Jurak, G., & Leskosek, B. (2012). "The prevalence of excess weight and obesity in Slovenian children and adolescents from 1991 to 2011." Anthropological Notebooks 18(1): 91-103.
225
Kovac, M., Leskoseck, B., Kragelj, L.Z., & Strel, J. (2014). " THE SECULAR TREND IN THE PREVALENCE OF OVERWEIGHT AND OBESITY IN THE POPULATION OF PRIMARY SCHOOL CHILDREN FROM LJUBLJANA (SLOVENIA)." Zdrav Var 53: 188-198.
Kovac, M., Leskosek, B., & Strel, J. (2008). “Overweight and obesity trends in Slovenian boys from 1991 to 2006.” ACTA UNIVERSITATIS PALACKIANAE OLOMUCENSIS GYMNICA 38(1): 17-26.
Krebs, N. F., Himes, J.H., Jacobson, D., Nicklas, T.A., Guilday, P., & Styne, D. (2007). "Assessment of child and adolescent overweight and obesity." Pediatrics 120 Suppl 4: S193-228.
Kuczmarski, R.J., Ogden, C.L., & Guo, S.S. (2002). 2000 CDC growth charts for the United States: Methods and
development Vital and health statistics. Series 11, Data from the National Health Survey, No. 246.
Washington DC, National Center for Health Statistics.
Kumar S, Han J, Li T, Qureshi AA. Obesity, waist circumference, weight change and the risk of psoriasis in US
women. Journal of the European Academy of Dermatology and Venereology. 2013;27(10):1293-8.
Kunjesic, M., Badric, M., & Prskalo, I. (2015). “RELATIONS BETWEEN OBESITY INDICATORS AND AEROBIC CAPACITY OF PUPILS.” Sport SPA 12(1): 17-24.
Kyriazis, I., Rekleiti, M., Saridi, M., Beliotis, E., Toska, A., Souliotis, K., & Wozniak, G. (2012). "Prevalence of obesity in children aged 6-12 years in Greece: nutritional behaviour and physical activity." Archives of Medical Science 8(5): 859-864.
Labree, L.J.W., van de Mheen, H., Rutten, F.F.H., & Foets, M. (2011). “Differences in overweight and obesity among children from migrant and native origin: a systematic review of the European literature.” Obesity Reviews 12: e535-e547.
Lagiou, A. & Parava, M. (2008). "Correlates of childhood obesity in Athens, Greece." Public Health Nutrition 11(9): 940-945.
Lake, J.K., Power, C., & Cole, T.J. (1997). “Women's reproductive health: the role of body mass index in early
and adult life. International journal of obesity and related metabolic disorders.” Journal of the
International Association for the Study of Obesity. 21(6): 432-438.
Larsson SC, Wolk A. Overweight, obesity and risk of liver cancer: a meta-analysis of cohort studies. British
Journal of Cancer. 2007;97(7):1005-1008. doi:10.1038/sj.bjc.6603932.
Layte, R., & Biesma-Blanco, R (2014). “Social class differences in weight gain from birth to three years.” Paper presented at the Growing Up in Ireland Research Conference 2014: Dublin, November.
Layte, R. & McCrory, C. (2011). Growing Up in Ireland-National Longitudinal Study of Children: Overweight and Obesity Among 9-Year-Olds. Dublin: Department of Children and Youth Affairs.
Lazzeri, G., Rossi, S., Pammolli, A., Pilato, V., Pozzi, T., & GIacchi, M.V. (2008). “Underweight and overweight among children and adolescents in Tuscany (Italy): Prevalence and short-term trends.” Journal of Preventative Medicine & Hygiene 49: 13-21.
Lazzeri, G., Giacchi, M.V., Spinelli, A., Pammoli, A., Dalmasso, P., Nardone, P. Lamberti, A., & Cavallo, F. (2014). “Overweight among students aged 11-15 years and its relationship with breakfast, area of residence, and parents’ education: Results from the Italian HBSC 2010 cross-sectional study.” Nutrition Journal 13: 69.
Lazzeri, G., Panatto, D., Pammolli, A., Azzolini, E., Simi, R., Meoni, V., Giacchi, M.V., Amicizia, D., & Gasparini, G. (2015). "Trends in overweight and obesity prevalence in Tuscan schoolchildren (2002-2012)." Public Health Nutrition 18(17): 3078-3085.
Lei, S. F., Liu, M.Y., Chen, X.D., Deng, F.Y., Lv, J.H., Jian, W.X., Xu, H., Tan, L.J., Yang, Y.J., Wang, Y.B., Xiao, S.M., Sun, X., Jiang, C., Guo, Y.F., Guo, J.J., Li, Y.N., Liu, Y.J., & Deng, H.W. (2006). "Relationship of total body fatness and five anthropometric indices in Chinese aged 20-40 years: different effects of age and gender." European Journal of Clinical Nutrition 60(4): 511-518.
Leite, A., Santos, A., Monteiro, M., Gomes, L., Veloso, M., & Costa, M. (2012). “Impact of overweight and obesity in carotid intima-media thickness of Portuguese adolescents.” Acta Paediatrica 101: e115-e121.
Leskosek, B., Strel, J., & Kovac, M. (2007). “Differences in physical fitness between normal-weight, overweight and obese children and adolescents.” Kinesiologia Slovenica 13(1): 21-30.
Leskosek, B., Strel, J., & Kovac, M. (2010). “Overweight and Obesity in Slovenian Schoolgirls, 1991–2006.” Collegium Antropologicum 34: 1303-1308.
Levy, D.T., Mabry, P.L., Wang, Y.C., Gortmaker, S., Huang, T.T.K., Marsh, T., Moodie, M., & Swinburn, B. (2011).
Simulation models of obesity: A review of the literature and implications for research and policy. Obesity Reviews 12(5): 378-394.
Lien, N., Henriksen, H.B., Nymoen, L.L., Wind, M., & Klepp, K.I. (2010). "Availability of data assessing the prevalence and trends of overweight and obesity among European adolescents." Public Health Nutrition 13(10A): 1680-1687.
Lightwood, J., Bibbins-Domingo, K., Coxson, P., Wang, C., Williams, L., & Goldman, L. (2009). “Forecasting the Future Economic Burden of Current Adolescent Overweight: An Estimate of the Coronary Heart Disease Policy Model.” American Journal of Public Health, 99(12): 2230-2237.
Lissner, L., Visscher, T.L., Rissanen, A., & Heitmann, B.L. (2013). "Monitoring the obesity epidemic into the 21st century--weighing the evidence." Obesity Facts 6(6): 561-565.
Llewellyn, A., Simmonds, M., Owen, C.G., & Wollacott, N. (2016). “Childhood obesity as a predictor of morbidity in adulthood: a systematic review and meta-analysis.” Obesity Reviews 17(1): 56-67.
Lobstein, T. (2015) "Prevalence and trends across the world." ECOG free obesity ebook. http://ebook.ecog-obesity.eu/chapter-epidemiology-prevention-across-europe/prevalence-trends-across-world/
Lobstein, T., Baur, L.A., & Uauy, R. (2004). "Obesity in children and young people: A crisis in public health." Obesity Reviews 5(s1): 4-85.
Lombardo, F.L., Spinelli, A., Lazzeri, G., Lamberti, A., Mazzarella, G., Nardone, P. Pilato, V., Buoncristiano, M., & Caroli, M. for the OKkio alla SALUTE Group 2010 (2014). “Severe obesity prevalence in 8- to 9-year-old Italian children: a large population-based study.” European Journal of Clinical Nutrition 1-6.
Lloyd, L.J., Langley-Evans, S.C., & McMullen, S. (2010). “Childhood obesity and adult cardiovascular disease risk: a systematic review.” International Journal of Obesity 34(1): 18-28.
Lloyd L.J., Langley-Evans, S.C., & McMullen, S. (2012). “Childhood obesity and risk of the adult metabolic syndrome: a systematic review.” International Journal of Obesity 36(1): 1-11.
Lønnberg A, Skov L, Skytthe A, Kyvik K, Pedersen O, Thomsen S. ASsociation of psoriasis with the risk for type 2 diabetes mellitus and obesity. JAMA Dermatology. 2016;152(7):761-7.
Lopes, V.P., Stodden, D.F., Bianchi, M.M., Maia, J.A.R., & Rodrigues, L.P. (2011). “Correlation between BMI and motor coordination in children.” Journal of Science & Medicine in Sport 15: 38-43.
Low, S., Chin, M.C., & Deurenberg-Yap, M. (2009). “Review on epidemic of obesity.” Annals of the Academy of Medicine 37(1): 57-65.
