Gait speed and physical exercise in people with dementia

Annika Toots

Community Medicine and Rehabilitation,

Physiotherapy and Geriatric Medicine

Umeå 2016

Responsible publisher under Swedish law: the Dean of the Medical Faculty

This work is protected by the Swedish Copyright Legislation (Act 1960:729)

ISBN: 978-91-7601-616-9

ISSN: 0346-6612-1866

Elektronisk version tillgänglig på http://umu.diva-portal.org/

Printed by Print & Media, Umeå University

Umeå, Sweden 2016

To my beloved family

i

CONTENTS ABSTRACT III

SVENSK SAMMANFATTNING (SUMMARY IN SWEDISH) V

ABBREVIATIONS VII

ORIGINAL PAPERS VIII

INTRODUCTION 1 Demographic changes 1 Consequences of ageing 1 Physical activity 1 Nursing homes 2

GAIT SPEED 2 Gait speed and health 3 Walking aids 3

DEMENTIA 4 Incidence, Prevalence and Costs 4 Risk factors 5 Diagnosis and Symptoms 5 Physical function 5

PHYSICAL EXERCISE 6 Effects on physical function 6 Effects on cognitive function 7 Exercise in nursing homes 8 Exercise in dementia 8

RATIONALE 10

AIMS 11

METHODS 12 SETTING 12

The GERDA study 12 The UMDEX study 13

PARTICIPANTS 13 ETHICS 14 STUDY DESIGN 16 OUTCOME AND TARGET MEASUREMENTS 16

Mortality 17 Gait speed 17 Dependence in ADLs 18 Balance 19 Cognitive function 19

ii

BASELINE MEASUREMENTS 19 RANDOMISATION 20 INTERVENTION 20

Exercise activity 20 Attention control activity 21

ADHERENCE, EXERCISE INTENSITY AND ADVERSE EVENTS 22 STATISTICAL ANALYSES 22

Paper I 22 Papers II-IV 23

RESULTS 25 OUTCOMES 28

Gait speed and Mortality 28 Dependence in ADLs 30 Balance 32 Gait speed 36 Cognitive function 36

DISCUSSION 41 GAIT SPEED IN VERY OLD AGE AND IN DEMENTIA 41 GAIT SPEED PREDICTS MORTALITY IN VERY OLD PEOPLE 42 EXERCISE EFFECTS ON DEPENDENCE IN ADLs 43 THE HIFE PROGRAM AND EFFECTS ON BALANCE 43 DIFFERENCES BETWEEN DEMENTIA TYPES 44 WALKING AIDS AND EXERCISE EFFECTS ON GAIT SPEED 45 EXERCISE EFFECTS ON COGNITIVE FUNCTION 46 METHODOLOGICAL CONSIDERATIONS 47 ETHICAL CONSIDERATIONS 50 CLINICAL IMPLICATIONS 51 IMPLICATIONS FOR FUTURE RESEARCH 53

CONCLUSIONS 54

ACKNOWLEDGEMENTS 55

REFERENCES 57

PAPERS I-IV

LIST OF DISSERTATIONS

iii

ABSTRACT

The aim of the thesis was to investigate the importance of physical function for survival in very old people, and furthermore, whether physical exercise could influence physical function, cognitive function, and dependence in activities of daily living (ADLs) in older people with dementia living in nursing homes.

The world’s population is ageing. Given the age-related increase in chronic disease such as dementia and compounded by physical inactivity, the prevalence in need for assistance and care in daily activities in older people is expected to increase in the near future. Gait speed, a measure of physical function, has been shown to be associated with health and survival. However, studies of the oldest people in the population, including those dependent in ADLs, living in nursing homes and with dementia, are few. Moreover, in people with dementia physical exercise may improve physical and cognitive function and reduce dependence in ADLs. Further large studies with high methodological quality and with designs incorporating attention control groups are needed in this population. In addition, no study has compared exercise effects between dementia types.

The association between gait speed and survival was investigated in a population based cohort study of 772 people aged 85 years and over. Usual gait speed was assessed over 2.4 metres and mortality followed for five years. Cox proportional hazard regression models adjusted for potential confounders were used in analyses. Effects of physical exercise in people with dementia were investigated in a randomised controlled trial that included 186 participants with various dementia types living in nursing homes. Participants were allocated to the High-Intensity Functional Exercise (HIFE) program or a seated control activity, which both lasted 45 minutes and held five times fortnightly for four months. Dependence in ADLs was assessed with Functional Independence Measure and Barthel ADL Index, and balance with Berg Balance Scale. Usual gait speed was evaluated over 4.0 metres in two tests; first using habitual walking aid if any, and thereafter without walking aid and with minimum living support. Global cognitive function was assessed using the Mini-Mental State Examination, the Alzheimer’s Disease Assessment Scale-Cognitive subscale, and executive function using Verbal fluency. Blinded testers performed assessments at baseline, four (directly after intervention completion) and seven months. Analyses used linear mixed models in agreement with the intention-to-treat principle.

Gait speed was found to be an independent predictor of five-year all-cause mortality, where inability to complete the gait test or a gait speed below 0.5

iv

meters per second (m/s) was associated with higher mortality risk. In analyses of exercise effects on ADLs there was no difference between groups in the complete sample. Interaction analyses showed a difference in exercise effect according to dementia type at seven months. Positive between-group exercise effects were found for dependence in ADLs in participants with non-Alzheimer’s type of dementia (non-AD) at four and seven months. In balance, a difference between groups was found at four but not at seven months in the complete sample, and interaction analyses indicated a difference in effect according to dementia type at four and seven months. Positive between-group exercise effects were found in participants with non-AD. No difference between groups in gait speed was found in the complete sample, where the majority habitually walked with a walking aid. In interaction analyses exercise effects differed according to walking aid use. Positive between-group exercise effects in gait speed were found in participants that walked unsupported at four and seven months. No difference between groups in cognitive function was found in the complete sample. The effects of exercise on gait speed and cognitive function did not differ according to sex, cognitive level, or dementia type.

In conclusion, among people aged 85 or older, including those dependent in ADLs and with dementia, gait speed seems to be a useful clinical indicator of health status. Inability to complete the gait test or a gait speed below 0.5 m/s appears to be associated with higher five-year mortality risk. In older people with mild to moderate dementia living in nursing homes, a four-month high-intensity functional exercise program appeared to attenuate loss of dependence in ADLs and improve balance, albeit only in participants with non-AD type of dementia. Further studies are needed to validate this result. Furthermore, exercise had positive effects on gait speed when tested unsupported, in contrast to when walking aids or minimum support were used. The result implies that the use of walking aids in the gait speed test may conceal exercise effects. The exercise program had no superior effects on global cognition or executive function when compared with an attention control activity. This thesis suggests that, in older people with dementia, exercise effects on physical function rather than cognitive function may explain effects on dependence in ADLs.

v

SVENSK SAMMANFATTNING (SUMMARY IN SWEDISH)

Världens befolkning blir allt äldre. Många kroniska sjukdomar så som exempelvis demenssjukdom är åldersrelaterade, och påverkas negativt av fysisk inaktivitet. Därför förväntas behovet av hjälp i Aktiviteter i det Dagliga Livet (ADL) öka i samband med att befolkningsmängden av äldre växer. Gånghastighet är ett mått på fysisk funktion, som har visat sig associera med hälsa och överlevnad. Få studier har dock undersökt de allra äldsta personerna i befolkningen, samt de som är ADL-beroende, har demenssjukdom och som bor i särskilt boende. Vissa studier tyder på att fysisk träning kan förbättra fysisk och kognitiv funktion samt minska ADL-beroende bland personer med demens. Det behövs dock fler stora studier av hög metodologisk kvalitet. Dessutom har ingen studie jämfört om effekter av träning skiljer sig mellan olika typer av demens. Syftet med denna avhandling var dels att undersöka betydelsen av fysisk funktion för överlevnad, bland personer som är mycket gamla, dels om fysisk träning kan påverka fysisk och kognitiv funktion, samt ADL-beroende bland äldre personer med demens som bor i särskilt boende.

Sambandet mellan gånghastighet och överlevnad undersöktes i en populationsbaserad kohortstudie av 772 personer i åldern 85 år och äldre. Självvald gånghastighet mättes över 2,4 meter och dödlighet följdes under 5 års tid. En överlevnadsanalys, där man tog hänsyn till faktorer vanliga hos den äldre befolkningen, utfördes. Effekter av fysisk träning hos personer med demens undersöktes i en randomiserad kontrollerad studie, omfattandes 186 deltagare, med olika demenstyper i särskilt boende. Deltagarna lottades till att delta i ett högintensivt funktionellt träningsprogram (HIFE programmet) eller en sittande kontrollaktivitet. Aktiviteterna pågick i fyra månader, två till tre gånger per vecka, 45 minuter per tillfälle. Ledare var fysioterapeuter respektive arbetsterapeut/arbetsterapibiträde. Effekter på ADL-beroende skattades med ”Functional Independence Measure” och ”Barthel ADL Index”, och balans med Bergs Balansskala. Självvald gånghastighet mättes över 4,0 m i två tester, först med stöd av gånghjälpmedel om sådant användes, och därefter utan gånghjälpmedel och med minsta möjliga stöd av en till två personer. Effekter på global kognitiv funktion skattades med Mini Mental Test, ”Alzheimers Disease Assessment Scale-Cognitiv subscale”, och exekutiv funktion med Verbalt flöde. Skattningar utfördes vid baslinjen, efter fyra (direkt efter interventionens slut) samt efter sju månader av testare ovetandes av deltagares grupptillhörighet. Analyser genomfördes utifrån ”intention-to-treat” principen, vilket innebär att all tillgänglig data analyserades utifrån ursprunglig grupptillhörighet och oavsett närvaro i aktiviteter.

vi

Gånghastighet var en prediktiv faktor för dödlighet, och sambandet var oberoende av flertalet sjukdomar och läkemedel vanliga bland äldre personer. En oförmåga att utföra gångtestet eller en gånghastighet under 0,5 m/s var associerad med högre risk för död inom fem år. Träningsprogrammet hade ingen effekt på ADL-beroende i hela gruppen. Interaktionsanalyser visade på skillnader i träningseffekt beroende på demenstyp. Deltagare med icke-Alzheimers typ av demens (icke-AD) försämrades mindre jämfört med kontrollgruppen. Träning hade en positiv effekt på balans vid fyra månader men inte vid sju månader i hela gruppen. Även här skilde sig träningseffekterna beroende på demenstyp i interaktionanalyser. Deltagare med icke-AD hade förbättrad balans jämfört med kontrollgruppen. Majoriteten av deltagarna i studien förflyttade sig vanligtvis med ett gånghjälpmedel. Träning hade ingen effekt på gånghastighet i hela gruppen. I interaktionsanalyser skilde sig träningseffekten beroende på stöd i testet, där en träningseffekt fanns bland deltagare som gick utan stöd, jämfört med de som gick med ett gånghjälpmedel eller levande stöd. Ingen träningseffekt på kognitiv funktion påvisades i hela gruppen. Träningseffekterna på gånghastighet och kognitiv funktion skilde sig heller inte beroende på typ av demens.

Sammanfattningsvis tyder resultaten från denna avhandling på att bland personer som är mycket gamla, inklusive de med ADL-beroende, demens, och i särskilt boende, kan gånghastighet vara en användbar klinisk indikator på hälsotillstånd. I denna population verkar en gånghastighet på 0,5 m/s vara en viktig gräns för överlevnad. Hos äldre personer med mild till måttlig demens i särskilt boende, förefaller ett högintensivt funktionellt träningsprogram i fyra månader kunna påverka ADL-beroende och balans, än om bara för personer med icke-AD demens. Ytterligare studier behövs för att bekräfta detta resultat. Vidare tycks träning ha positiva effekter på gånghastighet när det kan testas utan stöd. Gånghjälpmedel verkar kunna dölja träningseffekter. Träning tycks inte ha en fördelaktig effekt på global kognition eller exekutiv funktion jämfört med en kontrollaktivitet. Sammantaget tyder dessa resultat på att, bland personer med demens i särskilt boende, kan träningseffekter på fysisk funktion snarare än kognitiv funktion förklara effekter på ADL beroende.

vii

ABBREVIATIONS AD Alzheimer’s disease

ADL Activities of daily living

ADAS-Cog Alzheimer’s Disease Assessment Scale-Cognitive subscale

BBS Berg Balance Scale

BPSD Behavioural and Psychological Symptoms of Dementia

DSM-IV-TR Diagnostic and Statistical Manual of Mental Disorders, 4th Ed. Text revision.

ES Effect size

FIM Functional Independence Measure

FOPANU Frail Older People – Activity and Nutrition study in Umeå

GDS Geriatric Depression Scale

GS Gait speed with habitual walking aid if any

GSnoWA Gait speed with no walking aid and with minimum support

HIFE High-Intensity Functional Exercise

MMSE Mini-Mental State Examination

M/s Meters per second

Non-AD Non-Alzheimer’s type of dementia

OT Occupational Therapist

VF Verbal fluency

PT Physiotherapist

RCT Randomised Controlled Trial

RM Repetition maximum

SD Standard deviation

SE Standard error

UMDEX Umeå Dementia and Exercise study

viii

ORIGINAL PAPERS

This thesis is based on the following papers, which will be referred to in the text by their roman numerals:

I. Annika Toots, Erik Rosendahl, Lillemor Lundin-Olsson, Peter Nordström, Yngve Gustafson, Håkan Littbrand. Usual gait speed independently predicts mortality in very old people: a population-based study. The Journal of American Medical Directors Association. 2013; 14: 529.e1-529.e6.

II. Annika Toots, Håkan Littbrand, Nina Lindelöf, Robert Wiklund, Henrik Holmberg, Peter Nordström, Lillemor Lundin-Olsson, Yngve Gustafson, Erik Rosendahl. Effects of a high-intensity functional exercise program on dependency in activities of daily living and balance in older people with dementia. The Journal of the American Geriatrics Society. 2016; 64:55-64.

III. Annika Toots, Håkan Littbrand, Henrik Holmberg, Peter Nordström, Lillemor Lundin-Olsson, Yngve Gustafson, Erik Rosendahl. Walking aids moderate exercise effects on gait speed in people with dementia; a randomized controlled trial. The Journal of American Medical Directors Association. 2016 (Epub ahead of print).

IV. Annika Toots, Håkan Littbrand, Gustaf Boström, Carl Hörnsten, Henrik Holmberg, Lillemor Lundin-Olsson, Nina Lindelöf, Peter Nordström, Yngve Gustafson, Erik Rosendahl. Effects of a high-intensity exercise program on cognitive function in people with dementia. Submitted.

The original articles are reprinted in this thesis by kind permission of respective publisher.

1

INTRODUCTION

Worldwide, the population of older people is expected to increase in number and proportion in the near future. Older age is associated with increased risk for acute or chronic disease or disorders, such as dementia, and also increased risk of physical and cognitive impairments. Furthermore, health-related behaviour such as physical inactivity compounds risks. Subsequently, a growth in the number of people with need of care and assistance in activities of daily living (ADL) are expected in the future. Population ageing will have major social and economic consequences. In order to reduce individual suffering and provide adequate care, health promotion, early disease detection, and disease modifying interventions may become increasingly important.

Demographic changes The world’s population is ageing. As a result of decreasing mortality and fertility rates, the older population (age 60 years or over) is expected to more than double over the next four decades, from 841 million to more than two billion in year 2050; a proportionally increase from 12% to 21% of the world population.1 Meanwhile, the older population is itself ageing. Within the older population, future projections suggest a proportional increase of the oldest individuals (aged 80 years or older) from 14% to 19%.1 Similar demographic trends are expected in Sweden, with an increase in older people (aged 65 years or over) from 1. 9 million in 2015 to 2.9 million in 2050, with the oldest people (aged 85 years and over) estimated to contribute most to the increase.2

Consequences of ageing Ageing is a gradual process ultimately resulting in the functional decline of most organ systems in the body, including circulatory, nervous, and musculoskeletal systems. This biological process proceeds with great individual variation. Nevertheless, older age is associated with deterioration of health status, including reduced cognitive function and impairments in physical functions such as balance and gait.3-5 In older age there is also a greater risk of developing chronic disease including cardiovascular disease, depression, and dementia, and also the co-occurrence of two or more chronic diseases i.e. multimorbidity.3,6 Consequently, dependence on care and assistance to carry out ADLs, such as, dressing, feeding, personal hygiene, or mobility becomes more common with older age.3

Physical activity Physical inactivity increases the risk of morbidity and mortality in older age. In 2010, 55% of older people worldwide did not meet recommended levels of physical activity, i.e. 150 minutes of moderate-intensity physical activity per

2

week, or equivalent, and women are more physically inactive than men.7 Physical inactivity has a major influence on health, causing an estimated 6-10% of chronic disease.8 Conversely, physical activity has been associated with many health benefits, both in people with and without chronic disease. Evidence suggests that compared with less active individuals, older women and men who are more active have reduced mortality, coronary heart disease, high blood pressure, stroke, type 2 diabetes, colon cancer, and breast cancer.9 In addition, more active individuals have higher levels of cardiovascular and muscular fitness, and cognitive function, and a lower risk of falls.8,9

Nursing homes The number of older people in need of care and assistance in ADLs is predicted to increase in line with population ageing.1 Care may be provided in the community or institutionalised settings such as nursing homes. Nursing homes has been defined as a domestic care facility that provides 24-hour functional support and care, short- or long-term, or palliative care, and staffed by trained or untrained health professionals.10 In this thesis the two terms residential care facilities and nursing homes will be used interchangeably. Although approaches to care provision vary between countries, an increase in the demand of care for older people is expected as the population ages.11 In Sweden, the twofold increase in older people between 2010 and 2050 has been projected to result in a care demand growth of 1.3-1.4% per annum, conceivably totalling a 70% increase in cost of care by 2050,12 of which nursing home care constitutes the largest proportion.13 However, a shift towards care provided in the community appears to have occurred and reduced the number and proportion of older people living in nursing homes.13 In nursing homes, the prevalence of physical and cognitive impairment and dependence in ADLs is high.14-16 In addition, behavioural and psychological symptoms, multimorbidity, and polymedication are frequent,15,17,18 as are social isolation and physical inactivity.19,20

GAIT SPEED Gait speed slows in older age.21,22 Walking is inherently complex and dependent on the function of several physiological systems e.g. musculoskeletal, neurological or circulatory systems, which are affected by biological ageing and disease.23 Gait dysfunction may result from acute disease or injury e.g. stroke or hip fracture, or progress insidiously with chronic disease.23,24 In addition, deterioration in muscle strength, balance, or cognitive function, as a consequence of age-related changes or physical inactivity may influence gait.23,25

3

Gait speed and health Gait speed is a powerful predictor of several health related adverse outcomes.26,27 In longitudinal observational studies of older people, lower usual gait speed has been shown to be independently associated with increased risk for mortality25,28 and also disability,29-31 physical inactivity,32,33 falls34 and hospitalization.35-37 In addition, lower gait speed has been shown to predict incident dementia,38,39 predominantly non-Alzheimer’s type of dementia (non-AD) compared with Alzheimer’s disease (AD).40,41 Gait speed has been propositioned as a “6th vital sign”, amongst other important signs of physiological health such as respiratory rate, and blood pressure.42,43 Tests of gait speed may be used in clinic or research for screening purposes, as part of differential diagnostic tests, or to assess change over time.43,44

Most cohort studies investigating gait speed in association with survival have focused on community dwelling, well-functioning samples of older individuals with mean age younger than 80 years.25,26 Previously studied samples may therefore not be representative of the older population as a whole, which limits generalisation. There is a recognised need for validation of the results in a population of very old individuals, with dementia or disability; characteristics common in older people likely to be encountered in clinical practice.25,26,44 Many age-related factors such as dependency in ADLs, cognitive impairment, and multi-morbidity are associated with both lower gait speed and increased risk for mortality. Consequently, it is important to investigate whether gait speed can also predict mortality after comprehensive adjustments for potential confounders in very old people.

