Population Health and Health Disparities - A Social...
Transcript of Population Health and Health Disparities - A Social...
Population Health and Health Disparities -
A Social Epidemiology Perspective
Health Disparities Symposium 2003, University of Washington
John Lynch
Department of Epidemiology
Center for Human Growth and Development
Institute for Social Research
University of Michigan
“ The goal of the 2003 Health Disparities Symposium:
Social Determinants: Methods and Practice is to expose
faculty and students to research on health disparities
and the social determinants of health. ”
“social determinants”
“population health”
and
“health disparities”
US Infant Mortality over the 20th Century
0.0
20.0
40.0
60.0
80.0
100.0
120.0
140.0
160.0
180.0
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990
Infa
nt M
orta
lity
per
1000
Attempt to understand the
“social determinants” of this
marker of “population health”
Black – White Disparity in Infant Mortality over the 20th Century
0
50
100
150
200
250
300
350
400
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990
Infant M
ortality
per 1
,000
0.020.040.060.080.0100.0120.0140.0160.0180.0200.0
Absolu
te D
isparity
0
50
100
150
200
250
300
350
400
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990
Infa
nt M
orta
lity
per 1
,000
0
0.5
1
1.5
2
2.5
3
Rela
tive D
isparity
Relative Disparity
Absolute Disparity
White
Black
White
Black
• “Population Health”
• “Absolute Health Disparity”
• “Relative Health Disparity”
Improved in all race/ethnic
groups
Reduced
Increased
Infant Mortality
With thanks to:
George Davey Smith, Bristol Dave Leon, LSHTM, London
Steve Kunitz, Rochester Simon Szreter, Cambridge
Yoav Ben-Shlomo, Bristol Di Kuh, UCL, London
Chris Power, UCL, London David Blane, London
Vladimir Shkolnikov, Rostock Tony McMichael, Canberra
Johan Hallqvist, Karolinska, Stockholm Juan Merlo, Lund, Malmo
Pernille Due, Copenhagen Finn Diderichsen, Copenhagen
Nancy Ross, McGill, Canada Michael Wolfson, Stat. Canada
Sam Harper, Michigan George Kaplan, Michigan
Trivellore Raghunathan, Michigan Richard Cooper, Loyola
Part 1. A brief overview of social epidemiology
Part 2. Does an appreciation of social factors actually improve our
understanding of population health?
- A “social” epidemiology of the Russian Mortality Crisis
Part 3. What are some of the lessons for understanding the social
determinants of population health?
- Considering specificity of links between social factors and health
Part 4. What are some of the lessons for understanding social
disparities in health?
- Considering heterogeneity of links between social factors and health
Part 1. A brief overview of social epidemiology
- One person’s view
You will find examples of studies of social factors and health
in many disciplines
Psychology
Sociology
Demography
Economics
Politics - Public Policy
Urban Planning
Disciplinary Perspectives on Health and Health Inequalities
Characteristics ofSocial Systems
Characteristics ofIndividuals
Health
Politics Characteristics ofPolitical Systems
Health
Psychology Characteristics ofIndividuals Health
Economics Characteristics ofIndividuals
Characteristics ofEconomic Systems
Health
Sociology Characteristics ofSocial Systems
Characteristics ofIndividuals
Health
Epidemiology HealthCharacteristics ofSocial Systems
CharacteristicsOf Individuals
“Modern” Epidemiology
Causes
“Structural”
Epidemiology
Causes
Distal
PoliticalEconomy
“Psychosocial”
Epidemiology
Causes
PsychologySociology
“Behavioral”
Epidemiology
Causes
Behaviour
< “Social” Epidemiology > < “Risk Factor” Epidemiology >
Direct StressPathway
Proximal
GeneticsBiology
“Disease”
The Scope of Social Epidemiology
SocialStructure
Individual SocialPosition
PsychosocialBehavioral
Factors
BiologicalFactors
Why are we interested?
Morbidity MortalityAdapted from Diderichsen (1997)
Epidemiological Footprint of different
Social Structures
Geoffrey Rose
Determinants of individual cases
vs
Determinants of population rate
Marmot, Lancet (1998)
0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.28Median Share of Incom e
200
300
400
500
600
Wor
king
Age
(25-
64) M
orta
lity
US AUSSW EUKCAN
New York
Stockholm
Sydney
Income Inequality and W orking-Age Mortality528 Metropolitan Areas in Five Countries, 1990/91
Toronto
London
Ross, et al (in press)
US UK
CASWE
AUS
The Scope of Social Epidemiology
SocialStructure
SocialPosition
PsychosocialBehavioral
Factors
BiologicalFactors
Why are we interested?
Morbidity MortalityAdapted from Diderichsen (1997)
Socioeconomic Position is a Powerful Risk Factor
0.5
1
1.5
2
2.5
3
3.5
Never Former Current High25%
3rd 2nd Low 25%
Rel
ativ
e H
azar
d: C
VD M
orta
lity
KIHD Study, 1984-1996
Smoking Income
The Scope of Social Epidemiology
SocialStructure
SocialPosition
PsychosocialBehavioral
Factors
BiologicalFactors
Why are we interested?