Lundborg, P., Nystedt, P., & Rooth, D.-O. (2014). “Body Size, Skills, and Income: Evidence from 150,000 Teenage Siblings.” Demography, 51(5): 1573-1596.
Luppino, F.S., de Wit, L.M., Bouvy, P.F., Stijnen, T., Cuijpers, P., Penninx, P.W., & Zitman, F.G. (2010). “Overweight, obesity, and depression: a systematic review and meta-analysis of longitudinal studies.” Archives of General Psychiatry 67(3): 220-229.
Lydakis, C., Stefanaki, E., Stefanaki, S., Thalassinos, E., Kavousanaki, M., & Lydaki, D. (2012). "Correlation of blood pressure, obesity, and adherence to the Mediterranean diet with indices of arterial stiffness in children." European Journal of Pediatrics 171(9): 1373-1382.
Macfarlane, G.J., de Silva, V., & Jones, G.T. (2011). “The relationship between body mass index across the life course and knee pain in adulthood: results from the 1958 birth cohort study.” Rheumatology 50(12): 2251-2256.
McCarthy, H. D. (2007). "Body fat measurements in children as predictors for the metabolic syndrome: focus on waist circumference." Proceedings of the Nutrition Society 65(04): 385-392.
McCarthy, H.D., Ellis, S.M., & Cole, T.J. (2003). "Central overweight and obesity in British youth aged 11-16 years: Cross sectional surveys of waist circumference " BMJ 326.
McCarthy, H.D., Jarett, K.D., & Crawley, H.F. (2001). "The development of waist circumference percentiles in British children aged 5.0–16.9y." European Journal of Clinical Nutrition 55: 902-907.
McCarthy, L., Keane, E., Geaney, F., O’Sullivan, M. & Perry, I.J. (2016a). Trends and prevalence of overweight and obesity in primary school aged children in Ireland from 2002-2015: An update on the existing literature. Report commissioned by safefood Ireland. Cork: safefood.
McCarthy, L., Geaney, F., O’Sullivan, M. & Perry, I.J. (2016b). The effect of childhood overweight and obesity on risk of adult overweight and obesity and risk of chronic disease, disability, reduced quality of life and mortality in adult life. Report commissioned by safefood Ireland. Cork: safefood.
Ma, S., & Frick, K. D. (2011). “A Simulation of Affordability and Effectiveness of Childhood Obesity Interventions.” Academic Pediatrics, 11(4): 342-350.
Maffeis, C., Consolaro, A., Cavarzere, P., Chini, L., Banzato, C., Grezzani, C., Silvagni, D., Salzano, D., de Luca, F.,
227
& Tató, L. (2006). “Prevalence of overweight and obesity in 2- to 6-year-old Italian children.” Obesity 14 765-769.
Maffeis, C., Banzato, C., & Talamini, G. (2008). "Waist-to-height ratio, a useful index to identify high metabolic risk in overweight children." Journal of Pediatrics 152(2): 207-213.
Magkos, F., Manios, Y., Christakis, G., & Kafatos, A.G. (2005). "Secular trends in cardiovascular risk factors among school-aged boys from Crete, Greece, 1982-2002." European Journal of Clinical Nutrition 59(1): 1-7.
Malindretos, P., Doumpali, E., Mouselimi, M., Papamichail, N., Doumpali, Ch., Sianaba, O., Orfanaki, G., & Sioulis, A. (2009). "Childhood and parental obesity in the poorest district of Greece." Hippokratia 13(1): 46-49.
Mandic, Z., Piricki, A.P., Kenjeric, D., Hanicar, B., & Tansic, I. (2011). “Breast vs. bottle: differences in the growth of Croatian infants.” Maternal & Child Nutrition 7: 389-396.
Mangan, L., & Zgaga, L. (2014). “Exploring the association between sleep duration and overweight and obesity in infants.” Paper presented at the Growing Up in Ireland Research Conference 2014: Dublin, November.
Manios, Y., Angelopoulos, P.D., Kourlaba, G., Kolotourou, M., Grammatikaki, E., Cook, T.L., Bouloubasi, Z., & Kafatos, A.G. (2011). "Prevalence of obesity and body mass index correlates in a representative sample of Cretan school children." International Journal of Pediatric Obesity 6(2): 135-141.
Manios, Y., Costarelli, V., Kolotourou, M., Kondakis, K., Tzavara, C., & Moschonis, G. (2007). "Prevalence of obesity in preschool Greek children, in relation to parental characteristics and region of residence." BMC Public Health 7: 178.
Manios, Y., Grammatikaki, E., Androutsos, O., Chinapaw, M.J., Gibson, E.L., Buijs, G., Iotova, V., Socha, P., Annemans, L., Wildgruber, A., Mouratidou, T., Yngve, A. Duvinage, K., de Bourdeaudhuij, I., & the ToyBox study group (2012). "A systematic approach for the development of a kindergarten-based intervention for the prevention of obesity in preschool age children: the ToyBox-study." Obesity Reviews 13 Suppl 1: 3-12.
Manios, Y., Moschonis, G., Chrousos, G.P., Lionis, C., Mougios, V., Kantilafti, M., Tzotzola, V., Skenderi, K.P., Petridou, A., Tsalis, G., Sakellaropoulou, A., Skouli, G., & Katsarou, C. (2013). "The double burden of obesity and iron deficiency on children and adolescents in Greece: the Healthy Growth Study." Journal of Human Nutrition & Dietetics 26(5): 470-478.
Manios, Y., Yiannakouris, N., Papoutsakis, C., Moschonis, G., Magkos, F., Skenderi, K., & Zampelas, A. (2004). "Behavioral and physiological indices related to BMI in a cohort of primary schoolchildren in Greece." American Journal of Human Biology 16(6): 639-647.
Marginean, O.D., Simedrea, I.D., Lesovici, M.A., Bucuras, D.M., Marcovici, T.I., & Gug, C.M. (2010). “The Endocrine Disturbances in Obese Children of South West of Romania.” Paediatric Research 68: 549-550.
Martinez-Gomez, D., Ruiz, J.R., Ortega, F.B., Veiga, O.L., Moliner-Urdiales, D., Mauro, B., Galfo, M., Manios, Y., Widhalm, K., Beghin, L., Moreno, L.A., Molnar, D., Marcos, A., Sjostrom, M., & the HELENA Study Group (2010). "Recommended levels of physical activity to avoid an excess of body fat in European adolescents: the HELENA Study." American Journal of Preventative Medicine 39(3): 203-211.
Matejek, C., Planincec, J., Fosnaric, S., & Pisot, R. (2014). “Relations of weight status and physical fitness of children in Slovenia.” Zdrav Var 53: 11-16.
Mavrakanas, T.A., Konsuola, G., Patsonis, I., & Merkouris, B.P. (2009). “Childhood obesity and elevated blood pressure in a rural population of northern Greece.” Rural and Remote Health 9: 1150.
Mazaraki, A., Tsioufis, C., Dimitriadis, K., Tsiachris, D., Stefanadi, E., Zampelas, A., Richter, D., Mariolis, A., Panagiotakos, D., Tousoulis, D., & Stefanadis, C. (2011). "Adherence to the Mediterranean diet and albuminuria levels in Greek adolescents: data from the Leontio Lyceum ALbuminuria (3L study)." European Journal of Clinical Nutrition 65(2): 219-225.
Mazicioglu, M. M., Kurtoglu, S., Ozturk, A., Hatipoglu, N., Cicek, B., & Ustunbas, H.B. (2010). "Percentiles and mean values for neck circumference in Turkish children aged 6-18 years." Acta Paediatrica 99(12): 1847-1853.
Mebrahtu, T.F., Feltbower, R.G., Greenwood, D.C., & Parslow, R.C. (2015). "Childhood body mass index and wheezing disorders: a systematic review and meta‐analysis." Pediatric Allergy and Immunology 26(1): 62-72.
Mei, Z., Grummer-Strawn, L.M., Pietrobelli, A., Goulding, A., Goran, M.I., & Dietz, W.H. (2002). "Validity of body mass index compared with other body-composition screening indexes for the assessment of body
fatness in children and adolescents." American Journal of Clinical Nutrition 75: 978-985.
Messiah, S., Arheart, C., Lipshultz, S.E., & Miller, T.L. (2011). "Ethnic Group Differences in Waist Circumference Percentiles Among U.S. Children and Adolescents: Estimates from the 1999–2008 National Health and Nutrition Examination Surveys." Metabolic Syndrome and Related Disorders 9(4): 297-303.
Mihai, C.M., Stoicescu, R.M., Mihai, L., Cuzic, V., & Balasa, A. (2011). “The assessment of glucose tolerance in children with obesity.” Paediatric Research 70: 387.