Walking aids To compensate for impairments in physical function, older adults often use walking aids such as canes or rollators. The use of walking aids becomes more prevalent in older ages.45,46 In Sweden, a national survey from 2004/2005 showed that in the ages 75-84 years, 38% of women and 22% of men used mobility devices e.g. cane, rollator or wheelchair to move, while in men and women aged 85 years and older the corresponding proportion was 71% and 56% respectively.47 Over the last decade an increase of 50% in the use of walking aids by older people has been reported in the United States.46 There is no available nationwide data to show if an equivalent increase in Sweden has occurred. Nevertheless, a study in nursing homes within the county of Västerbotten in northern Sweden indicated that walking aid use in year 2000 was not significantly different from 2007 (50.6% and 51.9%, respectively).48 Walking aids improve mobility by alleviating balance or pain.49,50 Conversely, some older adults may have problems using their walking aids, due to for example cognitive impairment, particularly under more demanding conditions such as turning,51 and the effectiveness of walking aid use in the prevention of

4

falls is not determined.46,49,52 Gait speed measured using a walking aid could limit detection of gait deficits and thus reduce responsiveness of the test.53 However, few studies, if any, have investigated if exercise effects differ according to walking aid use in people with dementia living in nursing homes.

DEMENTIA

Dementia is the umbrella term used to describe a multitude of diseases and disorders that develop as a result of damage or death of nerve cells in the brain. AD is the most common type of dementia, and accounts for approximately 60% to 80% of cases, followed by Vascular dementia that accounts for about 10%.54 However, in older ages Vascular dementia becomes increasingly prevalent.54 The precise changes that trigger the development of AD remains unknown, but has been linked with abnormal protein deposits and formations. Vascular dementia usually results from infarcts or bleeding in the brain caused by blockage or damage to blood vessels, either related to stroke or small vessel disease. Other types of dementia include dementia with Lewy Bodies, Frontotemporal dementia, and Parkinson’s disease dementia. When symptoms of two or more types of dementias coexist the individual is considered to have Mixed dementia.

Worldwide, dementia is a leading cause of disability and dependence in ADLs in older people.55 The need for care and assistance in ADLs affects burden of care and quality of life.55-57 In addition to cognitive decline, impairments to balance and gait may contribute to a reduced ability to carry out ADLs.29,30,36

Incidence, Prevalence and Costs The incidence of dementia increases exponentially with higher age, and is slightly higher in women than in men.55 Each year 9.9 million new cases of dementia worldwide are expected, which implies one new case every 3.2 seconds.54 In 2015 an estimated 46.8 million people worldwide were living with dementia, and the number is projected increase to 74.7 million in 2030 and 131.5 million in 2050.54 In Sweden, 149 000 (8%) of people aged 65 years and over, and nearly half of all aged 90 years and over were estimated to have dementia in 2012.58 Costs to society for the care and treatment of people with dementia are high and will increase dramatically in the future. In Sweden, nearly half (42%) of all people with dementia live in nursing homes.58 The total cost for care and treatment of persons with dementia in Sweden has been estimated to 63 billion Swedish Kronor per year, with the largest proportion of the cost being care in nursing homes.58

5

Risk factors The greatest risk factor for developing dementia is older age, and further risk factors include family history, genetics, lower education, and social and cognitive inactivity.54 In addition, many potentially modifiable risk factors for cardiovascular disease are also associated with incident dementia, including smoking, obesity, diabetes mellitus, depression, midlife hypertension, diet, and physical inactivity.54,59 Indeed, about one third of all AD cases worldwide have been estimated to be attributable to potentially modifiable risk factors, of which physical inactivity is the prominent one in Europe and the USA.60 The same estimates can be applied to most common types of dementia.60 Assuming causality, effective methods targeting reduction of physical inactivity could therefore potentially reduce incidence of dementia.60,61

Diagnosis and Symptoms The dementia diagnosis is usually set according to criteria in the Diagnostic and Statistical Manual of Mental Disorders (DSM).62 While the most recent edition (DSM-5) classifies dementia disorders under the term major neurocognitive disorder, in this thesis the term dementia from DSM-IV will be used. Features of dementia include a progressive decline in cognition, including deficit to memory, and also language, recognition, motor or executive processes that are sufficient to disrupt social or occupational function, including ADLs.62 Distinct symptom patterns are associated with different types of dementia.54 For example, early symptoms of Vascular dementia are likely to be executive dysfunction rather than the memory loss prominent in early AD. With all dementia types, a wide range of symptoms may present as the disease progresses, including disorientation, confusion, behavioural and psychological symptoms in dementia (BPSDs), as well as speaking, swallowing and gait dysfunction.

Physical function People with dementia often experience disturbances to balance and gait,63-69 including lower gait speed, interrupted gait, increased step length or width variability, and stance instability.70-74 Although gait appears to be particularly affected in people with dementia types such as Vascular dementia, Parkinson’s disease dementia, and dementia with Lewy Bodies,41,65,66 cross-sectional studies show that gait dysfunctions are also present in AD.68,69,71,72,74 Nevertheless, gait becomes more disturbed in advanced stages,66,67 particularly in individuals with dementia types other than AD.67 Gait is considered a complex activity comprising input from higher-level cognitive areas,75-78 with gait deterioration in dementia particularly linked with impairments in executive function.79 Such areas in the brain are involved in activities such as planning and initiating complex behaviour, and dysfunction could therefore result in reduced ability to execute motor activities. The exacerbation of gait disturbance under dual task conditions, whereby a cognitive task e.g. naming animals or counting is added

6

to the motor task,70,72 may result from problems with executive function, including attention control in people with dementia.73

Impaired gait and balance are also associated with physical inactivity32,33 and falls.80,81 Physical inactivity has a major influence on health.8 In dementia symptoms such as impaired gait, reduced initiative, as well as apathy and delirium can increase physical inactivity.82 Conversely, physical inactivity may also exacerbate the gait and balance dysfunction. Moreover, deficits in attention control, as measured by a decline in dual-task performance, have been linked to falls.83,84 The risk of falls and fractures are more than twice as high for older people with dementia compared with older people without cognitive impairment,63,85 and hip fracture may lead to manifest physical impairments or even death.86

PHYSICAL EXERCISE

Physical activity has been defined as body movement produced by skeletal muscle contraction that increases energy expenditure, while exercise as planned and structured, which includes repetitive movement to improve or maintain aspects of physical fitness.87 To promote physical activity and prevent consequences of aging and chronic disease, guidelines on recommended levels of physical activity and exercise, including aerobic, strengthening, and balance exercise for older people and for people in nursing homes has been published.9,81,87,88

Effects on physical function To reduce risk of functional impairment, older people are recommended to perform aerobic (endurance) exercise such as walking or cycling 30-60 minutes per day at moderate intensity up to 150-300 minutes per week, or 20-30 minutes per day at high intensity up to 75-150 minutes per week, in bouts of at least ten minutes.87,89 Additional health benefits occur at the higher dose.89 In addition, exercise programs that combine aerobic exercise and resistance exercise may have additional effects on functional outcomes, compared with either exercise type alone.87,89

Progressive resistance exercise has been shown to improve muscle strength89-92 and benefit gait speed,90,93,94 and dependence in ADLs in older people.90 In addition, progressive resistance exercise may positively influence aerobic capacity.90,95 According to recommendations, strengthening exercise involving major muscle groups, should be performed at moderate to high intensity on two or more days per week.87,89,90 Intensity is often described relative to the maximum repetitions an exercise can be performed against resistance before

7

muscle fatigue, either as a percentage of an individuals’s one repetition maximum (RM)96 or as an absolute number of RMs. Exercise at high intensity appears to have greater effect on strength than lower intensity,87,90,97 but may require supervision.90 In addition, effects of exercise are related to the principle of specificity. For optimal and specific effects on function in daily activities, such as rising up from a chair, walking, or climbing stairs, the strength exercises should be performed in weight-bearing positions, similar to the activity or task intended to improve or components thereof.98-102

In older people, balance exercises appears to have positive effects on balance103,104 and on reduction of falls.87,105 Furthermore, functional activities typically involve tasks that require both muscle strength and balance, and balance exercise may thus be indicated in general older populations, and not only in populations with high risk of falls.81 For optimal effects on balance, exercises may need to be performed 30-40 minutes, three times per week, with a total of 90–120 minutes per week, for 11-12 weeks.103,104 In addition, moderate to high challenge balance exercise, more than 50 hours, typically one-hour sessions twice per week for six months, may result in larger effects on falls prevention.81,106 Exercise should include 1) progressively more challenging postures with gradually reduced base of support e.g. two-legged stand, semi-tandem stand, tandem stand or one-legged stand; 2) dynamic movements that perturb centre of gravity e.g. tandem walk, circle turns, leaning and reaching or stepping over obstacles; 3) specific exercise for postural muscle groups e.g. heel stands, toe stands or unsupported sit to stand; or 4) reducing sensory input e.g. standing with eyes closed, standing/walking on unstable surface such as foam mats.81,87 Recommendations regarding intensity of balance challenge relative to individual capacity, rather than the progression of challenge of tasks are limited and development of reliable instruments is needed.107

Effects on cognitive function Physical exercise, in particular aerobic exercise, is suggested to have favourable effects on cognitive function in older people.108,109 Various underlying mechanisms have been explored. Human and animal studies suggest that physical exercise can stimulate improvements in cerebrovascular function, perfusion, and neuroplasticity, which may prevent the progressive loss of cognitive function associated with ageing and disease such as dementia.109-111 However, a recent systematic review did not find evidence of aerobic exercise effects on cognition, even when cardiovascular exercise effects were shown.112 In addition, two recently published randomised controlled trials (RCT) in older people, both with 24-month long interventions and large samples, show encouraging yet somewhat conflicting results. In the first study, a multi-domain intervention (diet, exercise, cognitive training, vascular risk monitoring) a conservative improvement of cognitive function in people at risk of cognitive

8

impairment was shown.113 In the second study, of people at risk of dementia, physical activity as a single intervention alone did not find effects on cognitive function in the complete sample;114 subgroup analyses, however, suggested positive exercise effects on executive function in participants with functional deficits, and in participant over the age of 80 years.114

Exercise in nursing homes In older adults living in nursing homes, a population where dementia is common, beneficial effects of physical rehabilitation on balance and decline in dependence in ADLs have been indicated.115 Furthermore, in the Frail Older People—Activity and Nutrition Study in Umeå (FOPANU), a high-intensity functional exercise program led to improvements in balance,116,117 as well as promising effects on ADL dependence in subgroup analyses of people with dementia.118 Recommendations for older adults living in nursing homes are similar to those for older adults in general, and highlight the importance of moderate-intensity multicomponent exercise programs in 35-45 minute sessions, twice a week.88 However, lower limb strength exercise may be particularly important in this population,88 where physical inactivity is frequent, and where the risk of muscle weakness is high. Improvement in muscle strength has been shown to be highly associated with improved physical function up to a threshold, beyond which the association plateaus.119 This implies that strengthening exercise especially benefits physical function in people with muscle weakness, and muscle strengthening and balance exercise may therefore need to be prioritised before endurance exercise.81,87

Exercise in dementia Older people with dementia often face many challenges to exercise, especially in later stages of the disease. Symptoms of dementia including physical impairments, deficits to memory, language, attention, and motor functions, BPSDs, depressive symptoms, and other comorbidities, may lead to difficulties in participating in and performing exercise programs. To generalise the effects of physical exercise programs in healthy older people to older people with dementia is therefore difficult. Exercise for adults with dementia not only have to be tailored to suit individual physical impairment but also cognitive impairment and other associated symptoms, which may necessitate one-to-one supervision especially in the later stages of the disease. In addition to clear and concise verbal instructions, non-verbal instructions such as physical demonstrations or cues to exercise may be required. Futhermore, instructions may need to be repeated several times, or performed together with or “mirroring” the supervisor. That exercise is task specific may be particularly important in people with AD because of a concomitant difficulty in motor skill transfer i.e. ability to use acquired skills in new contexts.120-122

9

There is promising evidence that physical exercise can benefit gait speed,123-125 balance,123,124,126,127 and ability to independently carry out ADLs,128-130 in nursing home and community dwelling older adults with dementia. However, studies of exercise effects in people with dementia are highly heterogeneous and differ in type and severity of dementia, and type, duration and frequency of exercise, suggesting that results have to be interpreted with caution.128 Furthermore, while large studies show that exercise appears to improve gait in people with dementia in community settings,123,124 results are inconsistent in nursing home populations where severity of physical and cognitive impairments is greater.125-

127 In addition, given the frequency of walking aid use in older people with dementia living in nursing homes,126 its influence on balance or pain,49,50 and responsiveness of the gait speed test,53 an additional investigation of the impact on walking aids use is warranted. Certainly, in people with dementia the impact may be pronounced given the cognitive demand of walking aid use.49,51 Physical exercise may be an important complement to pharmacological treatment in dementia, which have been shown to benefit cognitive function and dependence in ADLs in people with AD,131,132 since side effects are common and its usefulness in people with Vascular dementia remains uncertain.131-133 However, results from RCTs investigating effects of physical exercise on cognition in people with dementia are inconsistent.126,134-144 Studies that show positive effects on cognition are small or lack blinded testers138-140,142,144 or comparable attention control groups.137-143

All in all, there appears to be a need for large exercise studies of methodologically high quality in people with dementia living in nursing homes,128,129 particularly with designs including attention control groups.129 In order to explore the effects of exercise as a single intervention, the additional attention brought by the intervention may need to be controlled,145 especially bearing in mind that this is a population with generally few social interactions. 19,146 Furthermore, no study has compared the effects of exercise between dementia types.129

10

RATIONALE

The number and proportion of older people in the world is increasing rapidly. The age-related increase in risk of disease, such as dementia, physical dysfunction, and need for assistance in ADLs prompts future improvement in disease prevention, detection, and treatment to promote quality of life and provide adequate care.

Given the diversity of ageing process, the older population is a heterogeneous group ranging in cognitive and functional abilities, and morbidity. Thus, a measure to screen for disability and disease in older people may be useful. To that end, gait speed tests have been advocated as a single, quick, simple, cheap and reliable test predictive of several adverse outcomes, including mortality. However, most cohort studies investigating gait speed have focused on community dwelling, well-functioning samples of older individuals with mean age younger than 80 years. There is a need for validation of the results in a population of very old individuals, with dementia or disability, characteristics common in older people likely to be encountered in clinical practice. Thus it seems important to investigate whether gait speed can predict mortality after comprehensive adjustments for potential confounders in very old people.

Physical exercise may be a method for preventing impairments or postponing the decline in physical and cognitive function and their consequences for ADLs. Despite evidence that exercise can benefit gait speed, balance, and independence in ADLs in older people without dementia, studies of exercise effects in people with dementia are highly dissimilar. RCTs in this area differ in methodological quality, population and exercise regimes, which make interpretation of meta-analyses challenging. In addition, given the frequency of walking aid use, its influence on balance or pain, an investigation of the impact of walking aids on exercise effects is warranted. Results from RCTs investigating effects of exercise on cognition in people with dementia are inconsistent. There is a need for large studies of methodologically high quality in this population, particularly with designs including attention control groups. In order to explore the effects of exercise as a single intervention, the additional attention brought by the intervention may need to be controlled for. Furthermore, no study has compared the effects of exercise between dementia types.

11

AIMS

The overall aim of this thesis was to investigate the influence of physical function on survival in very old people, and further to investigate effects of a high-intensity functional exercise program on physical function, cognitive function, and dependence in ADLs in older people with dementia living in nursing homes.

The specific aims of this thesis included:

I. To investigate if gait speed is an independent predictor of all-cause mortality in very old people, and whether the prediction is influenced by sex, dementia, dependence in ADLs, and use of walking aids (Paper I).

II. To investigate effects of a high-intensity functional exercise program on dependence in ADLs, and whether exercise effects differed according to dementia type or level of cognitive impairment. (Paper II).

III. To investigate effects of a high-intensity functional exercise program on balance and gait speed, and whether exercise effects differed according to sex, dementia type, level of cognitive impairment, or walking aid use. (Paper II-III).

IV. To investigate effects of a high-intensity functional exercise program on global and executive cognitive function, and whether exercise effects differed according to sex, dementia type or level of cognitive impairment. (Paper IV).

12

METHODS

This thesis includes data from the Umeå 85+/Gerontological Regional Database (GERDA) study (Paper I) and the Umeå Dementia and Exercise (UMDEX) study (Papers II-IV). The UMDEX study protocol (ISRCTN31767087) is published on the ISRCTN registry. An overview of the studies included is presented in Table 1.