Morbidity MortalityAdapted from Diderichsen (1997)
Hopelessness and CVD Mortality
1.0 1.0 1.0 1.0 1.0 1.0 1.0
1.8
1.6 1.6 1.51.7 1.7
1.5
2.6
2.22.0 2.0
2.3 2.3
1.7
0
1
2
3
+Biol +SEP +Behav +SR Hlth +Depr +Dx ALL
Low Medium High
RH
Everson, et al (1997)
NCHS (1998)
0
10
20
30
40
50
White, nonHispanic
Black, nonHispanic
Hispanic NativePeoples
Asian orPacific
Islander
< 12 years
12 years13-15 years
16 + years%
livebirths
Education and Smoking During PregnancyAmong Mothers Aged 20 or Older - USA 1996
NCHS (1998)
0
5
10
15
20
25
30
35
Boys Girls
High Income
Middle IncomeNear PoorPoor
%
Income and Sedentary Behavior in Adolescents - USA 1995
The Scope of Social Epidemiology
SocialStructure
SocialPosition
PsychosocialBehavioral
Factors
BiologicalFactors
Why are we interested?
Morbidity MortalityAdapted from Diderichsen (1997)
Income and Biological Risk Factors (KIHD 1990)
900
950
1000
1050
1100
Income Quintiles
Apo B
1200
1250
1300
1350
Income Quintiles
Apo A1
4.5
4.6
4.7
4.8
4.9
5
Income Quintiles
Fasting Glucose
2.5
2.6
2.7
2.8
2.9
3
3.1
3.2
3.3
3.4
3.5
Income Quintiles
Fibrinogen
14
16
18
20
22
24
26
28
I/II III-NM III-M IV/V I/II III-NM III-M IV/V
Body
Mas
s In
dex
Social Class at Birth
WomenMen
Body Mass Index by Social Class at Birth
Power, et al. Lancet. (1997)
Age 7
Age 33
SocialStructure
SocialPosition
PsychosocialBehavioral
Factors
BiologicalFactors
Morbidity Mortality
The Landscape of Social Epidemiology
Income inequalityResidential segregationSocial capital
ClassRaceGender Support, trust, depression,
smoking, diet, physical activity
Macro-biological processes of growth and development, reproductive maturation, lipids, insulin resistance, clotting factors, P-N-E, immune function
Genetic influences
Political EconomyInstitutions
DiscriminationCultureHistory
StructuralMacrosocial
Factors
Genetics Human Biology
Pathological Biomarkers
GeneticCharacteristics
Pathobiology
Lifecourse
Conception Old Age
Individual Characteristics
SocioeconomicBehaviouralPsychosocial
Proximal Social Connections
FamilyFriends
Distal Social ConnectionsNeighborhood Community
Lynch (2000)
Health Status
Work Work
Health Status
Lifecourse
multi-time point exposures
Mechanisms
biological, behavioral,
psychological translation
multi-level exposures
Health and Health Inequalities
at the Individual and Population Levels
Context
Part 2. Does an appreciation of social factors
actually improve our understanding of
population health?
A “social” epidemiology of the Russian Mortality Crisis
A. What are the overall population patterns?
Convergence of Life Expectancy at Birth in Industrialized Countries
25
35
45
55
65
75
85
1740 1790 1840 1890 1940 1990
Industrialised countries, except Eastern Europe and Portugal
Japan
New Zealand
Life expectancy at birth
Caselli(2002)
Recent divergence between industrialized countries
35
40
45
50
55
60
65
70
75
80
85
1900 1920 1940 1960 1980 2000
New Zealand
Japan
Russia
Ukraine
Life expectancy at birth
Caselli(2002)
Contrasting Trends in Life Expectancy
55
60
65
70
75
80
1965 1970 1975 1980 1985 1990 1995 2000
R uss ia
Lithuania
Latvia
Ukraine
Estonia
Life expectancy at birth
Lithuania
R uss ia
Latvia
Ukraine
EstoniaFemales
Males
55
60
65
70
75
80
1965 1970 1975 1980 1985 1990 1995 2000
Life expec tanc y at birth
Poland
Cze ch R e p.
R omaniaHungary
B ulgaria
Slovakia
B ulgaria
Hungary
Cze ch R e p.
Slovakia
Poland
R omania
Females
Males
Soviet Republics Eastern Europe
Russia4 year decline
6+ year decline
Meslé (2002)
B. Does the pattern differ by age?
Age Contribution to Russian Life Expectancy Decline
~ 65%
1990-1994Notzon et al., JAMA, 1998
C. Does the pattern differ by cause?
Cause-specific Mortality: Russia 1984-1994
Ratio: 1987 vs 84
Ratio:1994 vs 87
Age Age
All causes Infectious and parasitic
Neoplasms Circulatory disease
Leon, et al.Leon, et al. Lancet,Lancet, 19971997
Pneumonia Other respiratory
Alcohol related diseasesAccidents and violence
Causes of Death Contributing to Russian Life Expectancy Decline
~ 70%CVD
AccidentsInjury
Violence
1990-1994Notzon et al., JAMA, 1998
D. What does our knowledge of the
individual level risk factors tell us?