Mirkopoulou, D., Grammatikopoulou, M.G., Gerothanasi, K., Tagka, A., Stylianou, C., & Hassapidou, M. (2010). “Metabolic indices, energy and macronutrient intake according to weight status in a rural sample of 17-year-old adolescents.” Rural & Remote Health 10(4): 1513.
Mocanu, V. (2013). “Prevalence of Overweight and Obesity in Urban Elementary School Children in Northeastern Romania: Its Relationship with Socioeconomic Status and Associated Dietary and Lifestyle Factors.” Hindawi Publishing Corporation BioMed Research International, ArticleID537451, http://dx.doi.org/10.1155/2013/537451
Mocnik, M., Nikolic, S., & Varda, N.M. (2015). “Arterial Compliance Measurement in Overweight and Hypertensive Children.” Indian Journal of Pediatrics DOI 10.1007/s12098-015-1965-2
Morea, M., & Miu, N. (2013). “Metabolic syndrome in children.” Human & Veterinary Medicine 5(3): 103-108.
Moreira, H., Carona, C., Silva, N., Frontini, R., Bullunger, M., & Canavarro, M.C. (2013). “Psychological and quality of life outcomes in pediatric populations: A parent-child perspective.” Journal of Pediatrics 163(5): 1471-1478.
Moreira, P., Padez, C., Mourao-Carvalhal, I., & Rosado, V. (2007). “Maternal weigh gain during pregnancy and overweight in Portuguese children.” International Journal of Obesity 31: 608-614.
Moreno, L., Fleta, J., Mur, L., Feja, C., Sarría, A., & Bueno, M. (1997). "Indices of body fat distribution in Spanish children aged 4.0 to 14.9 years." Journal of Pediatric Gastroenterology and Nutrition 25(2): 175-181.
Moreno, L. A., De Henauw, S., Gonzalez-Gross, M., Kersting, M., Molnar, D., Gottrand, F., Barrios, L., Sjostrom, M., Manios, Y., Gilbert, C.C., Leclercq, C., Widhalm, K., Kafatos, A., Marcos, A., & HELENA Study Group (2008). "Design and implementation of the Healthy Lifestyle in Europe by Nutrition in Adolescence Cross-Sectional Study." International Journal of Obesity 32 Suppl 5: S4-11.
Moschonis, G., Chrousos, G.P., Lionis, C., Mougios, V., Manios, Y., & Healthy Growth Study Group (2012). "Association of total body and visceral fat mass with iron deficiency in preadolescents: the Healthy Growth Study." British Journal of Nutrition 108(4): 710-719.
Mühlig, Y., Antel, J., Focker, M., & Hebebrand, J. (2015). "Are bidirectional associations of obesity and depression already apparent in childhood and adolescence as based on high‐quality studies? A systematic review." Obesity Reviews 17(3): 235-49.
Munger KL, Chitnis T, Ascherio A. Body size and risk of MS in two cohorts of US women. Neurology.
2009;73(19):1543-1550. doi:10.1212/WNL.0b013e3181c0d6e0.
Musil, V., Majer, M., & Juresa, V. (2012). “Elevated Blood Pressure in School Children and Adolescents –
Prevalence and Associated Risk Factors.” Collegium Antropologicum 36 Suppl 1: 147-155.
Must, A., & Anderson, S.E. (2006). "Body mass index in children and adolescents: considerations for population-based applications." International Journal of Obesity 30(4): 590-594.
Must, A., Spadano, J., Coakley, E.H., Field, A.E., Colditz, G., & Dietz, W.H. (1999). "The disease burden associated with overweight and obesity." Journal of the American Medical Association 282(16): 1523-1529.
Nadeau, K.J., Maahs, D.M., Daniels, S.R., & Eckel, R.H. (2011). “Childhood obesity and cardiovascular disease: Links and prevention strategies.” Nature Reviews Cardiology 8(9): 513-525.
Nagy, P., Kovacs, E., Moreno, L.A., Veidebaum, T., Tornaritis, M., Kourides, Y., Siani, A., Lauria, F., Sioen, I., Claessens, M., Marild, S., Lissner, L., Bammann, K., Intemann, T., Buck, C., Pigeot, I., Ahrens, W., Molnar, D., & IDEFICS Consortium (2014). "Percentile reference values for anthropometric body composition indices in European children from the IDEFICS study." International Journal of Obesity 38 Suppl 2: S15-25.
Nanu, M.I., Stativă, E., Moldovanu, F., Stoicescu, S., & Novak, C. (2011). “Chapter 4: Growth and Development.” Evaluation of interventions’ efficiency of the national programs regarding nutrition of children under 2 years. Translated version provided by first author.
Nardone, P., Spinelli, A., Lauria, L., Buoncristiano, M., Bucciarelli, M., Galeone, D. & Gruppa OKkio alla SALUTE (2015). VARIABILITÀ SOCIODEMOGRAFICA NELLE PREVALENZE DI SOVRAPPESO E OBESITÀ DEI
229
BAMBINIIN ITALIA NEL 2014. Roma: Istituto Superiore di Sanità.
Nazareth, M. (2013). Como se cresce em Portugal nos primeiros 3 anos de vida: “Estudo do Padrão Alimentar e de Crescimento Infantil”: EPACI Portugal 2012. Paper presented at EPACI conference, Lisbon, November.
NCD Risk Factor Collaboration (2016). “Trends in adult body-mass index in 200 countries from 1975 to 2014: a pooled analysis of 1698 population-based measurement studies with 19·2 million participants.” The Lancet 387: 1377-1396.
Neovius, K., Neovius, M., Kark, M., & Rasmussen, F. (2012a). “Association between obesity status and sick-leave in Swedish men: nationwide cohort study.” European Journal of Public Health 22(1): 112-116.
Neovius, K., Rehnberg, C., Rasmussen, F., & Neovius, M. (2012b). “Lifetime Productivity Losses Associated with Obesity Status in Early Adulthood: A Population-Based Study of Swedish Men.” Applied Health Economics and Health Policy, 10(5): 309-317.
Nicolescu, R., Cucu, A., Branduse, L., Dumitrache, C., Standescu, C.T., Kassai, V., Drost, M. (2013). Nutritional status assessment in children from primary school by participation in the European Childhood Obesity Surveillance Initiative (COS). Bucharest: Romania National Institute of Public Health.
Ng, M., Fleming, T., Robinson, M., Thomson, B., Graetz, N., Margono, C., Mullany, E.C., Biryukov, S., Abbafati, C., Abera, S.F., Abraham, J.P., Abu-Rmeileh, N.M.E., Achoki, T., AlBuhairan, F.S., Alemu, Z.A., Alfonso, R., Ali, M.K., Ali, R., Guzman, N.A., Ammar, W., Anwari, P., Banerjee, A., Barquera, S., Basu, S., Bennett, D.A., Bhutta, Z., Blore, J., Cabral, N., Nonato, I.C., Chang, J.C., Chowdhury, R., Courville, K.J., Criqui, M.H., Cundiff, D.K., Dabhadkar, K.C., Dandona, L., Davis, A., Dayama, A., Dharmaratne, S.D., Ding, E.L., Durrani, A.M., Esteghamati, A., Farzadfar, F., Fay, D.F.J., Feigin, V.L., Flaxman, A., Forouzanfar, M.H., Goto, A., Green, M.A., Gupta, R., Hafezi-Nejad, N., Hankey, G.J., Harewood, H.C., Havmoeller, R., Hay, S., Hernandez, L., Husseini, A., Idrisov, B.T., Ikeda, N., Islami, F., Jahangir, E., Jassal, S.K., Jee, S.H., Jeffreys, M., Jonas, J.B., Kabagambe, E.K., Khalifa, S.E.A.H., Kengne, A.P., Khader, Y.S., Khang, Y.-H. , Kim, D., Kimokoti, R.W., Kinge, J.M., Kokubo, Y., Kosen, S., Kwan, G., Lai, T., Leinsalu, M., Li, Y., Liang, X., Liu, S. Logroscino, G., Lotufo, P.A., Lu,Y., Ma, J., Mainoo, N.K., Mensah, G.A., Merriman, T.R., Mokdad, A.H., Moschandreas, J., Naghavi, M., Naheed, A., Nand, D., Narayan, K.M.V., Nelson, E.L., Neuhouser, M.L., Nisar, M.I., Ohkubo, T., Oti, S.O., Pedroza, A., Prabhakaran, D., Roy, N., Sampson, U., Seo, H., Sepanlou, S.G., Shibuya, K., Shiri, R., Shiue, I., Singh, G.M., Singh, J.A., Skirbekk, V., Stapelberg, N.J.C., Sturua, L., Sykes, B.L., Tobias, M., Tran, B.X., Trasande, L., Toyoshima, H., van de Vijver, S., Vasankari, T.J., Veerman, J.L., Velasquez-Melendez, G., Vlassov, V.V., Vollset S.E., Vos, T., Wang, C., Wang, X., Weiderpass, E., Werdecker, A., Wright, J.L., Yang, Y.C., Yatsuya, H., Yoon, J., Yoon, S.J., Zhao, M., Zhou, Y., Zhu, S., Lopez, A.D., Murray, C.J.L., & Gakidou, E. (2014). "Global, regional, and national prevalence of overweight and obesity in children and adults during 1980–2013: a systematic analysis for the Global Burden of Disease Study 2013." The Lancet 384(9945): 766-781.