SETTING

The GERDA study Paper I was based on the GERDA study, a population-based cohort study that aims to further knowledge of health, quality of life, and living conditions of very old people, and to develop medical and social care. The GERDA study is conducted by Umeå University, Sweden, and Åbo Academi University/ University of Vaasa, and Novia University of Applied Sciences, Finland. The data collection was initiated in urban and rural areas in the municipality of Umeå in the year 2000, and in rural municipalities of Storuman, Sorsele, Malå, Vilhelmina, and Dorotea in year 2002, and have been repeated every 5-years since. In Finland data collection has been conducted in Vasa and Korsholm in 2005 and 2010, and in Korsnäs and Malax in 2010. Eligible participants are inhabitants of selected areas, chosen from national tax and population registers

Table 1. Overview of studies in the thesis

Paper I II III IV

Study GERDA UMDEX Design Observational cohort Randomised controlled trial Sample size 772 186 Participants: Age, yrs. 89.6 ± 4.6 85.1 ± 7.1 Women 542 (70.2) 141 (75.8) Nursing home 298 (38.6) 186 (100) Dementia 253 (32.8) 186 (100) MMSE 21.3 ± 7.7 14.9 ± 3.5 Barthel ADL 16.0 ± 6.1 10.9 ± 4.9 GS, m/s 0.52 ± 0.211 0.47 ± 0.211 Evaluation target Usual GS 2.4 m The HIFE program

Outcome Survival Dependence in ADLs, Balance

Gait Cognitive function

Outcome measure Date of death Barthel ADL, FIM, BBS

Usual GS 4 m MMSE, ADAS-Cog, VF

Follow-up 5 years 4 and 7 months Statistical Analysis Cox regression Linear mixed models Values are n (%) or mean ± SD. 1no gait speed values imputed ADL = Activities of daily living; ADAS-Cog = Alzheimer’s Disease Assessment Scale – Cognitive subscale; BBS = Berg Balance Scale; FIM = Functional Independence Measure; GS = Gait Speed; HIFE = High-Intensity Functional Exercise; MMSE = Mini-Mental State Examination; m/s = Meters per second; n = number; SD = Standard deviation; VF = Verbal fluency.

13

according to age at the cohort year; every second 85 year old, every 90 year old and every 95 year old or older. Eligible participants are contacted by mail and thereafter by telephone, and offered participation including review of medical records and structured interviews and assessments; conducted during home visit or via telephone.

The UMDEX study Papers II-IV was based on the UMDEX study. Units of Physiotherapy and Geriatric Medicine at the Umeå University, Sweden conducted the UMDEX study in the years 2011-2012, with the aim to investigate effects of a high-intensity functional exercise program in people with dementia living in nursing homes. In the UMDEX study, nursing homes with 30 residents or more within the municipality of Umeå were eligible. In total, of the 18 facilities that met that criterion, two were excluded during the screening process for logistical reasons including too few participants and remote location. Sixteen nursing homes were included, of which nine comprised general nursing home units and ten units for dementia special care, both with private rooms and staff on hand, and seven comprised units with private apartments, access to dining facilities, alarms and on-site nursing and care.

PARTICIPANTS

Data from the GERDA study cohorts collected years 2000, 2002, 2005, 2007 in Sweden, and 2005 in Finland were used. Participants able to complete a gait speed test during the home visit (n = 620) were included (Figure 1). The study also included participants not able to complete the test if failure was due to a habitual physical impairment in gait function (n = 152). Those unable to complete the gait speed test were excluded if the impairment was considered temporary, for instance due to a recent fracture, or as a consequence of severe vision or hearing impairment, lack of motivation, or failure to understand the instructions owing to cognitive impairment. Participants that took part in more than one data collection contributed with their earliest set of gait speed values only. The 347 individuals who declined home visit did not differ in age or sex from the 848 who agreed (P = 0.639 and P = 0.419 respectively).

14

In the UMDEX study, a total of 864 residents eligible for inclusion in 16 residential care facilities were screened by physiotherapists (PT) and physicians. Inclusion criteria were dementia diagnosis according to DSM-IV-TR criteria,62

aged 65 and older, dependence on assistance in one or more personal ADLs according to the Katz Index,147 ability to stand up from a chair with armrests with assistance from no more than one person, MMSE score of ten or greater,148 approval from a physician, and ability to hear and understand spoken Swedish sufficiently to participate in assessments. Age (P = 0.189) and MMSE score (P = 0.713) did not differ between participants included in the study and those who declined participation (n = 55; Figure 2). A larger proportion of men (34%) than women (18%) declined participation (P = 0.008).

ETHICS

All individuals included in the studies provided informed oral consent to participate. In participants with cognitive impairment informed consent was confirmed by relative or otherwise authorised representative. The GERDA study was approved by the Regional Review Board in Umeå (§99-326 and §05-063M) and the Ethics Committee of Vaasa Central Hospital (registration number 05-87). In the UMDEX study ethical approval was obtained from the regional ethics review board of Umeå in August 2011 (2011–205–31M).

Figure 1. Flow of participants through Paper I

5 year Follow-up

Baseline

Assessed (n=772)

Deceased (n=464) Alive (n=308)

Agreed to home visit (n=848)

Eligible participants (n=1310)

Excluded (n=76) No walking test - Deceased (n=2) - Cognitive impairment (n=21) - Motivation (n=24) - Other (n=26)      

Excluded (n=462) Deceased before contact (n=115) Declined home visit (n=347)  

15

Exc

lude

d (n

=67

8)

Inc

lusi

on c

riter

ia n

ot m

et (n

=53

2)

Not

pre

sent

at h

ome

(n=

12)

Dec

lined

MM

SE (n

=74

) D

eclin

ed p

artic

ipat

ion

(n=

55)

Dec

ease

d be

fore

allo

catio

n (n

=5)

 

Ran

dom

ised

(n=

186)

in c

lust

ers

(n=

36)

Elig

ible

par

ticip

ants

(n=

864)

in n

ursi

ng h

omes

(n=

16)

7 m

onth

Fol

low

-up

Bas

elin

e

4 m

onth

Fol

low

-up

Ava

ilabl

e to

ass

ess

(n=

93)

Ava

ilabl

e to

ass

ess

(n=

83)

Ava

ilabl

e to

ass

ess

(n=

79)

Ava

ilabl

e to

ass

ess

(n=

93)

Ava

ilabl

e to

ass

ess

(n=

88)

Ava

ilabl

e to

ass

ess

(n=

79)

Exe

rcis

e C

ontr

ol

Dec

ease

d (n

=4)

In

hos

pita

l (n=

1)1

Dec

ease

d (n

=9)

In

hos

pice

(n=

1)

Dec

ease

d (n

=8)

M

oved

hom

e (n

=1)

C

onse

nt w

ithdr

awn

(n=

1)

Dec

ease

d (n

=4)

Figu

re 2

. Flo

w o

f pa

rtic

ipan

ts th

roug

h Pa

per I

I-IV

1 re

turn

ed fo

r las

t fol

low

-up

16

STUDY DESIGN

Paper I was an observational cohort study. Papers II-IV, was based on the UMDEX study, which was a single blinded, stratified, cluster-randomised, controlled trial with a parallel group design, comparing effects between an exercise activity and an attention control activity.

OUTCOME AND TARGET MEASUREMENTS

In the GERDA study testers were medically trained e.g. physicians, PTs or nurses, who received theoretical and practical training in the tests conducted. In the UMDEX study all testers took part in theoretical and practical group training in assessments before commencement of the baseline measurement phase. PTs and Physicians conducted assessments at baseline and follow-ups blinded to allocation and previous test results. The same tester at baseline and consecutive follow-ups conducted assessments, except on a few occasions (for dependence in ADLs n = 9) when to preserve blindness a change of tester was necessary. PTs interviewed care staff familiar with participants’ need for assistance in ADLs and assessed balance, gait speed, MMSE, and Verbal fluency (VF) at baseline, and at four and seven months (Figure 3). Physicians assessed MMSE, and ADAS-Cog. The baseline assessment of MMSE took place in the screening process. The ADAS-Cog was assessed at baseline and four months only. The hypothesis of the study was not disclosed to participants, staff, or relatives when approached.

Control activity Baseline

tests 1st

Follow-up

2nd

Follow-up

Augus

t Se

ptem

ber

Octobe

r Nov

embe

r Dec

embe

r Jan

uary

Fe

brua

ry

Marc

h Apr

il M

ay

June

2011 2012

Exercise activity

Figure 3. Timeline of measurements in the UMDEX study

July

17

Mortality In Paper I mortality status was followed for 5 years with respect to the inclusion date, using recorded dates of deaths from death certificates, electronic medical records, and population registers.

Gait speed In the GERDA study usual gait speed was tested over a distance of 2.4 meter, chosen because of concerns regarding the limited space in many homes in the community.29 Usual walking speed has been shown to be a reliable and valid measure among older people.149 The set distance was measured and marked on the floor using masking tape. The participant was instructed to start from a standing still position and walk at a usual, comfortable pace past the markings on the floor. The time taken to walk between the markings was measured using a stopwatch. The procedure was repeated twice from which a mean gait speed time was calculated in meters per second (m/s). If only one gait speed time was registered it was included in the analyses. The use of walking aids was allowed during testing and the type recorded.

In the UMDEX study usual gait speed (GS) was measured over the distance 4.0 meters.29 Gait speed mode was chosen based on results on the FOPANU study, which showed effects on usual but not on fast gait speed.116 A 4-meter distance was marked along the base of a wall, and a visible target e.g. a chair placed approximately two meters beyond it. From a standing still start position participants were asked to walk in their usual pace to the visible target. Using a digital stopwatch the time was measured from when instructed to start until the participant’s trunk crossed the marking on the wall. The procedure was repeated twice and the mean gait speed time in m/s calculated. When only one gait speed time was registered it was included in the analyses. The participants’ habitual walking aid was used in the GS test.

In the UMDEX study a second gait speed test without walking aids and with minimum amount of support was conducted (GS-noWA) according to the same test procedure as outlined above, to reduce the support provided by walking aids. Participants that used a walking aid performed the GS-noWA test without the walking aid. Participants that walked unsupported only performed the GS test and their time was carried over to the GS-noWA test. Participants that were unable to walk without a walking aid were offered a minimum amount of living support. One or two testers provided single-sided or double-sided support in a standardized manner; the participant placed their pronated hand(s) over the supinated hand of the tester(s) (Figure 4). To reduce the influence on pace, testers were instructed to provide support only in an upright direction and only by this point of contact, with no other contact or support allowed.

18

No physical assistance or support from nearby structures such as walls or furniture was permitted during the gait speed tests. When unable to perform a gait speed test the reason was registered and categorized as physical impairment, motivation, other causes e.g. pain, dizziness or absence from ward, or deceased. In all gait speed tests, at baseline, 4 and 7 months, the same type of walking aid or amount of support was used.

Dependence in ADLs Dependence in ADLs was assessed using the Barthel ADL Index in all papers and the motor domain of the Functional Independence Measure (FIM) in Paper II only. The ten-item Barthel ADL Index (0–20) is a well-established, valid measure of functional independence.150 Interrater reliability has been found to be fair to very good in people with impaired function.151 The items cover personal care and mobility, with higher scores reflecting greater independence.152 The FIM has been shown to be valid in people with dementia,153 with good test–retest and interrater reliability across a variety of disability levels and medical conditions.154 The motor domain of the FIM comprises 13 items rated on a scale ranging from total assistance (1) to complete independence (7), with a total possible summed score of 91.

19

Balance In the UMDEX study balance was measured using the Berg Balance Scale (BBS). Participants’ ability to maintain an upright posture during 14 functional activities e.g., sitting, rising from a seated position, transfer between two chairs or reaching while standing was rated on a scale ranging from 0 to 4, with higher scores reflecting greater balance. Given that the ability to perform functional activities is multifaceted, the BBS also reflects aspects other than balance, such as lower limb strength. The BBS is a valid, reliable instrument for the measurement of function and evaluation of group effects of interventions in older people living in residential care facilities, including people with cognitive impairment.155,156 Furthermore, the BBS is a recommended test for the measurement of standing balance in older people for research or clinical practice.157

Cognitive function In all papers global cognitive function was measured using the MMSE,148 and in Paper IV the Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-Cog) was also measured.158 The MMSE (range from 0 to 30) is a well-known measure of cognitive function; comprising 20 questions related to domains e.g. attention, language, word recall, orientation to time and place, and where a higher score indicates better function. The ADAS-Cog (range from 0 to 70) is a measure of cognitive function widely used as an outcome measure in pharmaceutical trials targeting slowing progression of AD, and designed to improve assessment of subtle changes in symptoms.159 Similarly to the MMSE, the ADAS-Cog measures various domains but in contrast to the MMSE a lower score indicates better cognitive function. In Paper IV executive function was measured using Verbal fluency (VF).160 Participants were asked to name as many animals as possible within one minute, and a higher number of animals named indicate better executive function.

BASELINE MEASUREMENTS

In all papers socio-demographic information was collected, together with body weight and height. Depressive symptoms were measured using the 15-item Geriatric Depression Scale (GDS-15; range from 0 to 15), where lower score indicates better status.161 When at least ten questions were answered in the GDS-15, missing data was imputed using the individual mean of questions answered.162 In the GERDA study Body Mass Index (kg/m2) calculated. In the UMDEX study pain when weight bearing (yes/no) was evaluated in gait speed test. BPSDs were assessed using the Neuropsychiatric Inventory (0-144),163 where a lower score indicate better status. Dementia type, depressive disorders, and delirium diagnoses were based on medical records, current pharmaceutical

20

treatment, and assessments results. An experienced specialist in geriatric medicine further reviewed and confirmed the diagnoses according to DSM-IV-TR criteria.62

RANDOMISATION

In the UMDEX study randomisation was performed after completion of the enrolment process and baseline assessment to ensure concealed allocation, thereby avoiding selection bias. To reduce contamination between activities, 36 clusters of 3–8 participants each and inhabitants of the same wing, unit, or floor were formed. Randomization was stratified to ensure that participants of both groups lived in each facility, reducing the risk that associated factors would influence the outcome. Two researchers not involved in the study performed randomization by drawing lots using sealed opaque envelopes. The order of allocation for clusters was performed first, followed by allocation to the intervention and attention control groups.

INTERVENTION

Activities in the UMDEX study were conducted in groups of 3-8 participants. PTs supervised the exercise activity, two per group. Occupational Therapist (OT) or OT assistant supervised the attention control activity, one per group. All supervisors were experienced in working with people with cognitive impairment. Five exercise sessions lasting approximately 45 minutes were held per two–week period.87,90 The length of the intervention was four months (40 sessions in total). Individually supervised sessions were when possible offered when participants were unable to attend a group session.

Exercise activity The exercise intervention was based on the High-Intensity Functional Exercise (HIFE) program.164,165 The HIFE program comprises 39 exercises targeted at improvement of lower limb strength, balance, and mobility. The exercises are distributed over 5 categories.

A. Static and dynamic balance exercises in combination with lower-limb strength exercises.

B. Dynamic balance exercises in walking. C. Static and dynamic balance exercises in standing. D. Lower-limb strength exercises with continuous balance support. E. Walking with continuous balance support.

21

The exercises are performed in functional, weight-bearing positions; similar to those used in everyday situations, such as rising from a chair, stepping up, trunk rotation while standing, and turning or stepping over obstacles while walking. All exercise equipment used e.g. foam mats or rolls, steps and weighted-belts was portable.

Exercises were initially selected depending on individuals’ degrees of functional deficit and according to a hierarchical model based on level of living support required when walking a short distance (five to ten meters). According to instructions participants that walked unsupported performed category A and B exercises, those that walked with supervision or minor physical support (≤ one person) performed category A, B and C exercises, and those that walked with major physical support or not able to walk performed category C, D, and E exercises. Participants were recommended to perform at least two lower limb strength exercises and two balance exercises in two sets each session. All participants were supervised individually to promote the highest possible exercise intensity, while ensuring their safety.

Lower limb strength and balance exercise intensity was pre-defined as low, medium, and high. Strength exercise intensity were defined as high when performed in sets of 8-12 RMs, medium at 13-15 RM, and low at > 15 RM.96 Balance exercise intensity were defined according to level of challenge to postural stability, and deemed high when fully challenged, medium when not fully challenged or fully challenged in only a minority of the exercises, and low when in no way challenged.164,165 Exercises were progressed by increased load, for example by stepping higher, rising from a lower chair, or adding weights to a belt worn around the waist (maximum 12 kg). Balance exercises were progressed by, for example, narrowing the base of support, or altering the surface. For safety participants wore belts with handles so that PTs could provide support if needed when exercising near limits of postural stability, and thereby prevent falls. Unnecessary support was avoided. As a build-up, participants were encouraged to exercise at moderate intensity (13–15 RM for strength exercises) the first two weeks.

Exercises and intensity were adapted throughout the intervention to meet participants’ levels of cognition, BPSDs, and changes in health and functional status. Activity leaders were encouraged to obtain updates on participants’ health status before the activity and were able to contact physicians or nurses when necessary.

Attention control activity The OTs and OT assistant who supervised the attention control activity also developed the program. Each session was structured around topics believed to

22

be of interest to older people, such as local animals, writers, and composers, sport personalities, flowers, pets, handicraft, the royal family, and the Christmas celebration. While sitting together in a group, participants conversed, sang, listened to music or readings, and looked at and held pictures or objects associated with the topic. Materials used were books, brochures, CDs, spices, plants, and items such as candles, and Christmas decorations.

ADHERENCE, EXERCISE INTENSITY AND ADVERSE EVENTS

At the end of each session the activity leaders completed a structured protocol for each participant, including attendance, adverse events, and in the exercises group, intensity achieved. Any discomfort brought on, or worsened during activity sessions, both observed, or expressed either spontaneously or when questioned, were recorded as adverse events. The severity of each event were assessed, first independently then in consensus, by two specialists in geriatric medicine and one PT, and classified as minor or temporary, serious (potential risk of severe injury or threat to life), manifest injury or disease, or death. Furthermore, a specialist in geriatric medicine assessed the association between participation in the study and any deaths that occurred, from the start of intervention until one month after the final follow-up.

STATISTICAL ANALYSES

All analyses were performed using IBM SPSS version 20.0 to 23.0 (Armonk, NY: IBM Corp.) and R version 3.2.2 (Vienna, Austria: R Core Team) software. All statistical tests were two tailed and P < 0.05 was considered to be statistically significant.

Paper I In Paper I Student’s t-test and Pearson’s chi-square tests were used to compare differences according to survival status at 5-year follow-up. Gait speed was categorized based on quartiles of the time taken to complete the set distance and assigned to the following groups: (1) ≤ 0.36 m/s, (2) 0.37 – 0.49 m/s, (3) 0.50 – 0.63 m/s, and (4) ≥ 0.64 m/s. Participants unable to complete the test were assigned group 0. The correlation between all variables was tested, and considered too high between dementia and MMSE (r= –0.751), and also between Barthel ADL Index and categorical gait speed (r = 0.688). Dementia and Barthel ADL Index were subsequently removed from the list of covariates.