What could cause a big rise in CVD mortality?
Artifact of death registration
Collapse of health care
Hypertension
Smoking
Obesity
Lipids – Diet
Physical activity
Alcohol
“Although factors such as nutrition and health
services may be involved, the evidence is that
substantial changes in alcohol consumption over the
period could plausibly explain the main features of
the mortality fluctuations observed.”
Leon, et al. Lancet (1997)
Coronary heart disease and acute alcohol poisoning mortality in Russia, Men 30-59
0
50
100
150
200
250
300
350
400
450
1965 1970 1975 1980 1985 1990 19950
20
40
60
80
100
120
140
Age
-sta
ndar
dise
d ra
te p
er 1
00,0
00
Acute alcohol poisoningAcute alcohol poisoning
CHDCHD
SimultaneousSimultaneous Anti-alcoholCampaign
Injuries, poisoning and violence mortality, Men aged 30-59 (excluding acute alcohol poisoning)
0
300
600
1965 1970 1975 1980 1985 1990 199510
11
12
13
14
15
16
Age
-sta
ndar
dise
d ra
te p
er 1
00,0
00Per capita consum
ption litres pure alcohol/year
Injury/violence mortality rateInjury/violence mortality rate
Alcohol Alcohol consumptionconsumption
Very little time lag Anti-alcoholCampaign
But …
Isn’t alcohol protective for heart disease?
Beer Bingeing and Mortality - Finland
0.5
1.5
2.5
3.5
4.5
5.5
6.5
7.5
All-cause CVD External AMI Fatal AMI
Rel
ativ
e H
azar
d
Kauhanen, et al. BMJ (1997)
< 3 bottles (Reference)
3-5 bottles per session
6+ bottles per session*
* Adjusted for average consumption
Correlates of Alcohol Bingeing in Finnish Men - KIHD
• low physical activity
• hopelessness
• depression
• cynical hostility
• poor childhood
• low education
• blue-collar job
• low income
• smoking
• poor lipid profile
• higher blood pressure
E. How does all this fit with our knowledge of the
pathophysiology of the particular outcome?
Coronary heart disease mortality in Russia
• CHRONIC component :
classic atherosclerotic/thrombotic pathology
associated with known coronary risk factors
• ACUTE component :
alcohol induced arrhythmias/cardiomyopathies
associated with binge drinking
F. Putting together a plausible population health story
that incorporates the social to the biological
“These findings paint a picture of societies in which young
and middle aged men face social and economic disruption on
a large scale, for which they are poorly prepared. For many,
their options are constrained by low levels of education, and
the societies of which they are a part have few systems of
social support.
Poor nutrition and high rates of smoking have already
reduced their chances of a long life. The availability of cheap
alcohol, however, provides a pathway not only to oblivion
but often to premature death.” McKee and Shkolnikov BMJ (2001)
Social Stress
% Fall in Male Life Expectancy
and Labor Market Turnover
Russia1990-94
Walberg, et al.Walberg, et al.BMJ BMJ 19981998
A Model for Social EpidemiologyCulturally contextualized - specific exposures - susceptibilities - outcomes
Social andStructuralUpheaval
SpecificExposure
Binge drinking
Specific Outcomes
CVDAccidents / Violence
structures exposureprobabilities
to binge drinking
Socioeconomicposition
DecreasedLife
Expectancy
Cultural Component – role of alcohol in “vodka belt” culture
Part 3. What are some of the lessons for better
understanding the social determinants of
population health?
- Considering specificity in our explanations
Trends in selected causes of death for women aged 30-59, Russia 1965-1999
0
20
40
60
80
100
120
140
160
180
1965 1970 1975 1980 1985 1990 1995 2000
SDR
per
100
000
Accidents and violence
Ischaemic heart disease
Stomach cancer Breast cancer
Rheumatic heart disease
YearShkolnikov, McKee & Leon, Lancet 2001
Reinstate Specificity
“No, some of our epidemiologic forefathers did not get the notion of
‘specificity’ quite right. But I suggest that a modified version of that
notion has a useful place, after all, as we seek to sort out causal from
non-causal associations.” p. 7
• Specificity of Exposure – an outcome associated with 1 exposure, not another
• Specificity of Outcome – an exposure associated with 1 outcome, not another
• Specificity of Susceptibility – association only in those susceptible to the effect
Weiss. Epidemiology (2002)
Why might specificity of outcome-exposure links be useful in understanding the social causation of disease?
• It may give greater flexibility and explanatory power
because it helps shift our focus toward an understanding of
population health and health disparities, as rooted in the
social and structural distribution of more specific exposures
that affect specific susceptibility to particular outcomes.
Why might specificity of outcome-exposure links be useful in understanding the social causation of disease?
• This means understanding that the social and structural
distribution of relevant risk factors is often contextualized
in regard to time, space, culture and population subgroup.