Nogueira, H., Ferrao, M., Gama, C., Mourao, I., Marques, V.R., & Padez, C. (2013). “Perceptions of neighborhood environments and childhood obesity: Evidence of harmful gender inequities among Portuguese children.” Health & Place 19: 69-73.
O'Malley, G., Hussey, J., & Roche, E. (2012). "A pilot study to profile the lower limb musculoskeletal health in children with obesity." Pediatric Physical Therapy 24(3): 292-298.
O’Malley, G., Elmes, M., Keating, R., Killeen, S., Doyle, S., Murphy, S., & Lennon, O. (2015a). “Exploring the prevalence of musculoskeletal impairments in children and adolescents attending an obesity management service.” Appetite 89: 309.
O’Malley, G., Keating, R., Elmes, M., Killeen, Sheridan, N., Murphy, S., & Brinkley, A. (2015b). “standing balance and health-related quality of life in children who are obese.” Appetite 89: 309.
O'Neill, J., McCarthy, S., Burke, S., Hannon, E., Kiely, M., Flynn, A., Flynn, M., & Gibney, M. (2007). "Prevalence of overweight and obesity in Irish school children, using four different definitions." European Journal of Clinical Nutrition 61(6): 743-751.
O’Shea, B., Ladewig, E.L., Kelly, A., Reulbach, U., & O’Dowd, T. (2014). “Weighing children; parents agree, but GPs conflicted.” Archives of Disease in Childhood 99: 543-545.
Olds, T., Maher, C., Zumin, S., Peneau, S., Lioret, S., Castetbon, K., Bellisle, de Wilde, J., Hohepa, M., Maddison, R., Lissner, L., Sjoberg, A., Zimmermann, M., Aeberli, I., Ogden, C., Flegal, K., & Summerbell, C. (2011). "Evidence that the prevalence of childhood overweight is plateauing: data from nine countries." International Journal of Pediatric Obesity 6(5-6): 342-360.
Owen, C.G., Whincup, P.H., Orfei, L., Chou, Q.A., Rudnicka, A.R., Wathern, A.K., Kaye, S.J., Eriksson, J.G., Osmond, C., & Cook, D.G. (2009). “Is body mass index before middle age related to coronary heart
disease risk in later life? Evidence from observational studies.” International Journal of Obesity 33(8): 866-877.
Padez, C. (2006). “Trends in overweight and obesity in Portuguese conscripts from 1986 to 2000 in relation to place of residence and educational level.” Public Health 120: 946-952.
Padez, C., Fernandes, T., Mourao, I., Moreira, P., & Rosado, V. (2004). “Prevalence of Overweight and Obesity in 7–9-Year-Old Portuguese Children: Trends in Body Mass Index From 1970–2002.” American Journal of Human Biology 16: 670-678.
Pampel, F.C., Denney, J.T., & Krueger, P.M. (2012). “Obesity, SES, and economic development: A test of the reversal hypothesis.” Social Science & Medicine 74(7), 1073-1081.
Papadimitriou, A., Kounadi, D., Konstantinidou, M., Xepapadaki, P., & an Nicolaidou, P. (2006). “Prevalence of Obesity in Elementary Schoolchildren Living in Northeast Attica, Greece.” Obesity 14: 1113-1117.
Papadopoulou-Alataki, E., Papadopoulou-Legbelou, K., Doukas, L., Karatzidou, K., Pavlitou-Tsiontsi, A., & Pagkalos, E. (2004). "Clinical and biochemical manifestations of syndrome X in obese children." European Journal of Pediatrics 163(10): 573-579.
Papandreou, D., Karabouta, Z., Pantoleon, A., & Rousso, I. (2012). "Investigation of anthropometric, biochemical and dietary parameters of obese children with and without non-alcoholic fatty liver disease." Appetite 59(3): 939-944.
Papandreou, D., Malindretos, P., & Rousso, I. (2010). "Risk factors for childhood obesity in a Greek paediatric population." Public Health Nutrition 13(10): 1535-1539.
Papandreou, D., Rousso, I., Malindretos, P., Makedou, A., Moudiou, T., Pidonia, I., Pantoleon, A., Economou, I., & Mavromichalis, I. (2008). "Are saturated fatty acids and insulin resistance associated with fatty liver in obese children?" Clinical Nutrition 27(2): 233-240.
Park, M.H., Falconer, C., Viner, R.M., & Kinra, S. (2012). “The impact of childhood obesity on morbidity and mortality in adulthood: a systematic review.” Obesity Reviews 13(11): 985-1000.
Patsopoulou, A., Tsimtsiou, Z., Katsioulis, A., Rachiotis, G., Malissiova, E., & Hadjichristodoulou, C. (2015). "Prevalence and Risk Factors of Overweight and Obesity among Adolescents and Their Parents in Central Greece (FETA Project)." International Journal of Environmental Research & Public Health 13(1).
Paulis, W. D., Silva, S., Koes, B.W., & van Middelkoop, M. (2013). "Overweight and obesity are associated with musculoskeletal complaints as early as childhood: a systematic review." Obesity Reviews doi: 10.1111/obr.12067
Pedrosa, C., Oliveira, B.M.P.M., Albuquerque, I., Simoes-Pereira, C., Vaz-de-Almeida, M., & Correia, F. (2010). “Obesity and metabolic syndrome in 7-9 years-old Portuguese schoolchildren.” Diabetology & Metabolic Syndrome 2: 40.
Pelin, A.M., & Matasaru, S. (2012). “Metabolic syndrome in obese children and adolescents.” Medical-surgical Journal of the Society of Physicians and Naturalists, Iaşi 116(4): 957-961.
Perry, C., Keane, E., Fitzgerald, A.P., Layte, R., Perry, I.J., & Harrington, J.A. (2015). “The use of a Dietary Quality Score as a predictor of childhood overweight and obesity.” BMC Public Health 15: 581.
Perry, I.J., Dee, A., Staines, A., McVeigh, T., Sweeney, M.R., O’Neill, C., Callan, A., Doherty, E., O’Dwyer, V., Kearns, K., Sharp, L., Kee, F., Hughes, J., & Balanda, K. (2012). The cost of overweight and obesity on the island of Ireland. Cork: safefood.
Perry, I. J., Whelton, H., Harrington, J., & Cousins, B. (2009). "The heights and weights of Irish children from the post-war era to the Celtic tiger." Journal of Epidemiology and Community Health 63(3): 262-264.
Pervanidou, P., Bastaki, D., Chouliaras, G., Papanikolaou, K., Kanaka-Gantenbein, C., & Chrousos, G., (2015). “Internalizing and externalizing problems in obese children and adolescents: associations with daily salivary cortisol concentrations.” Hormones 14(4): 623-631.
Pervanidou, P., Bastaki, D., Chouliaras, G., Papanikolaou, K., Laios, E., Kanaka-Gantenbein, C., & Chrousos, G.P. (2013). "Circadian cortisol profiles, anxiety and depressive symptomatology, and body mass index in a clinical population of obese children." Stress 16(1): 34-43.
Petranovic, M.Z., Thomas, Z., Narancic, N.S., Skaric-Juric, T., Vecek, A., & Milicic, J. (2014). “A six decades long follow-up on body size in adolescents from Zagreb, Croatia (1951-2010).” Economics & Human Biology 13: 155-164.
Petricevic, N., Puharic, Z., Posavic, M., Simetin, I.P., & Franelic, I.P. (2012). “Family history and parental recognition of overweight in Croatian children.” European Journal of Pediatrics 171: 1209-1214.
Pietrobelli, A., Faith, M.S., Allison, D.B., Gallagher, D., Chiumello, G., & Heymsfield, S.B. (1998). "Body mass index as a measure of adiposity among children and adolescents: A validation study." Journal of
231
Pediatrics 132(2): 204-210.