Cox proportional hazards regression models were used to analyse associations between gait speed categories and mortality risk. The proportionality of hazards was statistically tested using Schoenfeld residuals.166 The first model (Model 1)

23

was adjusted for age and sex. Covariates that did not fulfil the Cox proportional hazards assumptions (age and education) in the second model (Model 2) were adjusted for by entering a time-dependent variable for each into an extended Cox proportional hazards model. In Model 2, age, sex, and all covariates associated with mortality at P ≤ 0.15 (Table 2), together with the two time-dependent variables were adjusted for. Any interaction effects between gait speed and sex, dementia, dependence in ADLs (Barthel ADL Index < 20), and use of walking aids were investigated by adding an interaction term for each into the second model.

Papers II-IV In the UMDEX study sample size was calculated based on results from the FOPANU Study,118 considering a between-group effect of 1.1 points on the Barthel ADL Index and an intracluster correlation of 0.02. A sample size of 183 participants was required to verify significant intervention effects at a statistical power of 80% at the four-month follow-up, a two-sided significance level of 0.05, and a presumed dropout rate of 10%.

A prespecified strategy for selection of adjusting variables was formulated. Significant imbalances between exercise and attention control groups167 and associations (r ≥ 0.3) with changes in outcome measures at four and seven months168 were analysed from baseline variables (preselected as descriptive or possible confounding variables) using Student’s t-test or Pearson’s chi-squared tests, and Pearson correlation coefficients. No variable was found to associate with change in outcome measures above predefined levels. The variable antidepressant use differed between groups (P = 0.04) and was adjusted for in analyses.

In agreement with the intention-to-treat principle available data for each participant were analysed according to original allocation and regardless of level of attendance. Longitudinal changes in outcomes (Barthel ADL Index, FIM, BBS, GS, GS-noWA, MMSE, ADAS-Cog, VF) from baseline to four and seven months were analysed in linear mixed effects models, using interaction terms for activity and time point and adjustment for age, sex, and antidepressant use as fixed effects, and individual and cluster allocation as random effects. Baseline values for outcome measures were included in the dependent variable to avoid loss of data. The least square mean within-group difference was estimated from these models.

Differences in execise effects were analysed, according to dementia type and cognitive level (Papers II-IV), sex (Papers III-IV), and support in the gait speed test (Paper III) by adding interaction terms to adjusted models. Dementia type was dichotomised as AD or other (non-AD) dementia, in part because many

24

previous trials have included participants with AD only, and further by indication of the directional effect on main outcome in unadjusted within-group analyses of non-AD types. Level of cognitive impairment was dichotomised based on the median MMSE score of 15, median ADAS-Cog score of 30, and median VF score of 6. Support in the test was dichotomised as unsupported versus any walking aid or living support in GS and GS-noWA tests, respectively. The difference in effect between the exercise and attention control groups in each subgroup was further investigated using Student t-tests and least square mean changes from baseline, with fewer degrees of freedom to obtain conservative P-values. Effect size was calculated by dividing the between-group difference in linear mixed models by the unadjusted pooled standard deviation of the difference between post- and preintervention values.169

The influence of outliers was explored in sensitivity analyses of main outcomes, whereby adjusted analyses were repeated after removal of extreme values, defined as greater than three times interquartile range. The number of extreme outliers varied (n=1-4) between outcome measures, activity groups, and follow-ups.

In Paper III, to further explore the absence of exercise effects in participants that performed the gait speed test with walking aids or minimum support, additional analyses on change in the Berg Balance Scale (BBS) was conducted using the same interaction terms for walking aid use and amount of support and adjustments as described previously. Furthermore, to account for possible bias introduced by missing gait speed values, multiple imputation sensitivity analyses were performed. Fifteen imputed datasets were generated using predictors based on baseline variables that were (a) adjusting or outcome variables in the model, (b) associated (r > 0.3) with outcomes, or (c) associated (r > 0.3) with categorical causes for missing gait speed values.170 The multiple imputation models were sequentially conducted for each follow up. Imputations were constricted dependent on causes for missing values; when caused by physical impairment maximum imputed values was limited to the lowest observed value, while when caused by the participants’ being dead no values were imputed.

25

RESULTS

Paper I included 772 participants, of which 464 (60%) were deceased five years later. Description of baseline characteristics of the study population is presented in Table 2, according to survival status at the end of the five-year follow-up. The mean age was 89.6 years, 542 (70%) were women, 298 (39%) lived in nursing homes, 418 (54%) were dependent in ADLs, and 253 (33%) of individuals had dementia disorder. Of participants that lived in nursing homes, 237 (80%) were women, 268 (90%) were dependent in ADLs and 183 (61%) had dementia.

The mean ± SD gait speed was 0.52 ± 0.21 m/s for the 620 participants (80%) able to complete the gait speed test (Table 3). Women had a slower gait speed than men of 0.49 ± 0.20 and 0.58 ± 0.23 m/s (P < 0.001), respectively. The

Table 2. Baseline characteristics in Paper I

Characteristic Total

N = 772 Deceased N = 464

Alive N = 308 P

Age 89.6 ± 4.6 90.9 ± 4.8 87.5 ± 3.4 < .001 Women 542 (70.2) 319 (68.8) 223 (72.4) .277 Nursing home resident 298 (38.6) 250 (53.9) 48 (15.6) < .001 Education < 8 years, n=750 550 (73.3) 341 (77.0) 209 (68.1) .007 Diagnoses/medical conditions: Dementia 253 (32.8) 216 (46.6) 37 (12.0) < .001 Depression, n=766 243 (31.7) 178 (38,9) 65 (21.1) < .001 Cerebrovascular disease 152 (19.7) 103 (22.2) 49 (15.9) .031 MI previous year 22 (2.8) 17 (3.7) 5 (1.6) .095 Heart failure (n=771) 227 (29.4) 183 (39.5) 44 (14.3) < .001 Hip fracture 134 (17.4) 97 (20.9) 37 (12.0) .001 Prescription medications: Benzodiazepines 229 (29.7) 152 (32.8) 77 (25.0) .021 Antidepressants 134 (17.4) 99 (21.3) 35 (11.4) < .001 Diuretics 381 (49.4) 249 (53.7) 132 (42.9) .003 Analgesics 333 (43.1) 237 (51.1) 96 (31.2) < .001 No. of prescribed drugs 6.5 ± 4.2 7.2 ± 4.4 5.3 ± 3.7 < .001 Assessments: Barthel ADL (0-20), n=7661 16.0 ± 6.1 14.0 ± 7.0 19.0 ± 2.6 < .001 BMI, n=740 25.2 ± 4.4 24.8 ± 4.6 25.8 ± 4.1 .001 MMSE (0–30), n=7571 21.3 ± 7.7 18.5 ± 8.5 25.2 ± 3.7 < .001

Values are n (%) or mean ± SD. Numbers reported after covariates indicate number of measurements available when values were missing. 1 a higher score indicates better status. ADL = Activities of daily living; BMI = Body Mass Index; MMSE = Mini-Mental State Examination; MI = Myocardial infarction; SD = Standard deviation

26

gait speed of community dwelling participants was 0.56 ± 0.21 m/s compared with 0.41 ± 0.18 m/s in nursing homes. In nursing homes, the gait speed in participants with dementia was 0.38 ± 0.17, and 0.42 ± 0.19 m/s in participants without dementia. Of the 620 participants that performed the test, 60 (9.7%) walked with a speed of 0.8 m/s or over, and 15 (2.4%) with a speed of 1.0 m/s or over. In the total sample, 169 participants (27%) used walking aids in the gait speed test, with rollator being the most common type used (Table 3). Of the 152 participants that could not perform the gait speed test for physical reasons, 80% lived in nursing homes and 82% had dementia.

Papers II-IV included 186 participants (141 women, 45 men) in nursing homes, 67 (36%) of whom had AD (Table 4). One hundred-and-nineteen participants (64%) had non-AD types of dementia, which included Vascular dementia, Mixed AD and Vascular dementia, Frontotemporal dementia, dementia with Lewy Bodies, and Parkinson’s disease dementia. Ninety-eight participants (82%) with non-AD dementia had Vascular dementia, alone or in combination with other dementia types. Participants with non-AD dementia had better cognitive function and were more likely to have medical conditions such as stroke, heart failure, and hip fracture than those with AD (Table 4, Table 5). One hundred and forty-five participants (78%) used wheelchair or walking aids. At baseline 42 participants (23%) performed the GS test unsupported and 99 participants (53%) the GS-noWA test unsupported.

Table 3. Target measure and use of walking aids in Paper I

Characteristic Total

N = 620 Women N = 416

Men N = 204

Gait speed, m/s 0.52 ± 0.21 0.49 ± 0.20 0.58 ± 0.23 85 years, n=308 (♀=207, ♂=101) 0.57 ± 0.21 0.55 ± 0.20 0.61 ± 0.23 90 years, n=202 (♀=129 ♂=73) 0.49 ± 0.21 0.44 ± 0.19 0.58 ± 0.22 95+ years, n=110 (♀=80 ♂=30) 0.43 ± 0.19 0.40 ± 0.17 0.49 ± 0.21 GS, without walking aids, m/s, n=446 (♀=290 ♂=156)1 0.58 ± 0.20 0.55 ± 0.19 0.64 ± 0.22 GS, with walking aids, m/s, n=169 (♀=123 ♂=46)1 0.36 ± 0.15 0.35 ± 0.15 0.39 ± 0.12 Use of walking aids in gait speed test1 None 446 (72.5) 290 (70.2) 156 (77.2) Cane, crutch (single-sided support) 26 (4.2) 13 (3.1) 13 (6.4) Rollator (double-sided support) 136 (22.1) 105 (25.4) 31 (15.3) Other (double-sided support) 7 (1.2) 5 (1.2) 2 (1.0)

Values are n (%) or mean ± SD 1Complete data from 615 individuals since five individuals had missing values on walking aids GS = Gait speed; M/s = meters per second; SD = standard deviation; ♀ = women; ♂ = men

27

In the UMDEX study adherence to the exercise activity was 73% and to the control activity 70%. Strength exercises were performed at moderate intensity at a median of 41% and high intensity at a median of 47%, and balance exercises were performed at high intensity at a median of 75% of attended sessions. All reported adverse events related to exercise or control activities were assessed as minor or temporary. In the case of one participant’s death, an indirect association with exercise could not be excluded with complete certainty; the individual fell ill one day after participation in an exercise session and later died from causes attributed to circulatory failure and general atherosclerosis.

Table 4. Baseline characteristics in Papers II-IV

Characteristics Total

N = 186 Exercise N = 93

Control N = 93

AD N = 67

Non-AD N = 119

Age, years 85.1 ± 7.1 84.4 ± 6.2 85.9 ± 7.8 85.5 ± 6.1 88.9 ± 7.6 Female 141 (75.8) 70 (75.3) 71 (76.3) 53 (79.1) 88 (73.9) Dementia type Vascular 77 (41.4) 36 (38.7) 41 (44.1) AD 67 (36.0) 34 (36.6) 33 (35.5) Other 27 (14.5) 15 (16.1) 12 (12.9) Mixed AD/Vascular 15 (8.1) 8 (8.6) 7 (7.5) Diagnoses/medical conditions: Depressive disorders 107 (57.5) 53 (57.0) 54 (58.1) 40 (59.7) 67 (56.3) Delirium previous week 102 (54.8) 48 (51.6) 54 (58.1) 37 (55.2) 65 (54.6) Previous stroke 57 (30.6) 33 (35.5) 24 (25.8) 5 (7.5) 52 (43.7) Heart failure 56 (30.1) 24 (25.8) 32 (34.4) 15 (22.4) 41 (34.5) Hip fracture 53 (28.5) 28 (30.1) 25 (26.9) 11 (16.4) 42 (35.3) Prescription medication: Analgesics 112 (60.2) 55 (59.1) 57 (61.3) 42 (62.7) 70 (58.8) Antidepressants 102 (54.8) 58 (62.4) 44 (47.3) 43 (64.2) 59 (49.6) Cholinesterase inhibitor 40 (21.5) 25 (26.9) 15 (16.1) 26 (38.3) 14 (11.8) Memantine 12 (6.5) 7 (7.5) 5 (5.4) 11 (16.4) 1 (0.8) Number of drugs 8.3 ± 3.8 8.4 ± 4.0 8.2 ± 3.7 7.5 ± 3.3 8.8 ± 4.1 Assessments: Pain walking, n=185 35 (18.9) 15 (16.3) 20 (21.5) 15 (22.4) 20 (16.8) Mobility device Wheelchair 24 (12.9) 11 (11.8) 13 (14.0) 5 (7.5) 19 (16.0) Rollator 117 (62.9) 64 (68.8) 53 (57.0) 34 (50.8) 83 (69.7) Stick or crutch 4 (2.2) 1 (1.1) 3 (3.3) 1 (1.5) 3 (2.5) Unsupported 41 (22) 17 (18.3) 24 (25.8) 27 (40.3) 14 (11.8) GDS-15 (0-15), n=1831 3.8 ± 3.2 4.0 ± 3.4 3.6 ± 2.9 3.1 ± 3.1 4.1 ± 3.1 NPI (0-144)1 14.8 ± 14.2 15.2 ± 15.8 14.4 ± 12.6 16.7 ± 13.2 13.7 ± 14.7

Values are n (%) or mean ± SD. Numbers reported after covariates indicate number of measurements available when values were missing. 1 a lower score indicates better status. AD = Alzheimer's disease; GDS-15 = Geriatric Depression Scale-15; MNA = Mini Nutritional Assessment; NPI = Neuropsychiatric Inventory; Non-AD = vascular dementia, mixed AD and vascular dementia, and all other types of dementia; SD = standard deviation.

28

OUTCOMES

Gait speed and Mortality In Paper I slower gait speed was found to associate with increased risk of mortality in a Cox proportional hazard regression model adjusted for age and sex (Model 1). After adjusting for baseline characteristics associated with death, the association between lower gait speed categories and higher five-year mortality risk remained significant (Model 2). Participants unable to complete the test had a hazard ratio (HR) of 2.27 (95% Confidence interval, CI: 1.44–3.59) and participants with a gait speed lower than 0.36 m/s and 0.37 to 0.49 m/s had an HR of 1.97 (95% CI: 1.34–2.88) and 1.99 (95% CI: 1.38–2.85), respectively, when compared with the fastest gait speed category (reference group) (Table 6). Figure 5 shows survival curves based on Cox proportional hazard regression models.

No interaction effects between the association of gait speed and mortality were observed for the subgroups sex (P = 0.953), dementia (P = 0.913), ADL dependency (P = 0.214), or use of walking aids during the test (P = 0.706).

Table 5. Baseline Measures of Outcomes in Papers I-IV

Outcome measure Total

N = 186 Exercise N = 93

Control N = 93

AD N = 67

Non-AD N = 119

FIM (13-91)1 52.0 ± 16.9 51.6 ± 17.1 52.5 ± 16.7 53.2 ± 16.7 51.4 ± 17.0 Barthel ADL (0-20)1 10.9 ± 4.4 10.7 ± 4.5 11.0 ± 4.4 11.1 ± 4.3 10.7 ± 4.5 BBS (0-56)1 28.9 ± 14.5 28.6 ± 14.3 29.3 ±14.7 33.0 ± 13.7 26.7 ± 14.4 GS, m/s, n=177 0.47 ± 0.21 0.47 ± 0.20 0.47 ± 0.22 0.51 ±0.25 0.44 ± 0.19 GSnoWA, m/s, n=169 0.44 ± 0.23 0.42 ± 0.23 0.45 ± 0.24 0.49 ±0.27 0.40 ± 0.20 MMSE (0-30)1 14.9 ± 3.5 15.4 ± 3.4 14.4 ± 3.5 14.0 ± 3.1 15.4 ± 3.6 ADAS-Cog (0-70), n=1832 31.6 ± 10.8 31.8 ± 11.4 31.3 ± 10.3 34.7 ± 10.6 29.8 ± 10.6 Verbal fluency, n=1821 6.4 ± 3.8 6.8 ± 4.1 5.9 ± 3.5 5.7 ± 3.8 5.9 ± 3.5

Values are mean ± SD. Numbers reported after covariates indicate number of measurements available when values were missing. 1 a higher score indicates better status. 2 a lower score indicates better status. AD = Alzheimer's disease; ADAS-Cog = Alzheimer’s Disease Assessment Scale – Cognitive subscale; BBS = Berg Balance Scale; FIM = Functional Independence Measure; MMSE = Mini-Mental State Examination; M/s = meters per second; Non-AD = vascular dementia, Mixed AD and Vascular dementia, and all other types of dementia; SD = standard deviation.

29

Figure 5. Survival curves based on Cox proportional hazards regression; in model 1 (A) adjusted for age and sex, and in model 2 (B) age, sex, nursing home residency, lives alone, education, depression, cerebrovascular disease, myocardial infarction, heart failure, hip fracture, diabetes, malignancy, benzodiazepines, antidepressants, diuretics, analgesics, neuroleptics, number of prescribed drugs, systolic blood pressure, diastolic blood pressure, vision impairment, hearing impairment, Body Mass Index, Mini-Mental State Examination.

Table 6. Mortality rate (per 1000 persons) and hazard ratios for death1 according to Gait speed categories

Model 12 Model 23 Gait speed categories (m/s)

N Mortality

Rate/1000 (95% CI)

HR (95% CI) P HR (95% CI) P

4) ≥ 0.64 162 327 (254–400) 1 1 3) 0.50–0.63 155 426 (347–505) 1.34 (0.93–1.93) .111 1.11 (0.76–1.62) .604 2) 0.37–0.49 152 645 (568–722) 2.48 (1.77–3.47) <.001 1.99 (1.38–2.85) <.001 1) ≤ 0.36 151 709 (635–782) 2.94 (2.10–4.14) <.001 1.97 (1.34–2.88) .001 0) Unable 152 921 (878–964) 5.80 (4.11–8.14) <.001 2.27 (1.44–3.59) <.001

1Calculated using Cox proportional hazards regression models. 2Adjusted for age and sex. 3Adjusted for age, age x follow-up time, sex, care facility resident, lives alone, education, education x follow-up time, depression, cerebrovascular disease, myocardial infarction, heart failure, hip fracture, diabetes, malignancy, benzodiazepines, antidepressants, diuretics, analgesics, neuroleptics, number of prescribed drugs, systolic blood pressure, diastolic blood pressure, vision impairment, hearing impairment, body mass index, mini-mental state examination. CI = confidence interval; HR = hazard ratio; M/s = meters per second; N = number of participants.