Recent divergence between industrialized countries
35
40
45
50
55
60
65
70
75
80
85
1900 1920 1940 1960 1980 2000
New Zealand
Japan
Rus s ia
Uk raine
Life expectancy at b irth
Why does Japan have the highest life expectancy in the world?
Caselli(2002)
Some have used this remarkable growth in life expectancy
as evidence for the importance of social cohesion / social
capital and related psychosocial processes
Lower inequality (income) blunts perceptions of relative
disadvantage → greater social cohesion → less stress
mediated disease
Why does Japan have the highest life expectancy in the world?
“… these improvements cannot be explained in terms of changes in
nutrition, health care, preventive health policies, or any of the obvious
factors, …” p. 18
“What is important about the Japanese example is . . . the association of
a narrow income distribution with a public sphere of life which has a real
social content. Instead of a moral and social vacuum mediated only by
market relationships, life beyond the family has a well-developed social
structure.” p. 133
Richard Wilkinson. Unhealthy Societies (1996)
What if we consider the specific components of the
rise in Japanese life expectancy?
Trends in Stomach Cancer, 1950-1995
0
10
20
30
40
50
60
70
1950 1960 1970 1980 1990 1995
CubaUKSweden
Denmark
Finland
Spain
Poland
Ireland
Portugal
Japan
Rat
e pe
r 100
,000
As % of All Cancers1950 1995
Japan 59% 21% Finland 37% 7%Sweden 26% 5%UK 18% 6%USA 14% 3%
* Stomach cancer was the leading single cause of cancer mortality in all these countries in 1950
USA
0102030405060708090
10018
7218
7818
8418
9018
9619
0219
0819
1419
2019
2619
3219
3819
44
(%)
male
female
1920
Primary School Attendance Rate1900
T HasegawaNIHSM,Japan
LE in 1870s35 years
0
50
100
150
200
250
300
350
1899
1904
1909
1914
1919
1924
1929
1934
1939
1944
1949
1954
1959
1964
1969
1974
1979
1984
1989
1994
(NMR)
(IMR)
(U5MR)
(Rat
e /1
,000)
1920
C T HasegawaNIHSM,Japan
Vital Statistics
20th Century Trends in Infant, Neonatal and Under 5 Mortality, Japan
(%)
(IMR/1,000)
Rate of Primaryeducation of mother
IMR
0
10
20
30
40
50
60
70
80
90
1001899
1903
1907
1911
1915
1919
1923
1927
1931
1935
1939
1943
0
20
40
60
80
100
120
140
160
180
2001920
C T HasegawaNIHSM,Japan
Infant Mortality and Education of Women, Japan 1899-1943
Infant mortality 1921-23 against stomach cancer mortality 1991-93for men aged 65-74 in 27 countries
0 50 100 150 200 2500
50
100
150
200
250
CANADA
CHILE
USA
JAPAN
AUSTRIA
BELGIUM
BULGARIACZECHOSL
DEN
FINLAND
FRAGREECE
HUNGARY
IRELANDITALY
NETHLANDSNORWAY
POLANDPORTUGAL
ROMANIA
RUSSIA
SPAIN
SWESWITZ
UK
AUSTRALIANZ
Stom
ach
canc
er m
orta
lity
rate
per
100
K, 1
991-
93
Infant mortality rate per 1000, 1921-23 Leon and Davey Smith, BMJ (2000)
Trends in Stomach Cancer, 1950-1995
0
10
20
30
40
50
60
70
1950 1960 1970 1980 1990 1995
CubaUKSweden
Denmark
Finland
Spain
Poland
Ireland
Portugal
Japan
Rat
e pe
r 100
,000
* Stomach cancer was the leading single cause of cancer mortality in all these countries in 1950
USA
Russia
Stomach cancer
begins to decline
in cohorts born
around the turn
of the century
• By 1900 - Systematic investments in education of women,
improvements in urban living, sanitary conditions especially
for children less infectious burden
• 1920s - The children of those women were the first
generations to experience lower infant mortality – a marker
for the health generating potential of early life environment
• Stomach cancer and hemorrhagic stroke - both causes
dependent on early life conditions / infection - decline
enormously after WW2
Japanese Life Expectancy - The Real Story?
0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0
1-4
5-9
10-14
15-19
20-24
25-29
30-34
35-39
40-44
45-49
50-54
55-59
60-64
65-69
70-74
75-79
80-84
85-89
90-94
95-99
other
suicide
accidents
perinatalperiodrespiratorydiseaseinfectiousdiseasecerebrovasculardiseaseIschemic heart diseasecancer
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0
1-4
5-9
10-14
15-19
20-24
25-29
30-34
35-39
40-44
45-49
50-54
55-59
60-64
65-69
70-74
75-79
80-84
85-89
90-94
95-99
Prolongation of Life Expectancy1965-1985Age & Disease Group by Polard’s Method
Male Female
C T HasegawaNIHSM,Japan
Hemorrhagic Stroke
Japanese Life Expectancy - The Real Story?