Pigeot, I., Barba, G., Chadjigeorgiou, C., de Henauw, P., Kourides, Y., Lissner, L., Marild, S., Pohlabeln, H., Russo, P., Tornaritis, M., Veidebaum, T., Wawro, N., & Siani, A. (2009). "Prevalence and determinants of childhood overweight and obesity in European countries: pooled analysis of the existing surveys within the IDEFICS Consortium." International Journal of Obesity 33(10): 1103-1110.
Pomerleau, J., Knai, C., Branca, F., Robertson, A., Rutter, H., McKee, M., Brunner, E., & EURO-PREVOB Consortium (2008). Prevention of Obesity in Europe – Consortium for the prevention of obesity through effective nutrition and physical activity actions – EURO-PREVOB – Tackling the social and economic determinants of nutrition and physical activity for the prevention of obesity across Europe. D3.1: Review of the literature of obesity (and inequalities in obesity) in Europe and of its main determinants: nutrition and physical activity. EURO-PREVOB Consortium.
Popescu, L.A., Vigolici, B., Pacurar, D., Timnea, O., Ranetti, A.E., Oraseanu, D., & Zagrean, L. (2013). “Beneficial effects of omega-3 fatty acids in non-alcoholic fatty liver disease in childhood obesity.” Farmacia 61(3): 598-608.
Pomp ER, Le Cessie S, Rosendaal FR, Doggen CJM. Risk of venous thrombosis: obesity and its joint effect with
oral contraceptive use and prothrombotic mutations. British Journal of Haematology.
2007;139(2):289-96.
Power, C., Frank, J., Hertzman, C., Schierhout, G., & Li, L. (2001). “Predictors of low back pain onset in a prospective British study.” American Journal of Public Health 91(10): 1671-1678.
Pulgarón, E. R. (2013). "Childhood obesity: a review of increased risk for physical and psychological comorbidities." Clinical Therapeutics 35(1): A18-A32.
Putarek, N.R., Ille, J., Uroic, A.S., Skrabic, V., Stipancic, G., Krnic, N., Radica, A., Marjanac, I., Severinski, S., Svigir, A., Bogdanic, A., & Dumic, M. (2015). “Incidence of type 1 diabetes mellitus in 0 to 14-yr-old children in Croatia – 2004 to 2012 study.” Pediatric Diabetes 16: 448-453.
Queally, M., Doherty, E., & Carter, L. (2016). The effect of overweight and obesity on morbidity in childhood. Report commissioned by safefood Ireland. Cork: safefood.
Raj, M. (2012). "Obesity and cardiovascular risk in children and adolescents." Indian Journal of Endocrinology & Metabolism 16(1): 13.
Rees, R., Oliver, K., Woodman, J., & Thomas, J. (2011). "The views of young children in the UK about obesity, body size, shape and weight: a systematic review." BMC Public Health 11: 188.
Reilly, J.J. (2002). "Assessment of childhood obesity: National reference data or international approach?" Obesity Research 10(8): 838-840.
Reilly, J.J., & Kelly, J. (2011). “Long-term impact of overweight and obesity in childhood and adolescence on morbidity and premature mortality in adulthood: systematic review.” International Journal of Obesity 35(7): 891-898.
Reilly, J J., Methven, E., McDowell, Z.C., Hacking, B., Alexander, D., Stewart, L., & Kelnar, C.J.H. (2003). "Health consequences of obesity." Archives of Disease in Childhood 88: 748-752.
Ribeiro, J., Guerra, S., Pinto, A., Oliveira, J., Duarte, J., & Mota, J. (2003). “Overweight and obesity in children and adolescents: relationship with blood pressure, and physical activity.” Annals of Human Biology 30(2): 203-213.
Rito, A.I., Paixao, E., Carvahlo, M.A., & Ramos, C. (2012a). Childhood Obesity Surveillance Initiative: COSI Portugal 2010. Lisbon: Instituto Nacional de Saúde.
Rito, A.I., Wijnhoven, T.M.A., Rutter, H., Carvalho, R., Paixão, E., Ramos, C., Claudio, D., Espanca, R., Sancho, T., Cerqueira, Z., Carvahlo, R., Faria, C., Feliciano, E., & Breda, J. (2012b). “Prevalence of obesity among Portuguese children (6–8 years old) using three definition criteria: COSI Portugal, 2008.” Pediatric Obesity 7: 413-422.
Rito, A.I., & Graça, P. (2015). Childhood Obesity Surveillance Initiative: COSI Portugal 2013. Lisbon: Instituto Nacional de Saúde.
Robertson, A., Lobstein, T., & Knai, C. (2007). Obesity and socio-economic groups in Europe: Evidence review and implications for action(SANCO/2005/C4-NUTRITION-03). European Commission.
Roditis, M.L., Parlapani, E.S., Tzotzas, T., Hassapidou, M., & Krassas, G.E. (2009). “Epidemiology and Predisposing Factors of Obesity in Greece: From the Second World War Until Today.” Journal of Pediatric Endocrinology & Metabolism 22: 389-405.
Rodriguez-Rodriguez, E., Palmeros-Exsome, C., Lopez-Sobaler, A.M., & Ortega, R.M. (2011). "Preliminary data
on the association between waist circumference and insulin resistance in children without a previous diagnosis." European Journal of Pediatrics 170(1): 35-43.
Rokholm, B., Baker, J.L., & Sorensen, T.I. (2010). "The levelling off of the obesity epidemic since the year 1999--a review of evidence and perspectives." Obesity Reviews 11(12): 835-846.
Rusescu, A. (2006). Nutritional status of the pregnant woman: Romania - 2006. Bucharest: Institute for Protection of Mother and Child.
Sakka, S., Siahanidou, T., Voyatzis, C., Pervanidou, P., Kaminioti, C., Lazopoulou, N., Kanaka-Gantenbein, C., Chrousos, G.P., & Papassotiriou, I. (2015). "Elevated circulating levels of lipoprotein-associated phospholipase A2 in obese children." Clinical Chemistry & Laboratory Medicine 53(7): 1119-1125.
Sakou, I.I., Psaltopoulou, T., Sergentanis, T.N., Karavanaki, K., Karachaliou, F., Ntanasis-Stathopoulos, I., Tzanninis, S., Sdogou, T., Greydanus, D., & Tsitsika, A. (2015). “Insulin resistance and cardiometabolic risk factors in obese children and adolescents: a hierarchical approach.” Journal of Pediatric Endocrinology & Metabolism 28 (5-6): 5589-596.
Sanders, R. H., Han, A., Baker, J.S., & Cobley, S. (2015). "Childhood obesity and its physical and psychological co-morbidities: a systematic review of Australian children and adolescents." European Journal of Pediatrics 174(6): 715-746.
Sardinha, L.B., Santos, R., Vale, S., C Silva, A.M. Ferreiro, J.P, Raimundo, A.M, Moreira, H., Baptista, F. & Mota, J. (2011). “Prevalence of overweight and obesity among Portuguese youth: A study in a representative sample of 10– 18-year-old children and adolescents.” International Journal of Pediatric Obesity 6: e124-e128.
Sargent, J.D., & Blanchflower, D.G. (1994). “Obesity and stature in adolescence and earnings in young adulthood: Analysis of a British birth cohort.” Archives of Pediatrics & Adolescent Medicine 148(7): 681-687.
Sassi, F., Devaux, M., Church, J., Cecchini, M., & Borgonovi, F. (2009). “Education and obesity in four OECD countries.” OECD Health Working Papers (46). Paris: OECD.
Sedej, K., Kotnik, P., Avbelj Stefanija, M., Groselj, U., Sirca Campa, A., Lusa, L., Battelino, T., & Bratina, N. (2014). "Decreased prevalence of hypercholesterolaemia and stabilisation of obesity trends in 5-year-old children: possible effects of changed public health policies." European Journal of Endocrinology 170(2): 293-300.
Serban, V., Brink, S., Timar, B., Sima, A., Vlad, M., Timar, R., & Vlad, A., for the ONROCAD study group (2015). “An increasing incidence of type 1 diabetes mellitus in Romanian children aged 0 to 17 years.” Journal of Pediatric Endocrinology & Metabolism 28(3-4): 293-298.
Severens, J.L., & Milne, R.J. (2004). Discounting health outcomes in economic evaluation: The ongoing debate. Value in Health 7(4): 397-401.
Sherry, B., Jefferds, M.E., & Grummer-Strawn, L.M. (2007). "Accuracy of self-reported of height and weight in assessing overweight status." Archives of Pediatrics & Adolescent Medicine 161(12): 1154-1161.