Follow-up time (days)2000150010005000

Cum

ulat

ive

Surv

ival

1.0

0.8

0.6

0.4

0.2

0.0

B

0.640.50 - 0.630.37 – 0.49 0 .36unable

Gait speed categories (m/s)

Page 1

Follow-up time (days)2000150010005000

Cum

ulat

ive

Surv

ival

1.0

0.8

0.6

0.4

0.2

0.0

A

0.640.50 – 0.630.37 – 0.49 0 .36unable

Gait speed categories (m/s)

Page 1

≥ ≥

≤ ≤

30

Dependence in ADLs In Paper II independence in ADLs deteriorated in both groups, with no significant between-group difference at four or seven months (Table 7, Figure 6A-B).

In interaction analyses, the effect of exercise was significant in favour of

participants with non-AD dementia (vs. those with AD) at seven months according to FIM and Barthel ADL Index scores (Table 8). The examination of between-group exercise effects in subgroup analyses revealed significant positive effects in participants with non-AD dementia according to the FIM at seven months (Figure 6D) and the Barthel ADL Index (Figure 6E) at four and seven months. The FIM reflected negative effects in participants with AD at seven months (Figure 6D).

In interaction analyses effects on FIM and Barthel Index did not differ according to cognitive level at baseline (Table 9, Figure 7A-B).

In sensitivity analyses, repeated adjusted analyses after removal of extreme outliers (n=1-4), the intervention effects on ADLs in the total sample remained essentially the same. Subgroup analyses showed no negative effect of exercise in participants with Alzheimer’s disease or those with lower cognitive levels.

Table 7. Within-group and Between-group Differences from Baseline in the Functional Independence Measure (FIM) and Barthel ADL Index

Measure Within-group difference Between-group difference

N Exercise,

Mean (SE) N Control,

Mean (SE) Mean (95% CI) P

FIM 4 months 83 –3.10 (1.07) 88 –4.44 (1.04) 1.34 (–1.56 to 4.25) .365 7 months 79 –6.77 (1.09) 79 –7.55 (1.08) 0.78 (–2.21 to 3.77) .610 Barthel ADL 4 months 83 –0.79 (0.31) 88 –1.39 (0.30) 0.60 (–0.24 to 1.44) .161 7 months 79 –1.56 (0.32) 79 –2.12 (0.32) 0.57 (–0.30 to 1.43) .200

Values are from linear mixed effects models of the complete sample (n=186) adjusted for age, sex, and antidepressant use. 1Based on proportion of variation explained by cluster ADL = activities of daily living; N = number of participants with complete data; CI = confidence interval

31

Table 8. Within-group differences from baseline in the Functional Independence Measure (FIM) and Barthel ADL Index, and differences in exercise effect according to dementia type

Measure

Within-group difference Interaction1

N Exercise, Mean (SE)

N Control, Mean (SE)

Mean (95 % CI) P

FIM: 4 months AD 30 –2.97 (1.77) 33 –3.67 (1.69) 1.04 (–4.98 to 7.04) .737 Non-AD 53 –3.18 (1.33) 55 –4.91 (1.30) 7 months AD 29 –9.14 (1.79) 29 –5.33 (1.77) 7.28 (1.10 to 13.47) .022 Non-AD 50 –5.40 (1.36) 50 –8.86 (1.35) Barthel ADL: 4 months AD 30 –1.01 (0.51) 33 –0.85 (0.49) 1.21 (–0.53 to 2.95) .170 Non-AD 53 –0.66 (0.39) 55 –1.71 (0.38) 7 months AD 29 –2.13 (0.52) 29 –1.22 (0.52) 2.34 (0.55 to 4.13) .011 Non-AD 50 –1.23 (0.40) 50 –2.65 (0.39)

Values are from linear mixed effects models of the complete sample (n = 186) adjusted for age, sex, and antidepressant use. 1Difference in exercise effect between participants with non-AD and AD. A positive mean value indicates a greater effect in favour of participants with non-AD. AD = Alzheimer’s disease; ADL = activities of daily living; CI = confidence interval; N = number of participants with complete data; non-AD = Non-Alzheimer’s dementia i.e. vascular dementia, Mixed AD and Vascular dementia, and all other types of dementia.

Table 9. Within-group differences from baseline in the Functional Independence Measure (FIM) and Barthel ADL Index, and differences in exercise effect according to cognitive

function at baseline

Measure

Within-group difference Interaction1

N Exercise

Mean (SE) N Control

Mean (SE)

Mean (95 % CI) P

FIM: 4 months Lower (<14) 35 –4.91 (1.63) 49 –4.97 (1.38) 1.90 (–3.94 to 7.72) .525 Higher (≥15) 48 –1.80 (1.39) 39 –3.75 (1.54) 7 months Lower (<14) 34 –7.84 (1.65) 44 –5.74 (1.44) 5.91 (–0.09 to 11.9) .054 Higher (≥15) 45 –5.99 (1.43) 35 –9.8 (1.61) Barthel ADL: 4 months Lower (<14) 35 –1.12 (0.48) 49 –1.44 (0.41) 0.47 (–1.24 to 2.17) .590 Higher (≥15) 48 –0.55 (0.41) 39 –1.32 (0.45) 7 months Lower (<14) 34 –1.85 (0.48) 44 –1.71 (0.42) 1.44 (–0.32 to 3.18) .109 Higher (≥15) 45 –1.34 (0.42) 35 –2.63 (0.47)

Values are mean (standard error) unless stated otherwise, from linear mixed effects models of complete sample (n = 186) and adjusted for age, sex, and antidepressants use. 1Difference in exercise effect between participants with Mini-Mental State Examination score ≥15 and <14. A positive mean value indicate greater effect in favour of participants with ≥15. ADL = activities of daily living; CI = confidence interval; SE = standard error

32

Balance At four months, balance had improved in the exercise group and declined in the attention control group; at seven months, balance had declined in both groups. The difference between groups was significant at four, but not at seven, months (Table 10, Figure 6C).

In interaction analyses, the effect of exercise was significant in favour of participants with non-AD dementia (vs. those with AD) at four and seven months according to BBS scores (Table 11). The examination of between-group exercise effects in subgroup analyses revealed significant positive effects in participants with non-AD dementia according to the BBS (Figure 6F) at four and seven months. BBS scores reflected negative effects in participants with AD at seven months (Figure 6F).

In interaction analyses according to cognitive level, exercise effects benefited participants with higher cognitive levels more than those with lower cognitive levels according to BBS score at seven months (Table 12). Analysis of between-group effects revealed a negative effect in participants with lower cognitive function at seven months (Figure 7C).

In sensitivity analyses, repeated adjusted analyses after removal of extreme outliers (n=1-3), the intervention effects on balance in the total sample remained essentially the same.

Table 10. Within-group and between-group differences from baseline in the Berg Balance Scale (BBS)

Measure Within-group difference Between-group difference

N Exercise, mean (SE)

N Control, mean (SE)

Mean (95% CI) P

BBS: 4 months 81 2.39 (0.88) 86 –1.82 (0.86) 4.20 (1.79 to 6.61) <.001 7 months 74 –2.08 (0.91) 75 –2.05 (0.90) –0.02 (–2.53 to 2.49) .985

Values are from linear mixed effects models of the complete sample (n=186) adjusted for age, sex, and antidepressant use. N = number of participants with complete data; CI = confidence intervals; SE = standard error.

33

Table 11. Within-group differences from baseline in the Berg Balance Scale (BBS), and differences in exercise effect according to dementia type

Measure

Within-group difference Interaction1

N Exercise, Mean (SE)

N Control, Mean (SE)

Mean (95% CI) P

BBS: 4 months AD 30 –0.44 (1.44) 33 –1.34 (1.38) 5.26 (0.35 to 10.18) .037 Non-AD 51 4.02 (1.10) 53 –2.14 (1.08) 7 months AD 27 –5.33 (1.49) 27 –1.15 (1.48) 6.57 (1.4 to 11.73) .013 Non-AD 47 –0.21 (1.13) 48 –2.59 (1.12)

Values are from linear mixed effects models of the complete sample (n = 186) adjusted for age, sex, and antidepressant use. 1Difference in exercise effect between participants with non-AD and AD. A positive mean value indicates a greater effect in favour of participants with non-AD. AD = Alzheimer’s disease; CI = confidence interval; N, number of participants with complete data; Non-AD = Non-Alzheimer’s dementia i.e. vascular dementia, Mixed AD and Vascular dementia, and all other types of dementia.

Table 12. Within-group differences from baseline in the Berg Balance Scale (BBS), and differences in exercise effect according to cognitive function

Measure

Within-group difference Interaction1

N Exercise

Mean (SE) N Control

Mean (SE)

Mean (95% CI) P

BBS: 4 months Lower (<14) 35 0.36 (1.34) 48 –1.64 (1.15) 3.96 (–0.88 to 8.79) .109 Higher (≥15) 46 3.92 (1.17) 38 –2.03 (1.28) 7 months Lower (<14) 33 –4.20 (1.37) 41 –1.35 (1.22) 5.34 (0.31 to 10.37) .038 Higher (≥15) 41 –0.42 (1.22) 34 –2.90 (1.34)

Values are mean (standard error) unless stated otherwise, from linear mixed effects models of complete sample (n = 186) and adjusted for age, sex, and antidepressants use. 1 Difference in exercise effect between participants with Mini-Mental State Examination score ≥15 and <14. A positive mean value indicate greater effect in favour of participants with ≥15. CI = confidence interval; SE = standard error.

34

Figu

re 6

. Cha

nges

in F

unct

iona

l Ind

epen

denc

e M

easu

re, B

arth

el A

DL

Inde

x, a

nd B

erg

Bal

ance

Sca

le (A

–C) a

nd A

ccor

ding

to D

emen

tia

Typ

e (D

–F).

Val

ues

are

leas

t squ

are

mea

ns o

f cha

nges

from

bas

elin

e, w

ith 9

5% c

onfid

ence

inte

rval

s, fr

om li

near

mix

ed e

ffec

ts m

odel

s of

th

e co

mpl

ete

sam

ple

(n =

186

) adj

uste

d fo

r age

, sex

, and

ant

idep

ress

ant u

se. N

on-A

D d

emen

tia in

clud

ed V

ascu

lar d

emen

tia, m

ixed

A

lzhe

imer

’s di

seas

e an

d V

ascu

lar d

emen

tia, a

nd a

ll ot

her t

ypes

of n

on-A

lzhe

imer

’s de

men

tia. A

D =

Alz

heim

er's

dise

ase;

ES

= e

ffec

t siz

e.

35

Figu

re 7

. Cha

nges

in F

unct

iona

l Ind

epen

denc

e M

easu

re, B

arth

el A

DL

Inde

x, a

nd B

erg

Bal

ance

Sca

le a

ccor

ding

to c

ogni

tive

func

tion.

V

alue

s ar

e le

ast s

quar

e m

eans

of c

hang

es fr

om b

asel

ine,

with

95%

con

fiden

ce in

terv

als,

from

line

ar m

ixed

eff

ects

mod

els

of th

e co

mpl

ete

sam

ple

(n =

186

) adj

uste

d fo

r age

, sex

, and

ant

idep

ress

ant u

se. M

MSE

= M

ini-M

enta

l Sta

te E

xam

inat

ion.

36

Gait speed There were no differences between exercise and activity groups in either gait speed test at four or seven months (Table 13, Figure 8A-B).

In interaction analyses exercise effects on gait speed significantly differed according to support, with larger effects in participants that walked unsupported in the GS test at four and seven months and in the GS-noWA test at seven months (Table 14). The between-group analyses showed positive effects on gait speed in participants that walked unsupported in the exercise group both at four and seven months (Figure 8C-D). In participants that walked with support there was no difference between groups at four months and, while at seven months the exercise group had negative effects on GS, no difference was observed in GS-noWA (Figure 8C-D).

Interaction analyses according to sex and dementia type showed no differences in exercise effects at four and seven months in either gait speed test. The additional interaction analyses showed that the positive exercise effects on balance (BBS score, 95% confidence intervals) did not differ according to walking aid use or amount of support in the gait speed test (0.06, –5.7–5.8, P = 0.983 and 0.31, –4.5–5.2, P = 0.901, respectively). When primary analyses were repeated using the multiple imputed data sets the results remained essentially the same.

Cognitive function There were no differences between exercise and activity groups in MMSE, ADAS-Cog or VF at four or seven months (Table 15, Figure 9).

In interaction analyses exercise effects on cognition did not differ at four or seven months according to sex or dementia type. The interaction analyses indicated that effects on MMSE differed according to level of cognition at baseline at seven months, with a negative exercise effect, while it did not differ at four months, and it did not differ on ADAS-Cog or VF at any follow-up (Table 16).

37

Table 13. Within-group and between-group differences from baseline in Gait speed with walking aid if any (GS) and Gait speed without walking aid and with minimal support (GSnoWA)

Measure Within-group difference Between-group difference

N Exercise, mean (SE)

N Control, mean (SE)

Mean (95% CI) P

GS: 4 months 75 –0.021 (0.015) 78 –0.015 (0.015) –0.006 (–0.047 to 0.035) .777 7 months 64 –0.031 (0.016) 69 –0.035 (0.015) 0.004 (–0.039 to 0.048) .842 GS-noWA: 4 months 67 0.005 (0.017) 68 –0.022 (0.016) 0.027 (–0.018 to 0.073) .242 7 months 53 –0.024 (0.018) 57 –0.046 (0.018) 0.023 (–0.027 to 0.072) .368

Values are from linear mixed effects models of the complete sample (n=186) adjusted for age, sex, and antidepressant use. ADL = activities of daily living; CI = confidence interval; N = number of participants with complete data; SE = standard error

Table 14. Within- and between-group differences from baseline in Gait speed (GS) and Gait speed without walking aids (GS-noWA) according to support in the test.

Within-Group Difference Interaction1

Measure N Exercise, Mean m/s (SE)

N Control, Mean m/s (SE)

Mean m/s (95% CI) P

GS: 4 months Unsupported 15 0.025 (0.033) 23 –0.049 (0.027) 0.104 (0.008 to 0.200) .034 With Walking Aid 60 –0.031 (0.016) 55 –0.002 (0.017) 7 months Unsupported 14 0.015 (0.034) 22 –0.111 (0.027) 0.166 (0.068 to 0.264) .001 With Walking Aid 50 –0.041 (0.028) 47 –0.001 (0.018) GS-noWA: 4 months Unsupported 42 0.023 (0.021) 46 –0.031 (0.020) 0.038 (–0.032 to 0.156) .199 With Min Support 26 -0.021 (0.026) 22 –0.013 (0.028) 7 months Unsupported 32 –0.006 (0.023) 42 –0.072 (0.021) 0.106 (0.013 to 0.221) .029 With Min Support 21 –0.045 (0.028) 15 0.005 (0.032)

Values are from linear mixed effects models of the complete sample (n = 186) adjusted for age, sex, and antidepressant use. 1 Difference between exercise and control groups according to participants that walked unsupported compared with a walking aid or minimum support. A positive mean value indicates a greater effect in favour of participants that walked unsupported. Min = minimum; N = number of participants with complete data.

38

Figure 8. Changes in gait speed (GS) and gait speed without walking aids (GS-noWA) (A-B) and according support in the test (C–D). Values are least square mean change

from baseline, with 95% confidence intervals, from linear mixed effects models adjusted for age, sex, and antidepressant use. ES = effect size; Min = minimum

39

Table 15. Within-group and between-group differences from baseline in the Mini-Mental State Examination (MMSE), Verbal fluency (VF) and, Alzheimer’s Disease Assessment Scale-

Cognitive subscale (ADAS-Cog)

Measure Within-group difference Between-group difference

N Exercise, mean (SE)

N Control, mean (SE)

Mean (95% CI) P

MMSE: 4 months 81 –1.15 (0.41) 85 –0.93 (0.40) –0.27 (–1.4 to 0.87) .644 7 months 75 –2.25 (0.42) 76 –1.11 (0.42) –1.15 (–2.32 to 0.03) .056 VF: 4 months 80 –0.74 (0.32) 81 –0.21 (0.32) –0.53 (–1.42 to 0.35) .241 7 months 74 –0.89 (0.33) 72 –0.71(0.33) –0.18 (–1.09 to 0.74) .707 ADAS-Cog 4 months 84 1.51 (1.06) 82 2.55 (1.07) –1.04 (–4.00 to 1.92) .491

Values are from linear mixed effects models of the complete sample (n=186) adjusted for age, sex, and antidepressant use. CI = confidence interval; N = number of participants with complete data; SE = standard error

Table 16. Within-group differences from baseline in the Mini-Mental State Examination (MMSE), Verbal fluency (VF), and Alzheimer’s Disease Assessment Scale - Cognitive

subscale (ADAS-Cog), and differences in exercise effect according to cognitive function at baseline

Within-group difference Interaction1

Measure N Exercise, Mean (SE)

N Control, Mean (SE)

Mean (95% CI) P

MMSE: 4 months Lower (≤ 14) 35 –1.75 (0.62) 47 –0.64 (0.54) 1.60 (–0.65 to 3.85) .165 Higher (≥15) 46 –0.78 (0.54) 38 –1.26 (0.60) 7 months Lower (≤ 14) 33 –3.62 (0.63) 41 –0.87 (0.56) 2.86 (0.54 to 5.18) .016 Higher (≥15) 42 –1.17 (0.56) 35 –1.27 (0.61) VF: 4 months Lower (≤5) 34 0.54 (0.47) 38 1.09 (0.44) 0.23 (–1.47 to 1.93) .793 Higher (≥6) 46 –1.64 (0.40) 44 -1.32(0.41) 7 months Lower (≤5) 31 0.03 (0.49) 33 0.16 (0.47) 0.04 (–1.73 to 1.80) .968 Higher (≥6) 43 -1.56 (0.41) 40 –1.46 (0.43) ADAS-Cog: 4 months Lower (≥31) 38 1.34 (1.56) 44 1.97 (1.46) –0.50 (–6.41 to 5.40) .868 Higher (≤30) 46 2.04 (1.43) 38 3.18 (1.57)

Values are from linear mixed effects models of the complete sample (n = 186) adjusted for age, sex, and antidepressant use as fixed effects. 1Difference in exercise effect between participants with lower and higher cognitive function at baseline. CI = confidence interval; N = number of participants with complete data; SE = standard error.

40

Figu

re 9

. Bet

wee

n-gr

oup

diff

eren

ces

from

bas

elin

e in

Min

i-Men

tal S

tate

Exa

min

atio

n (M

MSE

), V

erba

l flu

ency

(VF)

, and

Alz

heim

er’s

Dis

ease

Ass

essm

ent S

cale

– C

ogni

tive

Subs

cale

. Val

ues

are

leas

t squ

are

mea

n ch

ange

from

bas

elin

e, w

ith 9

5% c

onfid

ence

inte

rval

s,

from

line

ar m

ixed

eff

ects

mod

els

adju

sted

for a

ge, s

ex, a

nd a

ntid

epre

ssan

t use

.