• Enormous declines in hemorrhagic stroke and stomach
cancer, but Japan also avoids the “epidemic” of CHD
experienced in the West and Soviet Union
• The huge economic advances post WW2 are NOT
converted into increases in CHD as in the West – despite
high smoking rates post WW2 – low fat diet?
• But the future looks less bright as LE will likely stall or may
actually decline as lung cancer rates climb rapidly – 40 to 50
years after smoking taken up in great numbers
• Levels of social cohesion may well have been important in
facilitating these massive social changes but they were not
directly responsible for these unprecedented improvements
in population health in Japan - due to ↓ stomach cancer and
hemorrhagic stroke
• There were more specific social investments in the health
of women and children and urban sanitation
• Nor is it likely that these investments were rooted in ideas
about improving social equity
• More likely based on paternalistic views of the importance
of good maternal-child health conditions to future
generations and how that would feed into the aspirations of
the Japanese imperial state
• Whatever the motivation, the specific social investments in
women and children have helped deliver the highest life
expectancy on the planet, despite on-going gender
inequities in Japan
• No universal pathway to good population health?
• Should good population health trump other social values?
Factors - like social capital/cohesion and stress - that
may seem plausible (both biologically and socially) in the
cross-section sometimes fail to generate less support
a) when longer term trends in exposures and outcomes
are studied
500.00
700.00
900.00
1,100.00
1,300.00
1,500.00
1,700.00
1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 199835
37
39
41
43
45
47
49
Income inequality (gini) and sex-specific age-adjusted all-cause mortalityUSA, 1968 - 1998
Male Mortality
Female Mortality
Income Inequality
All-c
ause
mor
talit
y pe
r 100
,000
GiniC
oefficient
Lynch and Davey Smith (2003)
700.00
800.00
900.00
1,000.00
1,100.00
1,200.00
1,300.00
1,400.00
1,500.00
1,600.00
1,700.00
1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 199840.0
45.0
50.0
55.0
60.0
65.0
70.0
Race-specific Voting Participation in Presidential Elections and age-adjusted, all-cause mortality, USA, 1968 - 1998
Year
All-c
ause
mor
talit
y pe
r 100
,000
% Voting in Presidential elections
Black Mortality
White MortalityBlack Voting
White Voting
and/or
b) when the specific patho-physiology of the disease in
question is more fully understood.
Disease that have historically been linked to stress processes
• Cholera
• Pellagra
• Beri Beri
• Down’s syndrome
• Scurvy
• Yellow fever
• Typhoid
• Asthma
• Peptic ulcer
The mid-20th Century Epidemic of Peptic Ulcer
Susser and Stein. Lancet (1962)
Peptic Ulcer and Stressful Life Events
MenAny stressful events, OR = 20.3
WomenAny stressful events, OR = 21.5
Davies and Wilson. Lancet, Dec 11th 1937.
“During convalescence the patient should be given a
simple exposition of peptic ulcer. A clear understanding
of the need for maintaining a calm outlook on life, and of
the necessity for not exceeding his natural “tempo” by
accepting too much work or responsibility, will be much
more valuable than routine medication. The patient has
got to live with his ulcer-forming tendency and it is
essential to give him all the information at our disposal.”
F. Avery Jones. Progress in Clinical Medicine, 1948
“The diet should be served in as colourful and attractive a
way as possible. A bland insipid colourless diet may cause
less psychic flow of gastric juice, but it makes the patient
depressed and irritable. The sustained resentment is more
harmful than the increased psychic secretion.”
F. Avery Jones. In: Progress in Clinical Medicine, 1948
“That there has been a true decline in incidence is the most
interesting possibility. This would suggest to anyone in
sympathy with ‘psychosomatic’ theories … that the type of
personality disposed to the disease is less common –
unfortunately not a testable proposition; [or] that the
environment is less of a strain – which is scarcely
conceivable”.
Jerry Morris. Uses of Epidemiology, 3rd Edition, 1975.
What causes peptic ulcer?
“ … despite the common folk knowledge about the
emotional causes of ulcer, a major textbook of
pathology currently concludes that the origins of
peptic ulcers are ‘enigmas wrapped in mystery’ “.
Sterling and Eyer. Biological basis of stress related mortality.
Soc Sci Med 1981; 15E: 3-42.
The evidence “. . . does not exclude the possibility
that a major single causal factor awaits discovery”
Mervyn Susser. Journal of Chronic Disease
1967;20:435-456
What causes peptic ulcer?
Helicobacter PyloriHelicobacter Pylori
“H pylori eradication, without altering acid output, will
become the mainstay of duodenal ulcer treatment
because it cures the disease.”
Rauws and Tytgat. Lancet 1990;335:1233-5
Part 4. What are some of the lessons for better
understanding social disparities in health?
- Considering heterogeneity in our explanations
The Standard View of Socioeconomic Health Disparities
“Socioeconomic status is a strong and consistent predictor of mortality
and morbidity. Individuals lower in the SES hierarchy suffer
disproportionately from almost every disease … This association is found
with each of the key components of SES: income, education and
occupational status.”