Shields, M., Connor Gorber, S., Janssen, I., & Tremblay, M.S. (2011). "Obesity estimates for children based on parent-reported versus direct measures." Health Reports (Statistics Canada) 22(3): 1-13.
Shields, M. & Tremblay, M.S. (2010). "Canadian childhood obesity estimates based on WHO, IOTF and CDC cut-points." International Journal of Pediatric Obesity 5(3): 265-273.
Shrewsbury, V. & Wardle, J. (2008). "Socioeconomic status and adiposity in childhood: a systematic review of cross-sectional studies 1990-2005." Obesity 16(2): 275-284.
Sikorski, C., Luppa, M., Luck, T., & Riedel-Heller, S.G. (2015). “Weight stigma "gets under the skin" - Evidence for an adapted psychological mediation framework - A systematic review.” Obesity 23(2): 266-276.
Simmonds, M., Llewellyn, A., Owen, C.G., & Woolacott, N. (2016). “Predicting adult obesity from childhood obesity: a systematic review and meta-analysis.” Obesity Reviews 17(2): 95-107.
Sindicic Dessardo, N., Dessardo, S., Sasso, N., Sarunic, A.C., & Dezulovic, M.S. (2010). “Pediatric Idiopathic Intracranial Hypertension: Clinical and Demographic Features.” Collegium Antropologicum 34 Suppl 2: 217-221.
Singh, A.S., Mulder, C., Twisk, J.W.R., Van Mechelen, W., & Chinapaw, M.J.M. (2008). “Tracking of childhood overweight into adulthood: A systematic review of the literature.” Obesity Reviews 9(5): 474-488.
Skledar, M.T., & Milosevic, M. (2015). “BREASTFEEDING AND TIME OF COMPLEMENTARY FOOD INTRODUCTION AS PREDICTORS OF OBESITY IN CHILDREN.” Central European Journal of Public Health 23: 26-31.
233
Smith, S. M., Sumar, B., & Dixon, K.A. (2014). "Musculoskeletal pain in overweight and obese children." International Journal of Obesity 38(1): 11-15.
Sobal, J. & Stunkard, A.J. (1989). "Socioeconomic status and obesity: A review of the literature." Psychological Bulletin 105(2): 260-275.
Sonntag, D., Ali, S., Lehnert, T., Konnopka, A., Riedel-Heller, S., & König, H. H. (2015). “Estimating the lifetime cost of childhood obesity in Germany: Results of a Markov Model.” Pediatric Obesity, 10(6): 416-422.
Sonntag, D., Ali, S., & De Bock, F. (2016). “Lifetime indirect cost of childhood overweight and obesity: A decision analytic model”. Obesity, 24(1): 200-206.
Spathopoulos, D., Paraskakis, E., Trypsianis, G., Tsalkidis, A., Arvanitidou, V., Emporiadou, M., Bouros, D., & Chatzimichael, A. (2009). "The effect of obesity on pulmonary lung function of school aged children in Greece." Pediatric Pulmonology 44(3): 273-280.
Spinelli, A., Lamberti, A., Baglio, G., Andreozzi, S., & Galeone, D. (2009). Sistema di sorveglianza OKkio alla SALUTE: sistema di sorveglianza su alimentazione e attività fisica nei bambini della scuola primaria. Risultati 2008. Roma: Istituto Superiore di Sanità.
Spinelli, A., Lamberti, A., Nardone, P., Andreozzi, S., & Galeone, D. (2012). Sistema di sorveglianza OKkio alla SALUTE: Risultati 2010. Roma: Istituto Superiore di Sanità.
Spinelli, A., Nardone, P., Buoncristiano, M., Lauria, L., Andreozzi, S., & Galeone, D. (2014). Sistema di sorveglianza OKkio alla SALUTE: dai risultati 2012 alle azioni. Roma: Istituto Superiore di Sanità.
Spinelli, A., Nardone, P., Buoncristiano, M., Lauria, L., Andreozzi, S., Galeone, D. & Gruppa OKkio alla SALUTE (2015). ITALIA 2014:L’OBESITÀ NEI BAMBINI STA DIMINUENDO. Roma: Istituto Superiore di Sanità.
Starc, G., (2014). ZDRAV ŽIVLJENJSKI SLOG 3600 ZA DOBRO OTROK. SKUPAJ ZA BOLJŠE ZDRAVJE OTROK IN MLADOSTNIKOV - OHRANJANJE IN ZAGOTAVLJANJE ENAKIH MOŽNOSTI (pp. 12-19). Ljubljana: NIJZ.
Stipancic, G., La Grasta Sabolic, L., Malencia, M., Radica, A., Skrabic, V., & Tiljak, M.K. (2008). “Incidence and trends of childhood Type 1 diabetes in Croatia from 1995 to 2003.” Diabetes Research & Clinical Practice 80: 122-127.
Stolzman, S., Irby, M.B., Callahan, A.B., & Skelton, J.A. (2015). "Pes planus and paediatric obesity: a systematic review of the literature." Clinical Obesity 5(2): 52-59.
Tambalis, K.D., Panagiotakos, D.B., Kavouras, S.A., Kallistratos, A.A., Moraiti, I.P., Douvis, S.J., Toutouzas, P.K., & Sidossis, L.S. (2010). "Eleven-year prevalence trends of obesity in Greek children: first evidence that prevalence of obesity is leveling off." Obesity 18(1): 161-166.
Tambalis, K.D., Panagiotakos, D.B., Psarra, G., & Sidossis, L.S. (2011). "Inverse but independent trends in obesity and fitness levels among Greek children: a time-series analysis from 1997 to 2007." Obesity Facts 4(2): 165-174.
Tambalis, K.D., Panagiotakos, D.B., Kavouras, S.A., Papoutsakis, S., & Sidossis, L.S. (2013). "Higher prevalence of obesity in Greek children living in rural areas despite increased levels of physical activity." Journal of Paediatrics & Child Health 49(9): 769-774.
Taylor, R. W., Jones, I.E., Williams, S.M., & Goulding, A. (2000). "Evaluation of waist circumference, waist-to-hip ratio, and the conicity index as screening tools for high trunk fat mass, as measured by dual-energy X-ray absorptiometry, in children aged 3–19 y." American Journal of Clinical Nutrition 72: 490-495.
Teixera, P.J., Sardinha, L.B., Going, S.B., & Lohman, T.G. (2001). “Total and Regional Fat and Serum Cardiovascular Disease Risk Factors in Lean and Obese Children and Adolescents.” Obesity Research 9(8): 432-442.
Trasande, L. (2010). “How much should we invest in preventing childhood obesity?” Health Affairs 29(3): 372-378.
Trikaliotis, A., Boka, V., Kotsanos, N., Karagiannis, V., & Hassapidou, M. (2011). “dmfs and BMI in pre-school Greek children. An epidemiological study.” European Archives of Paediatric Dentistry 12(3): 176-178.
Trogdon, J.G., Finkelstein, E.A., Hylands, T., Dellea, P.S., & Kamal-Bahl, S.J. (2008). “Indirect costs of obesity: A review of the literature.” Obesity Reviews, 9: 489-500.
Tsai, A.G., Williamson, D.F., & Glick, H.A. (2011). “Direct medical cost of overweight and obesity in the USA: a quantitative systematic review.” Obesity Reviews 12: 50-61.
Tucker, D.M.D., Palmer, W.J., Valentine, S.R., & Ray, J.A. (2006). “Counting the costs of overweight and obesity: modeling clinical and cost outcomes.” Current Medical Research & Opinion 22(3): 575-596.
Turchetta, F., Gatto, G., Romano, F., Boccia, A., & La Torre, G. (2012). "Systematic review and meta-analysis of the prevalence of overweight and obesity among school-age children in Italy." Epidemiologia e Prevenzione 36(3-4): 188-195.
Turconi, G., Guarcello, M., Maccarini, L., Bazzano, R., Zaccardo, A., & Roggi, C. (2006). “BMI values and other anthropometric and functional measurements as predictors of obesity in a selected group of adolescents.” European Journal of Nutrition 45: 136-143.
Turconi, G., Maccarini, L., Bazzano, R., & Roggi, C. (2007). “Overweight and blood pressure: results from the examination of a selected group of adolescents in northern Italy.” Public Health Nutrition 11(9): 905-913.
Twig, G., Yaniv, G, Levine, H., Leiba, A., Goldberger, N., Derazne, E., Ben-Ami Shor, D., Tzur, D., Afek, A., Shamiss, A., Haklai, Z., & Kark, J.D. (2012). “Body-mass index in 2.3 million adolescents and cardiovascular death in adulthood.” New England Journal of Medicine DOI: 10.1056/NEJMoa1503840.