41

DISCUSSION

The results in this thesis indicate that usual gait speed is a predictor of five-year all-cause mortality in very old people, including those dependent in ADLs and with dementia disorders. The association between lower gait speed and greater mortality remained significant when adjusted for potential confounders relevant in this population. Inability to complete the test or a usual gait speed below 0.5 m/s was an independent predictor of increased mortality compared to faster gait speeds. Dementia disorder, dependency in ADLs, sex, age, and use of walking aids in the gait speed test did not influence the association. Effects of a four-month high intensity functional exercise program differed between participants with AD and those with other types of dementia living in nursing homes. In participants with non-AD dementia, the exercise program lessened decline in ability to carry out ADLs and improved balance at four and seven months. No such effect was evident in participants with AD. The effects of exercise on balance differed according to cognitive level at seven months only, in favour of participants with higher cognitive levels. In this study, where the majority of participants habitually used walking aids, effects of the exercise program on gait speed differed according to amount of support used in the test. Exercise had positive effects on gait speed in participants that tested unsupported compared with when walking aids or living support was used. The effects on global cognitive or executive function did not differ between exercise and attention control activity. The effects of exercise on gait speed or cognitive function did not differ according to sex, cognitive level or dementia type.

GAIT SPEED IN VERY OLD AGE AND IN DEMENTIA

In the total sample of very old particicpants in Paper I, women had a lower gait speed than men, 0.49 m/s and 0.58 m/s, respectively. Community dwelling participants, with or without dementia, walked with a gait speed of 0.56 m/s. In comparison, a large meta-analysis of healthy community dwelling older individuals with mean age of 73.5 years observed a gait speed of 0.92 m/s.25 Furthermore, normative values suggested for community dwelling older people at the age of 80 years or older are 0.94 and 0.97 m/s respectively for women and men,22 and may need to be revised. In Paper I, participants that lived in nursing homes had a usual gait speed of 0.41 m/s, which seems lower than the 0.48 m/s found in a systematic review of older individuals over 70 years of age in nursing home settings,14 but still conceivable considering the age difference. People with dementia walked with gait speeds of 0.38 m/s and 0.47 m/s in Paper I and Paper III, respectively, which is in line with studies of people with dementia in nursing homes.125,126 The lower gait speed in Paper I may relate to

42

cognitive impairment, since participants with MMSE score below ten at baseline were excluded in Paper III.

Several gait speed cut-offs to diagnose mortality risk have been proposed. However, recommended cut-offs of 1.0 m/s26 or 0.8 m/s25,171 to determine above average mortality risk may have limited use in very old people as the majority have lower gait speeds than that. In Paper I, 9% walked at a gait speed of 0.8 m/s or above and 2% at 1.0 m/s or above. A lower cut-off may be more appropriate in very old populations. Results from the present study suggest that a difference in mortality risk in this population occurs at approximately 0.5 m/s. Gait speed of 0.6 m/s has been suggested as a cut-off to predict further functional decline in older people that are already impaired,26 which better corresponds with the critical gait speed value found in the present study.

GAIT SPEED PREDICTS MORTALITY IN VERY OLD PEOPLE

In Paper I, gait speed remained a predictor of mortality even when extensively adjusted for confounders. The result is in accordance with previous studies of healthy, community dwelling individuals with mean age below 80 years.25,171,172 In the field of research with focus on very old people, two studies have evaluated the association between gait speed and mortality risk. In contrast to the present study, neither found gait speed to be an independent predictor. The first, a population-based study among community dwelling individuals over the age of 80 years, did not reach a significant association between usual gait speed and two-year mortality when comprehensive adjustments were made.173 However, statistical power might have been limited by the smaller sample size of 335 participants. The second, a population-based study included individuals aged 85 years.174 After comprehensive adjustments, the association between maximum gait speed and two- and twelve-year mortality risk lost significance. One possible explanation as to the reason gait speed failed to independently predict mortality in that study, may be that maximum gait speed in comparison to usual gait speed is less sensitive in capturing symptoms of subclinical disease or biological ageing in very old people.

Several medical conditions were adjusted for in the present study, including heart failure, cardiovascular disease and malignancy, and still the association between lower gait speeds and mortality remained significant. The results suggest that usual gait speed reflects damage to systems yet undiagnosed or difficult to define, perhaps subclinical disease or biological ageing.25,26 The ageing process is subject to much individual variation, and the ensuing physiological deterioration is not always well described solely by chronological age. Lower gait speed may, through illustrating overall organ system function,

43

provide an indication as to the degree of biological ageing. In addition, the association between gait speed and mortality could be mediated by factors not adjusted for in the present study, such as physical inactivity.8

EXERCISE EFFECTS ON DEPENDENCE IN ADLs

Although a difference in exercise effects were indicated according to dementia type, as shown in Paper II, no effects of exercise on dependence in ADLs were observed in the complete sample. This result is in line with another large study by Telenius et al.126,127 investigating effects of a three-month exercise intervention in older people with dementia living in nursing homes. No effect on ADL dependence was found when the HIFE program was compared with low-intensity exercise. In contrast, positive exercise effects on ADLs have been shown in two large RCTs.125,130 Both studies included only people with AD, and control groups received usual care. In the first study, set in nursing homes, Rolland et al.125 found effects only at twelve months, not at six months, suggesting that intervention length may be important for effects in people with AD. Similarly, in the second study, Pitkälä et al.,130 investigated effects of exercise in community dwelling older people and found positive effects at six and twelve months, but not at three months. However, neither of these two studies included an attention control group. The lack of an attention control group limits ability to draw conclusive inferences regarding exercise effects per se on ADLs, and the comparison with our results for the whole sample and the subgroup of people with AD.

THE HIFE PROGRAM AND EFFECTS ON BALANCE

The four-month HIFE program improved balance in the complete sample. The result is in accordance with the study by Telenius et al,126,127 and also with a large study by Hauer et al. set in the community.123,124 The HIFE program, evaluated in the present study and in that by Telenius et al.,126 has previously been found applicable to older populations in nursing homes.165 The design of the HIFE program follows exercise recommendations aimed at increasing effects on muscle strength, balance and gait, and follows principles of exercise such as specificity and intensity. Besides being optimal for effect on intended outcomes, including balance and dependence in ADLs, that exercises were functional is perhaps particularly important in a population of people with dementia. Functional exercises, typical to functional balance and ADLs, such as rising from a chair, climbing stairs and walking, may be easier to recognise and understand for individuals with severe cognitive impairment, compared with exercises in a leg press, treadmill or other gym machines. Furthermore, the portable equipment lends itself to exercise programs set in nursing homes,

44

where available space and gym equipment may be limited. The HIFE program seems appropriate for improvement of balance in the functional activities included in ADLs.

The observation of larger effects on balance than on dependence in ADLs is in line with findings from several previous studies of older people with low cognitive function living in nursing homes.116,118,126,127,175-177 Dependence in ADLs is multifactorial, with various compositions and causes that may not be equally predisposed to change. For example, although better balance may improve overall performance in ADLs, activities such as feeding and bladder control may be less influenced by the improvement. In addition, the application of improved balance to reduce level of assistance, as measured by the FIM and Barthel ADL Index, relies on the responsiveness of care staff, who may be limited by routines and time constraints. The contrast in effect between dependence in ADLs and balance may thus be inherent to differences in the measurement scales used, whereas the BBS measures what individuals can do, the FIM and Barthel ADL Index measures what they actually do.

DIFFERENCES BETWEEN DEMENTIA TYPES

The results in Paper II indicate that exercise improves balance and attenuates decline in ADL dependence in participants with non-AD when compared with AD, which has not been shown previously. The cause of lessened ability to perform ADLs varies between individuals. Impaired cognitive function, but also impaired physical function could be the cause, and clinical symptoms typical of certain dementia types may influence responses to exercise programs. The absence of a positive exercise response on ADL dependence and balance in participants with AD could reflect difficulties in motor learning. Studies of people with AD suggest that in order to learn motor skills, frequent and consistent practice of the specific skill may be important since skill transfer is limited.120-122 Less is known about motor skill learning in people with non-AD types, such as Vascular dementia. However, memory impairment is often less pronounced in people with Vascular dementia than in those with AD,54,178 which could indicate better abilities to learn and transfer learned skills. In addition, the difference in exercise effects between participants with AD and those with other dementia types may be explained by baseline differences between subtypes; for example, participants with non-AD had higher cognitive function than those with AD. The larger effect of exercise on balance seen in participants with higher cognitive function reinforces the potential moderating effect of cognitive function. Furthermore, considering that 82% of participants with non-AD had dementia of vascular origin or dementia with a vascular component, the exercise intervention could have affected vascular risk factors.

45

Moreover, the effects in participants with non-AD may be explained by principles of neuroplasticity following brain injuries, whereby targeted rehabilitation amplifies neurorecovery.179 Indeed, evidence suggests that high-intensity functional exercise in older people following a stroke has beneficial effects on gait and dependence in ADLs.180

In Paper II, in participants with AD and lower cognitive function, negative exercise effects were indicated on dependence in ADLs and balance at the seven-month follow-up. Considering that at this follow-up three months had passed since the end of the intervention, it is difficult to speculate about potential causes. Furthermore, sensitivity analyses revealed extreme values for some participants in the change over time in dependence in ADLs and balance, which influenced results. No significant negative exercise effects remained after removing the extreme values from the analyses. The extreme values showed decline in dependence in ADLs, and one in particular recurred throughout all measures and was caused by the participant first suffering a stroke then a hip fracture during the intervention period.

WALKING AIDS AND EXERCISE EFFECTS ON GAIT SPEED

The positive exercise effects on gait speed in participants that walked unsupported are in line with a study by Rolland et al.125, which included only ambulant people with dementia living in nursing homes; although the comparison is limited since walking aid use was not reported. In addition, a study by Hauer et al.123 included community dwelling participants with dementia that could walk unaided, and found that a three-month high-intensity progressive resistance training in combination with functional exercises had positive effects on maximal gait speed when compared with low intensity exercise.123,124 The absence of positive exercise effects in participants that walked with a walking aid in the present study are comparable with the study by Telenius et al.126 that showed improvement to balance but not to gait. Unfortunately, while the study similarly to ours had a high proportion of participants that used walking aids, its influence on exercise effects was not analysed.

The results in Paper III suggest that walking aids in the gait speed test may reduce responsiveness and conceal exercise effects. A reduced responsiveness with walking aid use was also shown in a study set in a geriatric inpatient ward after rehabilitation following a hip fracture.53 In the present study, positive exercise effects were shown in participants who walked without walking aids. When gait speed was tested without walking aids and with minimum living

46

support positive interaction effects persisted. Indeed, reduced responsiveness may explain the absence of positive exercise effects also in participants unable to test without walking aids or living support. That walking aids reduce responsiveness and concealed effects is further substantiated by results from the interaction analysis on balance, where positive exercise effects were observed irrespective of use of walking aid or support while walking.

EXERCISE EFFECTS ON COGNITIVE FUNCTION

In Paper IV, exercise had no effect on global or executive function. This result concurs with studies by Telenius et al.126 and Öhman et al.141, which also compared exercise interventions that comprised multiple components (strength, balance, and walking or other aerobic exercise. Telenius et al.126 found no effects of a three-month intervention, comparing the HIFE program with a low-intensity exercise control activity. In addition, Öhman et al.141 found no effects on global cognitive function of a twelve-month home-based exercise program when compared with usual care. The absence of effects may be attributed to a relatively limited aerobic component in the aforementioned exercise programs, as cardiovascular improvement has been proposed to mediate effects on cognitive function.110 However, results of our study also agrees with another large study in the community setting, where a four-month moderate-to-high-intensity aerobic exercise program had no effect on global cognitive function when compared with usual care in intention-to-treat analyses,143 even though cardiorespiratory fitness was shown to improve.181

For effect on cognitive function the length of the exercise intervention may be critical, and the four-month exercise intervention was perhaps too short. Dependent on type of exercise and intensity, interventions that last six to twelve months have been proposed for observable effects on brain structure.109 Indeed, some longer trials in people with dementia show positive effects,139,141,142 though inferences are limited by methodological constraints. Furthermore, the observation of structural changes in the brain may not necessarily equal corresponding effects on cognitive function,182 which may take even longer. In addition, certain cognitive domains such as executive function may be particularly sensitive to exercise.108 In people at risk of cognitive decline, the longest and largest trial to date found that a two-year multicomponent exercise program had effect on executive function, albeit only in participants aged 80 years and over, and those with functional impairment.114 In addition, Öhman et al.141 observed exercise effects, although limited, on executive function. However, this effect may perhaps be attributed to the dual task training component in their exercise program.

47

In Paper IV, negative exercise effects were indicated on cognition that approached statistical significance in MMSE at the seven-month follow-up, particularly in participants with lower cognitive function at baseline. Given the demand and challenge that high-intensity exercise encompasses, it is possible that negative effects on cognitive function is a consequence of stress-induced higher cortisol levels.183 However, since three months had transpired since the end of the intervention, causal inferences are limited. Furthermore, the result was not substantiated in analyses on VF or ADAS-Cog.

METHODOLOGICAL CONSIDERATIONS

The cohort in the GERDA study constitute to one of its strengths, incorporating very old people, including those in nursing homes, with dependence in ADLs, and severe dementia, who are often excluded from research. The study was representative of the three age groups, 85-, 90- and 95-year-olds in the studied geographical area, but perhaps not of the total population of very old people. Not surprisingly, given the age of the population, a high mortality rate was apparent in the GERDA study. Therefore in each cohort year, the data collection began in participants of the oldest age group (aged 95+) to increase likelihood of participation. Those eligible that declined participation did not differ in age or sex from those that were included in Paper I. Also in the UMDEX study the population constitutes a strength, since both people with dementia and people living in nursing homes are generally understudied. Generalisation of results was improved by inclusion criteria that allowed recruitment of a population with diverse functional abilities, co-morbidities, and age. Still, the number of men in the UMDEX study was few compared with women. This is typical to research in older populations, since women generally live longer than men. Furthermore, more men compared to women declined the invitation to participate. The reason for this is not known, but reduced the proportion of men in the UMDEX study, with risk of selection bias. No differences in exercise effect according to sex were found in any interaction analyses.

Losses to follow-up in the UMDEX study were expected, considering the age and morbidity of the population, and taken into account in power calculations of sample size. In keeping with the intention-to-treat principle, applied to reduce selection bias, the statistical method, linear mixed effects models, was chosen to allow inclusion of all available measurements from all participants in analyses. Besides participants passing away during the intervention and follow-up period, two participants were inadvertently excluded from follow-up assessments due to one participant’s relocation to own home for an extended period of the intervention and the withdrawal of a physician’s medical approval

48

for another participant to participate in exercise. Missing values, and reasons for missingness, were reported in detail in all papers. In Paper III, missing gait speed values were imputed with multiple imputation method, and based on reasons for missingness, which strengthens analyses.170 Multiple imputation, where plausible values to replace the missing ones are predicted through several regression models, may result in less biased results since the method incorporates the uncertainty of the imputed values.162,184

Paper I was an observational cohort. Although causative evidence is low in observational studies, the design offers an indication of the temporal association between gait speed and mortality. In observational studies, confounding variables are particularly problematic, and may lead to bias. A confounding variable is associated with both variable and outcome of interest, and may distort results. The adjustment of potential confounding variables increases the quality of the evidence.185 The extensive data collection in the GERDA study allowed for multivariate analyses with comprehensive adjustment for many potentially confounding factors bearing in mind the characteristics of this particular age range, which strengthens the result in the study. Unfortunately, Paper I lacked a measure for physical activity levels, which could be an important confounder in the association between gait speed and mortality.

The process by which adjusting variables were selected was described, which improves transparency and external validity.186 Analyses were adjusted for preselected baseline variables, identified as potential confounders in each paper, based on previous evidence and clinical experience. In Paper I, adjusting variables were selected if associated with mortality. In Papers II-IV, adjusting variables were selected if significantly different between groups at baseline, and further if strongly associated with change in the outcome variable of interest. This was a pre-defined strategy considering that the cluster design may increase risk of imbalance between groups.187 To choose adjusting variables based on imbalances only has been criticized, since unless the variable is a predictor for the outcome of interest it is unlikely to be influential.188 Conversely, baseline variables with strong association to the outcome may influence the result even if no significant imbalance between groups exists.189

Usual gait speed was tested over 2.4, and 4.0 meters and with habitual walking aid if any, and without walking aid and with minimum support. In all tests a digital stopwatch was used to measure time. Other ways to analyse gait include optical motion cameras or electronic mats, which provide additional data on gait parameters, and may be a more accurate measure of gait speed. However, using a stopwatch is quick, cheap, and user friendly, and may benefit implementation in clinic. As inclusion criteria in the GERDA study and

49

UMDEX study were comparatively generous, floor effects of the gait speed tests were evident, i.e. several participants could not perform the tests. In Paper III, participants unable to perform the test without their usual walking aid were offered a standardised minimum amount of living support in the gait speed test without walking aids and with minimum support; this type of support has not been tested for reliability. However, to increase reliability, trained testers using detailed instructions conducted the gait speed test in a structured manner. The use of standing still start in the gait speed tests may have resulted in lower gait speeds measured. The acceleration phase of the gait speed test has been associated with a greater cognitive load and a slower pace compared with tests of steady-state gait speed.190 In addition, some studies on healthy older adults show larger exercise effects on fast gait speed compared with usual gait speed, and fast gait speed may have been interesting to measure.94

The thorough data collection in both GERDA study and UMDEX study, which included structured interviews and assessments, and supported by a review of medical records, reduced the risk of recall bias. The diagnoses of dementia were obtained from medical records, prescribed medications, and results in assessments, and further reviewed by a specialist in geriatric medicine to reduce misclassification bias. In Papers II-IV dementia types were dichotomised into AD versus all other types of dementia (non-AD) to aid comparison with previous studies, mainly conducted in people with AD. Participants with Mixed Alzheimer’s and Vascular dementia were included in the non-AD group since unadjusted within-group analyses indicated similar directional effects. Subsequently, although the term non-AD is not a completely accurate description of this group, the dichotomisation is comparable with previous studies analysing the association between gait speed and dementia types.40,41 Although baseline tests of MMSE were conducted in connection with screening, which was on average one month prior to ADAS-Cog and VF tests, the time lapse between tests did not differ between groups. The ADAS-Cog was only tested at baseline and four-month follow-up due to limited funding, which was unfortunate for comparison of effects between tests; still all three cognitive tests indicated similar results at four months.