Adler, et. Al., JAMA (1993) p. 3140
“Throughout history SES has been linked to health. Individuals higher in
the socioeconomic hierarchy typically enjoy better health that do those
below; SES differences are found for rates of mortality and morbidity
from almost every disease and condition.”
Adler, et. al., Am Psychologist (1994) p. 15
This sort of mind-set has encouraged a focus on universal processes
that make the disadvantaged “generally susceptible” to all sorts of poor
health conditions
• John Cassell’s – the “father” of US social epidemiology - ideas on
social support and general susceptibility have been influential
• Ideas on general susceptibility have been combined with later
evidence (e.g., Whitehall studies) on the apparent inability of material
conditions; and traditional risk factors and behaviors to explain
socioeconomic gradients in CHD
For example, in regard to the Whitehall study, it has been
argued that a “ . . . gradient in mortality among civil
servants who are not poor argues for the importance of
psychosocial factors linked to position in the hierarchy.”
Marmot and Bobak. BMJ 2000 (p. 1127).
The Whitehall II Study
“As predicted, the specific psychosocial work characteristic of low
control made an important contribution to the social gradient in
incident CHD in the Whitehall II study.”
“ … work is the major cause of the social gradient in CHD incidence in
this cohort, …”
“Much of the inverse social gradient in CHD incidence can be attributed
to differences in psychosocial work environment.”
Marmot et al. Lancet 1997
Explanations for Health Disparities
• Materialist “objective” conditions & life circumstancesX• Behavioral diet, smoking, exercise, alcoholX
• Psychosocial stress, negative emotions, control, social support
account for the general susceptibility of the
disadvantaged
Perceptions ofDisadvantage
Psychosocial Environment
Assumption -of “general susceptibility”
to poorer health among the disadvantaged
ControlStress
P – N – E – IBiology
Social SupportSocial Capital
Heterogeneity of links between
socioeconomic factors and different outcomes?
Heterogeneity and History
Life Expectancy at Birth - Britain (1540-1901)
Kunitz, (1987)
20
30
40
50
60
70
1541156115811601162116411661168117011721174117611781180118211841186118811901
LE
Total Population
Aristocracy
Social Advantage = LE Equality Social Advantage = LE
“… in early modern Europe rulers and their subjects seem
to have had similarly low levels of life expectancy at birth
(30-35 years) at least before 1700. Some low-income
peasant families had substantially higher levels of life
expectancy at birth than wealthy peers (40-50 years)
although some had lower (20-25 years). … what matters
most to explaining pre-transition mortality is location not
income.”Johansson (1999)
Trends in Death Rates for Selected Occupations, 1861-1911
150
250
350
450
550
650
750
1861 1871 1881 1891 1901 1911
Hotel Servants
Clergy
File MakersPotters
Coal Miners
Farmers
Teachers Doctors?
Railway Clerks
Woods and Williams (1995)
SMR
Education and IHD Mortality, Korea (1995-2000)Census Linkage Study of 15 million individuals
Kang, Lynch, et al (in press)
0
20
40
60
80
100
No Elementary Middle High College
Men35-44
45-54
55-64
Age-
adju
sted
Mor
talit
y R
ate
Heterogeneity of
Race / Ethnic Disparities in Health
Mortality Unrelated to Race, 1979-89
Disease Age strataEstimated
Hazard Disease Age strataEstimated
HazardOral CA 1.26 Oral CA 1.14Colon CA 1.12 Colon CA 0.94Rectal CA 1.21 Rectal CA 0.62Pancreatic CA 1.25 Pancreatic CA 1.34Lung CA 1.03 Bladder CA 1.24Breast CA 0.98 Kidney CA 1.15Ovarian CA 0.91 Leukemia 0.54Bladder CA 0.58 Other lymph CA 1.07Kidney CA 1.24 Hypertension [75+] 1.78Brain CA 1.11 Ischemic HD [35-54] 1.19Leukemia 0.73 Heart Failure [55-74] 1.16Other lymph CA 1.05 Heart Failure [75+] 0.81Hered. & deg. of CNS 0.67 Ill-defined HD [75+] 1.19Ischemic HD [55-74] 1.07 Stroke [75+] 0.83Heart Failure [55-74] 1.40 Art/arterio/cap 0.80Heart Failure [75+] 1.01Ill-defined HD [75+] 0.79Stroke [75+] 0.80Art/arterio/cap 0.98COPD [75+] 0.71Automobile 0.94Other accidental [55+] 0.96
Women Men
Howard et al., Annals of Epidemiology, 2000; 10: 214-223
Mortality Related to Race, With and Without Adjustment for SEP, 1979-89
Disease Age strataUnadjusted
H.R. Adjusted H.R. Disease Age strataUnadjusted