Tzotzas, T., Kapantais, E., Tziomalos, K., Ioannidis, I., Mortoglou, A., Bakatselos, S., Kaklamanou, M., Lanaras, L., & Kaklamanos, I. (2008). "Epidemiological survey for the prevalence of overweight and abdominal obesity in Greek adolescents." Obesity 16(7): 1718-1722.
Vale, S., Santos, R., Soares-Miranda, L., Rego, C., Moreira, P., & Mota, J. (2011). “Prevalence of overweight and obesity among Portuguese pre-schoolers.” Archives of Exercise in Health & Disease 2: 65-68.
Valean, C., Ichim, G., Tatar, S., Samasca, G., Leucuta, A., & Nanulescu, M. (2010). “Prevalence of metabolic syndrome and serum profile of adipokines (leptin and adiponectin) in children with overweight or obesity.” Acta Endocrinologica (Buc) VI: 343-354.
Valean, C., Tatar, S., Nanulescu, M., Leucuta, A., & Ichim, G. (2009). “PREVALENCE OF OBESITY AND OVERWEIGHT AMONG SCHOOL CHILDREN IN CLUJ-NAPOCA.” Acta Endocrinologica (Buc) V: 213-219.
Valerio, G., Maffeis, C., Balsamo, A., Miraglia Del Giudice, E., Brufani, C., Grugni, G., Licenziati, M.R., Brambilla, P., & Manco, M., on the behalf of the Childhood Obesity Group of the Italian Society of Pediatric Endocrinology and Diabetology (2013). “Severe Obesity and Cardiometabolic Risk in Children: Comparison from Two International Classification Systems.” PLoS One 8(12): e83793.
van Baal, P.H.M., Polder, J.J., de Wit, A., Hoogenveen, R.T., Feenstra, T.L., Boshuizen, H.C., Engelfriet, P.M., & Brouer, W.B.F. (2008). “Lifetime Medical Costs of Obesity: Prevention No Cure for Increasing Health Expenditure.” PLoS Medicine 5(2): e29.
van Stralen, M.M., te Velde, S.J., van Nassau, F., Brug, J., Grammatikaki, E., Maes, L., De Bourdeaudhuij, I., Verbestel, V., Galcheva, S., Iotova, V., Koletzko, B.V., von Kries, R., Bayer, O., Kulaga, Z., Serra-Majem, L., Sanchez-Villegas, A., Ribas-Barba, L., Manios, Y., Chinapaw, M.J., & ToyBox Study Group (2012). "Weight status of European preschool children and associations with family demographics and energy balance-related behaviours: a pooled analysis of six European studies." Obesity Reviews 13 Suppl 1: 29-41.
Vassiloudis, I., Yiannakouris, N., Panagiotakos, D.B., Apostolopoulos, K., & Costarelli, V. (2014). "Academic performance in relation to adherence to the Mediterranean diet and energy balance behaviors in Greek primary schoolchildren." Journal of Nutrition Education & Behaviour 46(3): 164-170.
Veltsista, A., Kanaka, C., Gika, A., Lekea, V., Roma, E., & Bakoula, C. (2010). "Tracking of overweight and obesity in Greek youth." Obesity Facts 3(3): 166-172.
Verbeeten, K., Elks, C.E., Daneman, D., & Ong, K.K. (2011). "Association between childhood obesity and subsequent Type 1 diabetes: a systematic review and meta‐analysis." Diabetic Medicine 28(1): 10-18.
Vicente-Rodriguez, G., Libersa, C., Mesana, M.I., Beghin, L., Iliescu, C., Moreno Aznar, L.A., Dallongeville, J., Gottrand, F., & HELENA Study Group (2007). "Healthy lifestyle by nutrition in adolescence (HELENA). A new EU funded project." Therapie 62(3): 259-270.
Viner, R.M., & Cole, T.J. (2005). “Adult socioeconomic, educational, social, and psychological outcomes of childhood obesity: a national birth cohort study.” BMJ Clinical Research 330(7504): 1354.
Visscher, T.L., Heitmann, B.L., Rissanen, A., Lahti-Koski, M., & Lissner, L. (2015). "A break in the obesity epidemic? Explained by biases or misinterpretation of the data?" International Journal of Obesity 39(2): 189-198.
Walsh, B., & Cullinan, J. (2015). “Decomposing socioeconomic inequalities in childhood obesity: Evidence from Ireland.” Economic & Human Biology 16: 60-72.
Walton, J., McNulty, B.A., Nugent, A.P., Gibney, M.J., & Flynn, A. (2014). "Diet, lifestyle and body weight in Irish children: findings from Irish Universities Nutrition Alliance national surveys." Proc eedings of the Nutrition Society 73(2): 190-200.
Wang, L. Y., Denniston, M., Lee, S., Galuska, D., & Lowry, R. (2010). “Long-term Health and Economic Impact of Preventing and Reducing Overweight and Obesity in Adolescence.” Journal of Adolescent Health 46(5): 467-473. doi:10.1016/j.jadohealth.2009.11.204
Wang, Y. & Lim, H. (2012). "The global childhood obesity epidemic and the association between socio-
235
economic status and childhood obesity." International Review of Psychiatry 24(3): 176-188.
Wang, Y. & Lobstein, T. (2006). "Worldwide trends in childhood overweight and obesity." International Journal of Pediatric Obesity 1(1): 11-25.
Wijnhoven, T. M., Van Raaij, J.M., & Breda, J. (2014b). WHO European Childhood Obesity Surveillance Initiative: Implementation of round 1 (2007/2008) and round 2 (2009/2010). Copenhagen, Denmark, World Health Organization.
Wijnhoven, T. M., van Raaij, J.M., Spinelli, A., Rito, A.I., Hovengen, R., Kunesova, M., Starc, G., Rutter, H., Sjoberg, A., Petrauskiene, A., O'Dwyer, U., Petrova, S., Farrugia Sant'angelo, V., Wauters, M., Yngve, A., Rubana, I.M., & Breda, J. (2013). "WHO European Childhood Obesity Surveillance Initiative 2008: weight, height and body mass index in 6-9-year-old children." Pediatric Obesity 8(2): 79-97.
Wijnhoven, T. M., van Raaij, J.M., Spinelli, A., Starc, G., Hassapidou, M., Spiroski, I., Rutter, H., Martos, E., Rito, A.I., Hovengen, R., Peréz-Farinós, N., Petrauskiene, A., Eldin, N., Braeckevelt, L., Pudule, I., Kunesova, M., & Breda, J. (2014a). "WHO European Childhood Obesity Surveillance Initiative: Body mass index and level of overweight among 6-9-year-old children from school year 2007/2008 to 2009/2010." BMC Public Health 14.
Williams, J., Murray, A., McCrory, C., & McNally, S. (2013). Growing Up in Ireland Longitudinal Study of Children: Development from birth to three years. Dublin: Department of Children and Youth Affairs.
World Health Organization (2000). Obesity: preventing and managing the global epidemic - report of a WHO consultation (Consultation on Obesity, 1997 Geneva, Switzerland). Geneva, Switzerland, World Health Organization.
World Health Organization (2006). WHO child growth standards: length/height-for-age, weight-for-age, weight-for-length, weight-for height and body mass index-for-age: methods and development. Geneva, Switzerland, World Health Organization.
World Health Organization (2011). Waist circumference and waist-hip ratio: Report of a WHO expert consultation, Geneva, 8-11 December 2008. Geneva, Switzerland, World Health Organization.
World Health Organization Regional Office for Europe (2016). Eighth meeting on the WHO European Childhood Obesity Surveillance Initiative, Dubrovnik, Croatia, 18-20 May, 2015. Copenhagen: WHO Regional Office for Europe.
Xekouki, P., Nikolakopoulou, M., Papageorgiou, A., Livadas, S., Voutetakis, A., Magiakou, M.A., Chrousos, G.P., Spiliotis, B.E., & Dacou-Voutetakis, C. (2007). “Glucose Dysregulation in Obese Children: Predictive, Risk, and Potential Protective Factors.” Obesity 15(4): 860-869.
Yngve, A., De Bourdeaudhuij, I., Wolf, A., Grjibovski, A., Brug, J., Due, P., Ehrenblad, B., Elmadfa, I., Franchini, B., Klepp, K.I., Poortvliet, E., Rasmussen, M., Thorsdottir, I., & Perez Rodrigo, C. (2008). "Differences in prevalence of overweight and stunting in 11-year olds across Europe: The Pro Children Study." European Journal of Public Health 18(2): 126-130.