Multiple analyses conducted on the same data increase the risk of false positive results, namely type I errors, and is reduced by pre-specification of planned analyses.188 In Paper I subgroup analyses on age, sex, dementia, dependence in ADLs, and walking aid use were pre-specified. In addition, in the UMDEX study all analyses on outcome measures, the Barthel ADL Index, FIM, BBS, Gait speed with and without walking aids, MMSE, ADAS-Cog and, VF, and exploratory analyses on subgroups age, sex, dementia type, and cognitive level at baseline in Papers II-IV were pre-specified. Interaction analyses were used in all papers to explore differences between subgroups, rather than reliance on P

50

values for between group differences in subgroups only, which increases risk for type I or II errors.168 In Paper II, that the interaction analyses indicated differences according to dementia type in both outcome measures for dependence in ADLs, and on balance substantiates drawn inferences. Similar patterns of effect in outcome measures for cognitive function and gait speed were evident in Paper III and IV. However, interaction analyses could still lack statistical power, and the results merit further investigation.168

The UMDEX study evaluated effects using the HIFE program, which beside a compilation of exercises, comprises a scale defining high, medium, and low intensity of strength and balance exercises, together with a model for selection of exercises according to functional deficits.164,165 The structured program improves potential to replicate the results of this study clinically or for research purposes. In addition, the selection model is an effective indicator of suitable level of difficulty, from which progression is possible. The attention control activity was also structured, with activities based around topics considered interesting for older people in order to ensure adherence comparable with the exercise group. As such, some components of the control activity resembles cognitive stimulation, which has been shown to improve cognitive function and well being, but not dependence in ADLs in people with dementia.191 Cognitive stimulation involves engagement in activities that aim to enhance cognitive and social function.191 Although the control activity in the UMDEX study did not aim to enhance cognitive function, a certain effect cannot be ruled out.

ETHICAL CONSIDERATIONS

Medical research involving humans comprises careful consideration; whether the benefits that research may bring counteracts the risks and burdens entailed, and guided by ethical principles outlined in the Declaration of Helsinki. The UMDEX study protocol (ISRCTN31767087) is published on the ISRCTN registry website (www.isrctn.com) to improve transparency in conduct and reporting, hence reduce risk of transgression and bias. Regional ethics review boards approved all studies in this thesis.

In very old adults, including older adults with dementia, physical and cognitive impairments are common, as well as dependence in ADLs. In these populations effective prevention and disease modifying treatments may be particularly beneficial. Nevertheless, very old people and people with dementia are often underrepresented in research. Since the populations studied in this thesis included adults with reduced capacity to understand or consent to particulars of research, special consideration to protect their health and rights were taken.

51

All participants in the GERDA and UMDEX studies were provided with a written letter outlining what participation in studies encompassed, including protection of privacy and the right to withdrawal at any point without explanation. As stated in the ethical approval for the UMDEX study, an informed consent from all participants could not be expected, due to cognitive impairment. However, information was provided as far as possible, by not only providing the letter but also orally presenting the content to the eligible participant. The letter was also provided to their relatives or otherwise authorized representatives, whose agreement was a requisite to participation. In the GERDA study relatives or otherwise authorized representatives were consulted in case of cognitive decline.

Testers in the GERDA and UMDEX studies had medical training, and were able to consult a geriatric specialist if severe depression, or other ill health was suspected. In the UMDEX study all PTs and OTs/OT assistant involved in the intervention had experience in working with people with cognitive impairment. Furthermore, participation in the UMDEX study required a medical consent from the participants’ physician. Activity leaders were encouraged to obtain updates on participants’ health status before activities, and were able to contact physicians or nurses when necessary. In addition, harms reported by participants, or observed by activity leaders, in relation to exercise session were recorded and analysed.

In the GERDA study, participation could encompass an enjoyable moment to converse or reminisce, which can be valuable in a population where many live alone or have reduced social interactions. The interview sometimes required repeated sessions, which often included informal chats over a cup of coffee. In the UMDEX study the exercise activity was compared with an attention control activity. To promote attendance and to compensate for the benefits of participation in the exercise group, care was taken to develop an enjoyable and stimulating control activity.

CLINICAL IMPLICATIONS

As results from Paper I show, gait speed appears to predict mortality in very old populations where dependency in ADLs and use of walking aids are common. The population-based study concentrated on a population at the age of 85 years and over, which the comparatively large proportion of participants that was ADL dependent and had dementia disorders, either living in residential care facilities or in the community show. None of the subgroups age, sex, ADL dependence, use of walking aids, or dementia interacted with gait speed to alter its association to mortality. The prevalence of dementia was as

52

high as 33%, as expected in a very old population, and included individuals with MMSE scores as low as zero. The results in the present study suggest that gait speed predicts mortality regardless of dementia disorders. Since communication and language problems often accompany dementia, gait speed may be a particularly useful tool for health care clinicians in the assessment of individuals of this subgroup.

The results from Paper II, the effect of exercise on dependence in ADLs and balance, support the notion that dementia should not be considered a single disease entity, but rather to constitute separate disorders with clinical symptoms that may require different strategies to optimize symptom management.192 The observed effects of exercise on FIM (3.5 points) and on Barthel ADL Index (1.4 points) corresponds to effects sizes of 0.36 and 0.50, respectively, in participants with non-AD dementia, which we propose should be considered clinically meaningful. A one-point difference on the Barthel ADL Index, a scale described as rather crude, can reflect meaningful change in an individual’s level of independence. A one-point improvement in FIM score has been related to timesaving in the care of older patients with stroke.193,194

The result from Paper III, the effects of exercise on gait speed, may be of value both in clinical and research settings when interested in best medical practice concerning care of older people with dementia living in nursing homes, as well as, when effectiveness of rehabilitation is measured in populations where walking aid use are common. The exercise effects in participants that walked unsupported at four and seven months, 0.07 m/s and 0.13 m/s in the GS test and, 0.05 m/s and 0.07 m/s in the GS-noWA test, seem clinically meaningful. In comparison, small and substantial meaningful changes in gait speed has on an individual level been reported to correspond with 0.05 m/s and 0.10 m/s, respectively, in older people with mild to moderate mobility disability.195 Seeing as walking aids appears to conceal changes over time in gait speed, it may be important to measure gait unsupported in order to fully understand the progression and severity of the impairment, and the association to health related outcomes such as falls and survival.

The result from Paper IV, the exercise effects on cognitive function, agree with that of a previous systematic review and recent large exercise studies in people with dementia. Despite the limited evidence of beneficial exercise effects on cognitive function, the activity limitation in people with dementia may still be attenuated by physical exercise. Dependence in ADLs is influenced not only by cognitive function but also physical function. Exercise programs may, through improved muscle strength, balance and gait, ameliorate physical function, and therefore also dependence in ADLs.90,128 For this purpose, a four-month high-intensity functional exercise program seems effective in people with dementia

53

living in nursing homes; as shown by the effects on ADL dependence in people with non-AD, and by effects on balance and on gait speed in those that walked unaided. In this population it seems likely that attenuation of dependence in ADLs is predominantly associated with exercise effects on physical function rather than cognitive function.

IMPLICATIONS FOR FUTURE RESEARCH

To aid clinicians in the management of people with low gait speeds, future studies to further illuminate causes of the independent association between usual gait speed and mortality may be of interest. Further, it seems important to investigate the effect of improved gait speed on survival, in very old people and in people with dementia. Since walking aid use seems to conceal exercise effects on gait speed, it may be of interest to further compare gait speed, conducted with walking aids and without, in relation to mortality or falls.

Validation of whether effects of exercise differ dependent on dementia type needs to be investigated in further studies. Although durations of up to three months have been observed in most previous exercise interventions,90 the length of the current intervention was four months (40 sessions in total) to augment the effects of exercise considering that people with dementia face differences in motor skill learning. Future studies may want to compare exercise effects between dementia types in longer interventions, lasting at least twelve months, and with designs using attention control.

Future studies of exercise effects on cognitive function may want to consider interventions that last at least six months but preferably longer, to allow for effects on function. Study designs that compare high-intensity exercise with low-intensity exercise would be of interest to evaluate optimal exercise intensity. To explore ideal intervention length, pre- and post tests, intermediate, as well as adequate follow-up tests would be required, preferably in combination with brain images.

54

CONCLUSIONS

Among people aged 85 or over, including people dependent in ADLs and with dementia disorders, usual gait speed was an independent predictor of five-year all-cause mortality. Inability to complete the gait test or a gait speed below 0.5 m/s appears to be associated with higher five-year mortality risk. Gait speed subsequently seems to be a useful clinical indicator of health status or biological ageing among very old people. In older people with mild to moderate dementia, dependent in ADLs, and living in nursing homes, a four-month high-intensity functional exercise program appeared to attenuate further loss of dependence in ADLs and improve balance, albeit only in participants with non-AD type of dementia. Further studies are needed to validate this result. In participants with AD, the intervention seems to have had no such effect. In this population of older people with dementia living in nursing homes, where a majority used walking aids and the mean gait speed was below 0.5 m/s, the exercise program had positive effects on gait in participants that tested unsupported compared with when walking aids or living support was used. The study suggests that the use of walking aids in the gait speed test may conceal exercise effects. The exercise program had no superior effects on global cognition or executive function compared with an attention control activity. This thesis suggests that, in older people with dementia, exercise effects on physical function rather than cognitive function may explain effects on dependence in ADLs.

55

ACKNOWLEDGEMENTS

This work was carried out at the Department of Community Medicine and Rehabilitation, Geriatric medicine and Physiotherapy, Umeå University. I want to extend my warmest gratitude to all who has contributed to this thesis in various ways along the journey. I like to take this opportunity to thank:

Erik Rosendahl, my supervisor, for your support and advice, always delivered in a most gentle and reassuring manner; for listening, and for your encouragement, patience, and help in the exploration of new pathways.

Håkan Littbrand, my supervisor, for excellent discussions and steadfast encouragement; for always taking the time to answer questions or pointing me in the right direction, and for terrific company and collaboration in unchartered statistical and methodological territories.

Yngve Gustafson, my supervisor, for introducing me to research; for contributing with extensive knowledge and experience, and for your inspirational engagement in the welfare of older people.

Henrik Holmberg, statistician, for a great cooperation and help with analyses; for your patient explanations, open mind and relaxed ways, and for finding a shared statistical language.

Lillemor Lundin-Olsson and Peter Nordström for contributing with extensive knowledge and experience, and for superb advice and guidance.

Carl Hörnsten, Gustaf Boström, and Nina Lindelöf, co-authors, for your valuable contribution of knowledge and experience.

Robert Wiklund and Mia Conradsson, for your enjoyable company and support during shared travels, offices, and even chairs.

Colleagues at Units of Physiotherapy and Geriatric medicine, past and present, for stimulating conversations on diverse subjects during coffee breaks, for shared knowledge and learning, good discussions, and great company.

Contributors to data collection of the GERDA study and the UMDEX study. The helpful care staff, and the wonderful participants, who made the data collection and intervention a pleasure and a memory for life.

56

Friends, nearby and far away, my fellow disco divas, skiers, surfers, dreamers, lads and ladettes, for enjoyable times and diversions along the way, and for your easy going, lasting friendships.

Lilian, my mother, for your unwavering support, and my beloved sisters Camilla and Jessica, and brother Johan for a lifelong company, great counsel, and happy family gatherings.

Simon, my partner, for believing in me; for encouraging me to break my boundaries, and for being a great friend. Maia, Hannah and Johannes, my three wonderful children, for being exactly who you are. I love you to the moon and back!

This work was supported by the Swedish Research Council (grant numbers K2009-69P-21298-01-4, K2009-69X-21299-01-1, K2009-69P-21298-04-4, K2014-99X-22610-01-6); Forte – Swedish Research Council for Health, Working Life and Welfare (formerly FAS – Swedish Council for Working Life and Social Research); the Vårdal Foundation; the Swedish Dementia Association; the Promobilia Foundation; the Swedish Society of Medicine; the Swedish Alzheimer Foundation; the King Gustav V and Queen Victoria’s Foundation of Freemasons; the European Union Bothnia-Atlantica Program; the County Council of Västerbotten, the Umeå University Foundation for Medical Research; the Ragnhild and Einar Lundström’s Memorial Foundation; and the Erik and Anne-Marie Detlof´s Foundation.

57

REFERENCES

1. United Nations. Department of Economic and Social Affairs, Population

Division (2013). World Population Ageing ST/ESA/SERA/348 2013.

2. Statistics Sweden. The future population of Sweden 2015–2060 [publication in Swedish]. Örebro, Sweden, 2015.

3. von Heideken Wågert P, Gustavsson JM, Lundin-Olsson L, et al. Health status in the oldest old. Age and sex differences in the Umeå 85+ Study. Ageing Clin Exp Res 2006;18:116-126.

4. von Heideken Wagert P, Gustafson Y, Lundin-Olsson L. Large variations in walking, standing up from a chair, and balance in women and men over 85 years: an observational study. Aust J Physiother 2009;55:39-45.

5. Nyberg L, Lovden M, Riklund K, et al. Memory aging and brain maintenance. Trends Cogn Sci 2012;16:292-305.

6. Marengoni A, Winblad B, Karp A, et al. Prevalence of chronic diseases and multimorbidity among the elderly population in Sweden. Am J Public Health 2008;98:1198-1200.

7. World Health Organization. Global status report on noncommunicable diseases 2014. Geneva, Switzerland: World Health Organization, 2014.

8. Lee IM, Shiroma EJ, Lobelo F, et al. Effect of physical inactivity on major non-communicable diseases worldwide: an analysis of burden of disease and life expectancy. Lancet 2012;380:219-229.

9. World Health Organization. Global recommendations on physical activity for health. Geneva, Switzerland: World Health Organization, 2010.

10. Sanford AM, Orrell M, Tolson D, et al. An International Definition for "Nursing Home". Journal of the American Medical Directors Association 2015;16:181-184.

11. World Health Organization. World report on ageing and health. Geneva Switzerland: World Health Organization,, 2015.

12. Government Offices of Sweden. Den ljusnande framtid är vård: Delresultat från LEV-projektet [publication in Swedish]. Sweden: Government Offices of Sweden, 2010.

13. The National Board of Health and Welfare. Vård och omsorg om äldre: lägesrapport 2016 [publication in Swedish], 2016.

14. Kuys SS, Peel NM, Klein K, et al. Gait speed in ambulant older people in long term care: a systematic review and meta-analysis. J Am Med Dir Assoc 2014;15:194-200.

15. Bjork S, Juthberg C, Lindkvist M, et al. Exploring the prevalence and variance of cognitive impairment, pain, neuropsychiatric symptoms and ADL

58

dependency among persons living in nursing homes; a cross-sectional study. Bmc Geriatrics 2016;16.

16. Crocker T, Young J, Forster A, et al. The effect of physical rehabilitation on activities of daily living in older residents of long-term care facilities: systematic review with meta-analysis. Age Ageing 2013;42:682-688.

17. Rolland Y, Van Kan GA, Hermabessiere S, et al. Descriptive study of nursing home residents from the REHPA network. Journal of Nutrition Health & Aging 2009;13:679-683.

18. de Souto Barreto P, Lapeyre-Mestre M, Mathieu C, et al. A multicentric individually-tailored controlled trial of education and professional support to nursing home staff: research protocol and baseline data of the IQUARE study. J Nutr Health Aging 2013;17:173-178.

19. Simonsick EM, Kasper JD, Phillips CL. Physical disability and social interaction: factors associated with low social contact and home confinement in disabled older women (The Women's Health and Aging Study). J Gerontol B Psychol Sci Soc Sci 1998;53:S209-217.

20. den Ouden M, Bleijlevens MHC, Meijers JMM, et al. Daily (In) Activities of Nursing Home Residents in Their Wards: An Observation Study. Journal of the American Medical Directors Association 2015;16:963-968.

21. Bohannon RW. Comfortable and maximum walking speed of adults aged 20-79 years: reference values and determinants. Age Ageing 1997;26:15-19.

22. Bohannon RW, Williams Andrews A. Normal walking speed: a descriptive meta-analysis. Physiotherapy 2011;97:182-189.

23. Ferrucci L, Bandinelli S, Benvenuti E, et al. Subsystems contributing to the decline in ability to walk: bridging the gap between epidemiology and geriatric practice in the InCHIANTI study. J Am Geriatr Soc 2000;48:1618-1625.

24. Guralnik JM, Ferrucci L, Balfour JL, et al. Progressive versus catastrophic loss of the ability to walk: implications for the prevention of mobility loss. J Am Geriatr Soc 2001;49:1463-1470.

25. Studenski S, Perera S, Patel K, et al. Gait speed and survival in older adults. JAMA 2011;305:50-58.

26. Abellan van Kan G, Rolland Y, Andrieu S, et al. Gait speed at usual pace as a predictor of adverse outcomes in community-dwelling older people: an International Academy on Nutrition and Ageing (IANA) Task Force. J Nutr Health Ageing 2009;13:881-889.

27. Cesari M, Kritchevsky SB, Newman AB, et al. Added value of physical performance measures in predicting adverse health-related events: results from the Health, Aging And Body Composition Study. J Am Geriatr Soc 2009;57:251-259.

28. Rolland Y, Lauwers-Cances V, Cesari M, et al. Physical performance measures as predictors of mortality in a cohort of community-dwelling older French women. European Journal of Epidemiology 2006;21:113-122.

59

29. Guralnik JM, Ferrucci L, Pieper CF, et al. Lower extremity function and subsequent disability: consistency across studies, predictive models, and value of gait speed alone compared with the short physical performance battery. J Gerontol A Biol Sci Med Sci 2000;55:M221-231.

30. Wennie Huang WN, Perera S, VanSwearingen J, et al. Performance measures predict onset of activity of daily living difficulty in community-dwelling older adults. J Am Geriatr Soc 2010;58:844-852.

31. Perera S, Patel KV, Rosano C, et al. Gait Speed Predicts Incident Disability: A Pooled Analysis. J Gerontol A Biol Sci Med Sci 2016;71:63-71.

32. Morie M, Reid KF, Miciek R, et al. Habitual physical activity levels are associated with performance in measures of physical function and mobility in older men. J Am Geriatr Soc 2010;58:1727-1733.

33. Middleton A, Fulk GD, Beets MW, et al. Self-Selected Walking Speed is Predictive of Daily Ambulatory Activity in Older Adults. Journal of Aging and Physical Activity 2016;24:214-222.

34. Montero-Odasso M, Schapira M, Soriano ER, et al. Gait velocity as a single predictor of adverse events in healthy seniors aged 75 years and older. J Gerontol A Biol Sci Med Sci 2005;60:1304-1309.

35. Cesari M, Kritchevsky SB, Penninx BW, et al. Prognostic value of usual gait speed in well-functioning older people--results from the Health, Aging and Body Composition Study. J Am Geriatr Soc 2005;53:1675-1680.