H.R. Adjusted H.R. Infections 2.50 2.33 Infections 2.21 1.87Stomach CA 2.23 2.06 Stomach CA 1.65 1.46Diabetes 2.11 1.70 Lung CA 1.28 1.15Hypertension [55-74] 4.86 4.17 Prostate CA 2.11 2.15Hypertension [75+] 1.71 1.79 Brain CA 0.22 0.24Ischemic HD [35-54] 2.36 1.63 Diabetes 1.48 1.36Ischemic HD [75+] 0.72 0.71 Hered. & deg. CNS 0.42 0.43Pulmonary 2.49 2.19 Hypertension 4.04 3.60Cardiomyopathy 2.59 2.26 Ischemic HD [55-74] 0.85 0.76Ill-defined HD [55-74] 1.76 1.55 Ischemic HD [75+] 0.50 0.49Stroke [55-74] 1.71 1.53 Pulmonary [55-74] 2.55 2.34COPD [55-74] 0.40 0.37 Cardiomyopathy 2.02 1.84Cirrhosis 2.64 2.25 Ill-defined HD [55-74] 2.66 2.26Suicide 0.34 0.37 Stroke [55-74] 1.77 1.51Homicide 4.60 3.28 COPD [55-74] 0.65 0.52Other accidental [25-54] 1.98 1.32 COPD 75+ 0.35 0.33
Cirrhosis 1.72 1.52Automobile 0.78 0.65Suicide 0.49 0.41Homicide 8.18 5.94Other accidental 1.67 1.30
Women Men
Howard et al., Annals of Epidemiology, 2000; 10: 214-223
Hazard Ratio for Black versus White Race for Death Related to Race, 1979-89
Howard et al., Annals of Epidemiology, 2000; 10: 214-223
-80
-60
-40
-20
0
20
40
60
80
0.12
5
0.25
0
0.50
0
1.00
0
2.00
0
4.00
0
8.00
0
16.0
00
COPD (55-74)
Suicide
Ischemic HD (75+)Hypertension (75+)
InfectionsStomach CA
Hypertension (55-74)Stroke (55-74)
PulmonaryCardiomyopathy
Ill-defined HD (55-74) Cirrhosis
DiabetesHomicide
IHD (35-54)
Other Accidental (25-54)
WOMENPe
rcen
t Cha
nge
afte
r adj
ustm
ent f
or S
EP
Hazard Ratio (Unadjusted)
-80
-60
-40
-20
0
20
40
60
80
0.12
5
0.25
0
0.50
0
1.00
0
2.00
0
4.00
0
8.00
0
16.0
00
COPD (55-74)
Suicide
I HD (75+)
Infections
Stomach CA
Hypertension (55-74)
Stroke (55-74)
Pulmonary
CardiomyopathyIll-defined HD (55-74)
CirrhosisDiabetes
Homicide
IHD (55-74)
Other Accidental (25-54)
MEN
Hazard Ratio (Unadjusted)
Automobile
COPD (75+) Prostate CA
Brain CA Hered&Deg of CNS
Lung CA
Howard et al., Annals of Epidemiology, 2000; 10: 214-223
Perc
ent C
hang
e af
ter a
djus
tmen
t for
SEP
Hazard Ratio for Black versus White Race for Death Related to Race, 1979-89
Change in the Disparity in Life Expectancy if Selected Diseases were Eliminated
-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
Hom
icide
Suicide
Motor vehicle accident
Alcohol-related diseases
Rheum
atologic diseases
Liver disease
Renal disease
Diabetes m
ellitus
Lung disease
HIV
disease
Sepsis
Viral hepatitis
Pneumonia
Tuberculosis
Leukemia or lym
phoma
Urinary cancer
Prostate cancer
Cervical cancer
Breast cancer
Lung cancer
Stomach cancer
Esophageal cancer
Uterine or ovarian cancer
Pancreatic cancer
Liver cancer
Colon cancer
Congestive heart failure
Hypertension
Cerebrovascular stroke
Ischemic heart disease
Cha
nge
in D
ispa
rity
in L
ife E
xpec
tanc
y (y
ears
per
per
son)
Wong et al., New England Journal of Medicine, 2002; Vol.347, No.20: 1585-1592
Change racial disparity
HomicideHypertension
HIV35% of the racial disparity
Heterogeneity of
Socioeconomic Disparities in Health
Causes of death and median income of Zip Code area of residence in the men screened for MRFIT: relative risk for $10,000 lower income
RR > 1.50 RR 1.21-1.50 RR 1.00-1.20 RR < 1.00AIDS Infection Aortic aneurysm Blood diseaseDiabetes Coronary Heart Suicide Motor neurone
Disease diseaseRheumatic Stroke Nervous system Flying accidentsHeart Disease diseaseHeart failure Cirrhosis Oesophageal Lymphoma
cancerCOPD Genitourinary Stomach cancer Hodgkin’s disease
diseasePneumonia/ SR Symptoms Pancreatic cancer MelanomaInfluenzaHomicide Accidents Prostate cancer Bone/connective
tissue cancerLung cancer Bladder cancerLiver cancer Kidney cancerColorectal cancer Brain cancer
MyelomaLeukaemia Davey Smith (1996)
Parental Occupation and Relative Risk for Injury, Sweden, 1990-1994
0
0.5
1
1.5
2
High Low Employees Skilled Workers Unskilled Workers
Traffic Injuries Falls
Ages 15-19OR
Engerström, Diderichsen, Laflamme. Injury Prevention (2002)
Mouth/Pharynx ? •Oesophagus • •Stomach • •Liver • •Nasal • •Larynx • •Lung • •Cervix • •
Incidence Mortality
More Common Among Poor
Melanoma • •Breast • •?Colon • •Ovarian • •Testis •? •
Incidence Mortality
More Common Among Rich
Rectum, Pancreatic, Bone, Connective Tissue, Prostate, Bladder, Kidney, Brain, Thyroid, Lymphomas, Leukemia
No Evidence for Socioeconomic Patterns
IARC (1997)
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
10-29
Constance, Int J Cancer (1990)
30-4950-69
70+
Men