Zannolli, R. & Morgese, G. (1996). "Waist percentiles: A simple test for atherogenic disease?" Acta Paediatrica 85(11): 1368-1369
Zhao ZG, Guo XG, Ba CX, Wang W, Yang YY, Wang J, et al. Overweight, obesity and thyroid cancer risk: a meta- analysis of cohort studies. The Journal of international medical research. 2012;40(6):2041-50.
Zimmermann, E., Gamborg, M., Holst, C., Baker, J.L., Sorensen, T.I., & Berentzen, T.L. (2015). “Body mass index in school-aged children and the risk of routinely diagnosed non-alcoholic fatty liver disease in adulthood: a prospective study based on the Copenhagen School Health Records Register.” BMJ Open 5(4): e006998
APPENDIX 1: POSSIBLE NON-MODELLING PROJECTS Topics for possible non-modelling projects come from the clinical and population perspectives.
A1.1 Conditions that could not be included in the modelling The international and local literature on conditions that could not be included in the modelling will
be reviewed and summarised.
These conditions include:
• Childhood (see Table 7a) and adult (see Table 7.3) conditions where there is moderate
evidence in the literature of an association with childhood obesity
• Acute conditions (see Table 7.2 and Table 7.3) for which necessary model inputs are not
available
• Other conditions for which necessary model inputs are not available
The conditions to be included in the model for each participating country would depend on the
availability of relevant data or proxy data for that country.
A1.2 Experiences of morbidly obese children and their families Existing EASO networks of childhood obesity management clinics (e.g. an Irish clinic has a small
dataset on these costs for those children whose BMI is over the 98th percentile) could address a
number of issues:
Private healthcare costs and other costs (such as work parental absences) borne by morbidly
obese children and their families
Prevalence of psycho-social impacts on morbidly obese children (including mental health
and wellbeing, social and emotional development, and obesity-related QOL73)
Impact of morbid childhood obesity on school attendance and academic performance
A1.3 Childhood obesity and educational outcomes This could address a number of aspects of the impact of childhood obesity on:
Readiness for work and youth unemployment (analysis of SLOFit data in Slovenia)
Third level educational achievement (subsequently and lifetime income loss) (NUIG
systematic review)
A1.4 Inequalities A number of possible projects are being explored:
• Links to the HIPP project
• Variation of recent trends (decline) in childhood obesity with rurality, child age and socio-
economic circumstances (analysis of SLOFit data in Slovenia)
• Geographical variation (analysis of SLOFit data)
73 Reviews of literature on HRQOL in obese US children (JAMA 2003 289(4) and Lancet 2016)
237
• Effect of socio economic (social class, education, HRQOL, income) adjustment of ORs for
childhood diseases (Irish study term, third systematic review)
APPENDIX 2: LIMITATIONS IN EVIDENCE, DATA AND MODELLING
(preliminary list)
The assumptions and limitations listed below were identified during the development of the
Evidence Paper and Study Protocols and form a preliminary list that will be developed during the
project.
A2.1 Evidence The gaps in the evidence base may be summarised as follows:
1. Research on direct and indirect costs of childhood overweight and obesity in the medium- to
long-term: The area in which the gaps in the evidence are most apparent is with respect to
lifetime costs associated with childhood overweight and obesity. The available evidence
suggests a strong need for further European work in this area, particularly with respect to
costs incurred during childhood, estimation of indirect costs, better incorporation of BMI
transitions, and differences across ethnic/racial groups.
2. Inequalities
3. Obesity forecasting and generating BMI trajectories
a. Standardised and robust approaches to adjust for adult BMI into analyses: Adjusting
for adult BMI status, as some studies have done, may over-adjust and therefore
underestimate the contribution of childhood BMI status
b. The manner in which changes in BMI over time are incorporated into analyses. and
4. Standardised approaches in multivariate analyses: Studies on impacts of child/adolescent
overweight/obesity also vary in the extent to which adjustments for potential confounders,
such as socio-economic status, are incorporated into analyses
5. Need for further long-term longitudinal dat and more longitudinal analyses: In examining the
evidence base for both the short-term and long-term impacts of child/adolescent
overweight/obesity, the scarcity of high-quality longitudinal data has been identified. This
makes it difficult to establish firm evidence on the causal relationships involved, particularly
for psychological impacts, where relationships with BMI may be bi-directional. For some
medical conditions such as asthma, the relationships may also be bi-directional.
A2.2 Data 1. Standardised surveillance of preschool children and adolescents: While COSI (Childhood
Obesity Surveillance Initiative) provides valuable national and international surveillance data
on the BMI status of school-aged children, there is no equivalent standardised surveillance
of preschool-aged children or adolescents. The Health Behaviour in School-aged Children
(HBSC) study gathers data on BMI of adolescents, but this is self-reported and prone to
underestimates of BMI which in turn vary by age, country and gender. There is a wealth of
local data on prevalence estimates, but comparisons are difficult due to differences in ages,
sample designs, and use of cut-points.
2. Monitoring to address health inequalities:
239
a. The effective monitoring of the prevalence of overweight and obesity is also
restricted by variations in the extent to which prevalence estimates are available for
different socio-economic, racial and ethnic groups: a priority should be that these
groups are included in study designs.
b. Information on racial/ethnic differences: There is a lack of evidence on differential
impact of overweight and obesity by ethnic or racial groups and it is unclear whether
and how much these differences may be due to socio-economic factors.
A2.3 Modelling
During WP4 we will document assumptions, and limitations in the modelling under the following headings: Population, BMI, Health impacts, Healthcare costs, Societal impacts, Societal costs and Other
POPULATION
Assumptions Limitations
Exclusion of children aged under two years
Exclusion of race and ethnicity: in some countries, it would be important to consider migrant / asylum seeker status74
Differences between children in care and children not in care are not considered.
BMI
Assumptions Limitations
Future age-sex specific BMI distributions of the virtual cohort of 2016’s children follow current trends
Constant lifetime BMI percentile75
No breakdown of the obese category into morbidly and severely obese categories
74
The only systematic review we located that examined variation in weight status across race/ethnic group concluded that the definitions and use of terms like ‘migrant’, ‘ethnic’, etc. varied greatly.
75 Validation study looking at modelling lifetime BMI trajectories will be able to look at this
The measurement of overweight and obesity in epidemiological studies of RRs may not match the BMI categorisations used in JANPA WP4 – for example, they may use some measure of central adiposity and, if they use BMI, they may use different reference curves.
The Disease Module in the UKHF’s existing modelling software’s does not incorporate the duration of obesity and overweight into the BMI categories used in the state transition probability matrix for the Semi-Markov that is used to model the occurrence of disease
Different BMI percentile cut-offs are used for individual children in clinical settings (for example, >98th percentile with UK90) and in population studies (for example, >95th centile with UK90).
HEALTH IMPACTS
Assumptions Limitations
No diseases unrelated to overweight and obesity are considered
The international literature shows that the cost of treatment for conditions not related to overweight and obesity can be higher amongst overweight and obese patients than amongst patients with healthy weight (Hamilton et al (2016) – in preparation). This means that costs –related excess metrics will be underestimated.
Independent disease processes
BMI-specific state transition probabilities constant throughout the simulation
Age-sex specific death rates from all other causes are assumed constant over time
DIRECT HEALTHCARE COSTS
Assumptions Limitations
Private costs (out of pocket costs incurred by patients
241
and their families) are not included in the modelling.
New medications/treatments which improve survival will not be considered in the model.
While we age virtual individuals one year at a time, disease data taken from the data sources or from the epidemiological risk literature may have a different time dimension – it could be ten-year incidence rates, five-year relative risks, etc.
Studies used in bottom-up approaches may use different measures of adiposity (measurement vs self-reports) and this may limit their generalisability.
Only deaths from all other causes are included in the model. Cohort members can die from all other causes (combined) but disease occurrence, direct healthcare and societal costs associated with these causes will not be considered.
SOCIETAL IMPACTS
Assumptions Limitations
Adult productivity losses may be overestimated because they are assumed to commence when an individual develops an obesity-related disease and last until retirement or death.
Adult productivity losses due to:
Reduced productivity at work (presenteeism)
Short term absences
Early retirement are not considered.
SOCIETAL COSTS
Assumptions Limitations Implications
Omission of adjustment means we have assumed no societal cost inflation in the future
Pension costs are not included in the modelling. Ignoring
socio-economic differences, these will tend to higher amongst individuals who are of healthy weight than amongst individuals who are obese or overweight because of premature mortality in the latter.
OTHER
Assumptions Limitations Implications
The uncertainty limits that accompany the model outputs represent the accuracy of the microsimulation (stochastic or aleatoric uncertainty) while those that accompany the input data (parameter uncertainty). Errors around the input data are often not available. Model outputs and excess metrics and effects metrics will not be broken down by disease.
243