AgeAge--specific Melanoma Risk and Census Tract Incomespecific Melanoma Risk and Census Tract IncomeWashington State, 1974Washington State, 1974--8585
HighLow Income
RR
International Heterogeneity in the Magnitude and
Direction of Social Inequalities in Health
Manual vs Non-Manual Health Inequalities, Europe
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2
France
Finland
England
/Wale
sSwed
enIre
land
Spain
Portuga
l
Italy
Switzerla
ndNorw
ayDen
mark
2
7
12
17
22
27
32
KunstKunst, , BMJBMJ (1998)(1998)
RR
: Mor
talit
y (M
an v
sN
on-M
an) Men 45-59Probability of Dying
Rate Ratio: Manual vs Non-manual
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2
France
Finland
England
/Wale
sSwede
n
Irelan
d
Spain
Portugal Ita
lySwitz
erland
Norway
Denmark
KunstKunst, , BMJBMJ (1998)(1998)
RR
Men 45-59Cancer MortalityCancer Mortality
Rate Ratio: Manual vs Non-manual
0.75
0.85
0.95
1.05
1.15
1.25
1.35
1.45
1.55
1.65
1.75
KunstKunst, , BMJBMJ (1998)(1998)
RR
Men 45-59IHD MortalityIHD Mortality
France
Finland
Norway
SwedenEngland/W
ales
Switzerla
nd
Ireland
SpainPortu
gal
Denmark
Italy
Implies differing socioeconomic distributions of the risk factors for IHD across countries
So what have we learned?
This is your quiz
Can You Explain this Health Disparity?
What process generated this health disparity?
0
5
10
15
20
25
30
35
40
45
50
2.8
16
45
Upper Class Middle Class Lower Class
Fata
lity
Rat
e pe
r 100
• Very famous and occurred in 1st decades of 20th C
• This entire population was susceptible
• Despite this, lower class women had 20 times higher
death rate - 45% of lower class women died
• And men died at even higher rates than women
Some Clues
0
5
10
15
20
25
30
35
40
45
50
1st Class 2nd Class 3rd Class
2.8
16
45
Fatality Rates per 100 Women Passengers: HMS Titanic
Class of Travel
How should we understand this health disparity?• Why were poorer women more likely to die?
- Drowning gene?
- More stressed?
- Low self-efficacy?
- Lack of exercise?
- Targetted intervention?
swimming lessonsFata
lity
Rat
e pe
r 100
Nature of theSocial Structure
distribution of specific health
protective resources
Lifeboats
Social disparityin the death rates
Various socioeconomic positions within that structure
Can apply the same idea to understanding this pattern
50
60
70
80
90
100
110
120
1975 1980 1985 1990 1995
Breast Cancer Incidence
0
5
10
15
20
25
30
1975 1980 1985 1990 1995
Breast Cancer Mortality
White Women
Black Women
White Women
Black Women
How does the social structure distribute risk factors for breast cancer incidence?
How does the social structure distribute risk factors for breast cancer mortality?
National Center for Health Statistics, USA (1998)
We need to include heterogeneity in our explanations for social disparities
in health
The direction and extent of social disparities in health differs
• by outcome
• by exposure – education, income
• by place
• over time
This may help make the problem of health disparities seem less intractable
and suggests that social disparities in health are rooted in the nature of
the social structures created by humans and thus amenable to change
Conclusion
• The example of the Russian mortality crisis and the rise in Japanese LE
show how social factors can strongly affect levels and trends in population
health BUT they do so specifically through their links to risk factors for
specific outcomes.
• Applied to health disparities, this suggests we can use the underlying
heterogeneity of associations to suggest more specific links between
socioeconomic markers and race/ethnicity and risk factors and specific
health outcomes
• This may weaken the case for general psychosocial processes in
determining population health or health disparities
Lynch and Davey Smith (2003)
Where do psychosocial factors fit?
• The case for the importance of psychosocial factors to health does not
need to be made on the grounds that they are key determinants in the
etiology of diseases like CVD and cancer, although they may be
etiologically important to some outcomes such as homicide.
• They are also important outcomes in their own right because of their
contributions to quality of life in affluent societies.
• The quantity of life may well be a relatively unimportant contributor to
the quality of life in affluent societies and this means psychosocial
phenomena are important as population health outcomes.
Lynch and Davey Smith. IJE (2003)
Thank you