36. Studenski S, Perera S, Wallace D, et al. Physical performance measures in the clinical setting. J Am Geriatr Soc 2003;51:314-322.

37. Penninx BW, Ferrucci L, Leveille SG, et al. Lower extremity performance in nondisabled older persons as a predictor of subsequent hospitalization. J Gerontol A Biol Sci Med Sci 2000;55:M691-697.

38. Waite LM, Grayson DA, Piguet O, et al. Gait slowing as a predictor of incident dementia: 6-year longitudinal data from the Sydney Older Persons Study. J Neurol Sci 2005;229-230:89-93.

39. Abellan van Kan G, Rolland Y, Gillette-Guyonnet S, et al. Gait speed, body composition, and dementia. The EPIDOS-Toulouse cohort. J Gerontol A Biol Sci Med Sci 2012;67:425-432.

40. Beauchet O, Annweiler C, Callisaya ML, et al. Poor Gait Performance and Prediction of Dementia: Results From a Meta-Analysis. J Am Med Dir Assoc 2016.

41. Verghese J, Lipton RB, Hall CB, et al. Abnormality of gait as a predictor of non-Alzheimer's dementia. N Engl J Med 2002;347:1761-1768.

42. Middleton A, Fritz SL, Lusardi M. Walking speed: the functional vital sign. J Aging Phys Act 2015;23:314-322.

43. Studenski S. Bradypedia: is gait speed ready for clinical use? J Nutr Health Aging 2009;13:878-880.

60

44. Cesari M. Role of gait speed in the assessment of older patients. JAMA 2011;305:93-94.

45. Dahlin-Ivanoff S, Sonn U. Use of assistive devices in daily activities among 85-year-olds living at home focusing especially on the visually impaired. Disabil Rehabil 2004;26:1423-1430.

46. Gell NM, Wallace RB, LaCroix AZ, et al. Mobility device use in older adults and incidence of falls and worry about falling: findings from the 2011-2012 national health and aging trends study. J Am Geriatr Soc 2015;63:853-859.

47. Statistics Sweden. Health and Medical care 1980–2005 [publication in Swedish]. Stockholm, Sweden: Statistics Sweden.

48. Pellfolk T, Sandman PO, Gustafson Y, et al. Physical restraint use in institutional care of old people in Sweden in 2000 and 2007. International Psychogeriatrics 2012;24:1144-1152.

49. Bateni H, Maki BE. Assistive devices for balance and mobility: benefits, demands, and adverse consequences. Arch Phys Med Rehabil 2005;86:134-145.

50. Hardi I, Bridenbaugh SA, Gschwind YJ, et al. The effect of three different types of walking aids on spatio-temporal gait parameters in community-dwelling older adults. Aging Clin Exp Res 2014;26:221-228.

51. Muir-Hunter SW, Montero-Odasso M. Gait Cost of Using a Mobility Aid in Older Adults with Alzheimer's Disease. Journal of the American Geriatrics Society 2016;64:437-438.

52. O'Hare M, Pryde S, Gracey J. A systematic review of the evidence for the provision of walking frames for older people. Physical Therapy Reviews;18:11-23.

53. Schwenk M, Schmidt M, Pfisterer M, et al. Rollator use adversely impacts on assessment of gait and mobility during geriatric rehabilitation. J Rehabil Med 2011;43:424-429.

54. Alzheimer's Disease International. World Alzheimer Report 2015. The Global Impact of Dementia an Analysis of Prevalence, Incidence, Costs and Trends. London: Alzheimer's Disease International., 2015.

55. World Health Organization. Dementia: a public health priority. Geneva, Switzerland. ISBN 978 92 4 156445 8: World Health organization, 2012.

56. Andersen CK, Wittrup-Jensen KU, Lolk A, et al. Ability to perform activities of daily living is the main factor affecting quality of life in patients with dementia. Health Qual Life Outcomes 2004;2:52.

57. Åberg AC, Sidenvall B, Hepworth M, et al. On loss of activity and independence, adaptation improves life satisfaction in old age--a qualitative study of patients' perceptions. Qual Life Res 2005;14:1111-1125.

58. The National Board of Health and Welfare. Nationell utvärdering - Vård och omsorg vid demenssjukdom [publication in Swedish]. The National Board of Health and Welfare,, 2014.

61

59. Alzheimer's Association. 2015 Alzheimer's disease facts and figures. Alzheimers Dement 2015;11:332-384.

60. Norton S, Matthews FE, Barnes DE, et al. Potential for primary prevention of Alzheimer's disease: an analysis of population-based data. Lancet Neurol 2014;13:788-794.

61. Barnes DE, Yaffe K. The projected effect of risk factor reduction on Alzheimer's disease prevalence. Lancet Neurol 2011;10:819-828.

62. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision. Washington, DC.: American Psychiatric Association., 2000.

63. van Doorn C, Gruber-Baldini AL, Zimmerman S, et al. Dementia as a risk factor for falls and fall injuries among nursing home residents. J Am Geriatr Soc 2003;51:1213-1218.

64. Morgan D, Funk M, Crossley M, et al. The potential of gait analysis to contribute to differential diagnosis of early stage dementia: current research and future directions. Can J Aging 2007;26:19-32.

65. Allan LM, Ballard CG, Burn DJ, et al. Prevalence and severity of gait disorders in Alzheimer's and non-Alzheimer's dementias. J Am Geriatr Soc 2005;53:1681-1687.

66. van Iersel MB, Hoefsloot W, Munneke M, et al. Systematic review of quantitative clinical gait analysis in patients with dementia. Z Gerontol Geriatr 2004;37:27-32.

67. Allali G, Annweiler C, Blumen HM, et al. Gait phenotype from mild cognitive impairment to moderate dementia: results from the GOOD initiative. Eur J Neurol 2015.

68. Pettersson AF, Olsson E, Wahlund LO. Motor function in subjects with mild cognitive impairment and early Alzheimer's disease. Dement Geriatr Cogn Disord 2005;19:299-304.

69. Pettersson AF, Engardt M, Wahlund LO. Activity level and balance in subjects with mild Alzheimer's disease. Dement Geriatr Cogn Disord 2002;13:213-216.

70. Cedervall Y, Halvorsen K, Aberg AC. A longitudinal study of gait function and characteristics of gait disturbance in individuals with Alzheimer's disease. Gait Posture 2014;39:1022-1027.

71. Maquet D, Lekeu F, Warzee E, et al. Gait analysis in elderly adult patients with mild cognitive impairment and patients with mild Alzheimer's disease: simple versus dual task: a preliminary report. Clin Physiol Funct Imaging 2010;30:51-56.

72. Muir SW, Speechley M, Wells J, et al. Gait assessment in mild cognitive impairment and Alzheimer's disease: the effect of dual-task challenges across the cognitive spectrum. Gait Posture 2012;35:96-100.

62

73. Sheridan PL, Solomont J, Kowall N, et al. Influence of executive function on locomotor function: divided attention increases gait variability in Alzheimer's disease. J Am Geriatr Soc 2003;51:1633-1637.

74. Wittwer JE, Webster KE, Menz HB. A longitudinal study of measures of walking in people with Alzheimer's Disease. Gait & Posture 2010;32:113-117.

75. Hausdorff JM, Yogev G, Springer S, et al. Walking is more like catching than tapping: gait in the elderly as a complex cognitive task. Experimental Brain Research 2005;164:541-548.

76. Morris R, Lord S, Bunce J, et al. Gait and cognition: Mapping the global and discrete relationships in ageing and neurodegenerative disease. Neurosci Biobehav Rev 2016;64:326-345.

77. Montero-Odasso M, Hachinski V. Preludes to brain failure: executive dysfunction and gait disturbances. Neurol Sci 2014;35:601-604.

78. Amboni M, Barone P, Hausdorff JM. Cognitive contributions to gait and falls: evidence and implications. Mov Disord 2013;28:1520-1533.

79. Sheridan PL, Hausdorff JM. The role of higher-level cognitive function in gait: executive dysfunction contributes to fall risk in Alzheimer's disease. Dement Geriatr Cogn Disord 2007;24:125-137.

80. Sherrington C, Whitney JC, Lord SR, et al. Effective exercise for the prevention of falls: a systematic review and meta-analysis. J Am Geriatr Soc 2008;56:2234-2243.

81. Tiedemann A, Sherrington C, Close JC, et al. Exercise and Sports Science Australia position statement on exercise and falls prevention in older people. J Sci Med Sport 2011;14:489-495.

82. Stubbs B, Eggermont L, Soundy A, et al. What are the factors associated with physical activity (PA) participation in community dwelling adults with dementia? A systematic review of PA correlates. Arch Gerontol Geriatr 2014;59:195-203.

83. Lundin-Olsson L, Nyberg L, Gustafson Y. "Stops walking when talking" as a predictor of falls in elderly people. Lancet 1997;349:617.

84. Lundin-Olsson L, Nyberg L, Gustafson Y. Attention, frailty, and falls: the effect of a manual task on basic mobility. J Am Geriatr Soc 1998;46:758-761.

85. Baker NL, Cook MN, Arrighi HM, et al. Hip fracture risk and subsequent mortality among Alzheimer's disease patients in the United Kingdom, 1988-2007. Age and Ageing 2011;40:49-54.

86. Nordstrom P, Gustafson Y, Michaelsson K, et al. Length of hospital stay after hip fracture and short term risk of death after discharge: a total cohort study in Sweden. BMJ 2015;350:h696.

87. American College of Sports M, Chodzko-Zajko WJ, Proctor DN, et al. American College of Sports Medicine position stand. Exercise and physical activity for older adults. Med Sci Sports Exerc 2009;41:1510-1530.

63

88. de Souto Barreto P, Morley JE, Chodzko-Zajko W, et al. Recommendations on Physical Activity and Exercise for Older Adults Living in Long-Term Care Facilities: A Taskforce Report. J Am Med Dir Assoc 2016;17:381-392.

89. Paterson DH, Warburton DE. Physical activity and functional limitations in older adults: a systematic review related to Canada's Physical Activity Guidelines. Int J Behav Nutr Phys Act 2010;7:38.

90. Liu CJ, Latham NK. Progressive resistance strength training for improving physical function in older adults. Cochrane Database Syst Rev 2009:CD002759.

91. Peterson MD, Rhea MR, Sen A, et al. Resistance exercise for muscular strength in older adults: a meta-analysis. Ageing Res Rev 2010;9:226-237.

92. Borde R, Hortobagyi T, Granacher U. Dose-Response Relationships of Resistance Training in Healthy Old Adults: A Systematic Review and Meta-Analysis. Sports Med 2015;45:1693-1720.

93. Van Abbema R, De Greef M, Craje C, et al. What type, or combination of exercise can improve preferred gait speed in older adults? A meta-analysis. BMC Geriatr 2015;15:72.

94. Hortobagyi T, Lesinski M, Gabler M, et al. Effects of Three Types of Exercise Interventions on Healthy Old Adults' Gait Speed: A Systematic Review and Meta-Analysis (vol 45, pg 1627, 2015). Sports Medicine 2016;46:453-453.

95. Giuliano C, Karahalios A, Neil C, et al. The effects of resistance training on muscle strength, quality of life and aerobic capacity in patients with chronic heart failure - A meta-analysis. Int J Cardiol 2016.

96. Delorme TL. Restoration of Muscle Power by Heavy-Resistance Exercises. Journal of Bone and Joint Surgery 1945;27:645-667.

97. Raymond MJ, Bramley-Tzerefos RE, Jeffs KJ, et al. Systematic review of high-intensity progressive resistance strength training of the lower limb compared with other intensities of strength training in older adults. Arch Phys Med Rehabil 2013;94:1458-1472.

98. de Vreede PL, Samson MM, van Meeteren NL, et al. Functional-task exercise versus resistance strength exercise to improve daily function in older women: a randomized, controlled trial. J Am Geriatr Soc 2005;53:2-10.

99. Krebs DE, Scarborough DM, McGibbon CA. Functional vs. strength training in disabled elderly outpatients. Am J Phys Med Rehabil 2007;86:93-103.

100. Sherrington C, Lord SR, Herbert RD. A randomized controlled trial of weight-bearing versus non-weight-bearing exercise for improving physical ability after usual care for hip fracture. Arch Phys Med Rehabil 2004;85:710-716.

101. Olivetti L, Schurr K, Sherrington C, et al. A novel weight-bearing strengthening program during rehabilitation of older people is feasible and improves standing up more than a non-weight-bearing strengthening program: a randomised trial. Aust J Physiother 2007;53:147-153.

64

102. Augustsson J, Esko A, Thomee R, et al. Weight training of the thigh muscles using closed vs. open kinetic chain exercises: a comparison of performance enhancement. J Orthop Sports Phys Ther 1998;27:3-8.

103. Howe TE, Rochester L, Neil F, et al. Exercise for improving balance in older people. Cochrane Database Syst Rev 2011:CD004963.

104. Lesinski M, Hortobagyi T, Muehlbauer T, et al. Effects of Balance Training on Balance Performance in Healthy Older Adults: A Systematic Review and Meta-analysis. Sports Med 2015;45:1721-1738.

105. Gillespie LD, Robertson MC, Gillespie WJ, et al. Interventions for preventing falls in older people living in the community. Cochrane Database Syst Rev 2012:CD007146.

106. Sherrington C, Tiedemann A, Fairhall N, et al. Exercise to prevent falls in older adults: an updated meta-analysis and best practice recommendations. N S W Public Health Bull 2011;22:78-83.

107. Farlie MK, Robins L, Keating JL, et al. Intensity of challenge to the balance system is not reported in the prescription of balance exercises in randomised trials: a systematic review. J Physiother 2013;59:227-235.

108. Colcombe S, Kramer AF. Fitness effects on the cognitive function of older adults: a meta-analytic study. Psychol Sci 2003;14:125-130.

109. Erickson KI, Weinstein AM, Lopez OL. Physical activity, brain plasticity, and Alzheimer's disease. Arch Med Res 2012;43:615-621.

110. Davenport MH, Hogan DB, Eskes GA, et al. Cerebrovascular reserve: the link between fitness and cognitive function? Exerc Sport Sci Rev 2012;40:153-158.

111. Boraxbekk CJ, Salami A, Wahlin A, et al. Physical activity over a decade modifies age-related decline in perfusion, gray matter volume, and functional connectivity of the posterior default-mode network-A multimodal approach. Neuroimage 2016;131:133-141.

112. Young J, Angevaren M, Rusted J, et al. Aerobic exercise to improve cognitive function in older people without known cognitive impairment. Cochrane Database of Systematic Reviews 2015.

113. Ngandu T, Lehtisalo J, Solomon A, et al. A 2 year multidomain intervention of diet, exercise, cognitive training, and vascular risk monitoring versus control to prevent cognitive decline in at-risk elderly people (FINGER): a randomised controlled trial. Lancet 2015;385:2255-2263.

114. Sink KM, Espeland MA, Castro CM, et al. Effect of a 24-Month Physical Activity Intervention vs Health Education on Cognitive Outcomes in Sedentary Older Adults: The LIFE Randomized Trial. JAMA 2015;314:781-790.

115. Crocker T, Forster A, Young J, et al. Physical rehabilitation for older people in long-term care. Cochrane Database Syst Rev 2013;2:CD004294.

65

116. Rosendahl E, Lindelöf N, Littbrand H, et al. High-intensity functional exercise program and protein-enriched energy supplement for older persons dependent in activities of daily living: a randomised controlled trial. Aust J Physiother 2006;52:105-113.

117. Littbrand H, Carlsson M, Lundin-Olsson L, et al. Effect of a high-intensity functional exercise program on functional balance: preplanned subgroup analyses of a randomized controlled trial in residential care facilities. J Am Geriatr Soc 2011;59:1274-1282.

118. Littbrand H, Lundin-Olsson L, Gustafson Y, et al. The effect of a high-intensity functional exercise program on activities of daily living: a randomized controlled trial in residential care facilities. J Am Geriatr Soc 2009;57:1741-1749.

119. Bean JF, Vora A, Frontera WR. Benefits of exercise for community-dwelling older adults. Arch Phys Med Rehabil 2004;85:S31-42; quiz S43-34.

120. Dick MB, Hsieh S, Bricker J, et al. Facilitating acquisition and transfer of a continuous motor task in healthy older adults and patients with Alzheimer's disease. Neuropsychology 2003;17:202-212.

121. Dick MB, Hsieh S, Dick-Muehlke C, et al. The variability of practice hypothesis in motor learning: does it apply to Alzheimer's disease? Brain Cogn 2000;44:470-489.

122. van Halteren-van Tilborg IA, Scherder EJ, Hulstijn W. Motor-skill learning in Alzheimer's disease: a review with an eye to the clinical practice. Neuropsychol Rev 2007;17:203-212.

123. Hauer K, Schwenk M, Zieschang T, et al. Physical training improves motor performance in people with dementia: a randomized controlled trial. J Am Geriatr Soc 2012;60:8-15.

124. Zieschang T, Schwenk M, Oster P, et al. Sustainability of motor training effects in older people with dementia. J Alzheimers Dis 2013;34:191-202.

125. Rolland Y, Pillard F, Klapouszczak A, et al. Exercise program for nursing home residents with Alzheimer's disease: a 1-year randomized, controlled trial. J Am Geriatr Soc 2007;55:158-165.

126. Telenius EW, Engedal K, Bergland A. Effect of a high-intensity exercise program on physical function and mental health in nursing home residents with dementia: an assessor blinded randomized controlled trial. PLoS One 2015;10:e0126102.

127. Telenius EW, Engedal K, Bergland A. Long-term effects of a 12 weeks high-intensity functional exercise program on physical function and mental health in nursing home residents with dementia: a single blinded randomized controlled trial. Bmc Geriatrics 2015;15.

128. Forbes D, Forbes SC, Blake CM, et al. Exercise programs for people with dementia. Cochrane Database Syst Rev 2015;4:CD006489.

66

129. Littbrand H, Stenvall M, Rosendahl E. Applicability and effects of physical exercise on physical and cognitive functions and activities of daily living among people with dementia: a systematic review. Am J Phys Med Rehabil 2011;90:495-518.

130. Pitkälä KH, Pöysti MM, Laakkonen ML, et al. Effects of the Finnish Alzheimer disease exercise trial (FINALEX): a randomized controlled trial. JAMA Intern Med 2013;173:894-901.

131. Birks JS, Chong LY, Grimley Evans J. Rivastigmine for Alzheimer's disease. Cochrane Database Syst Rev 2015;9:CD001191.

132. Tan CC, Yu JT, Wang HF, et al. Efficacy and safety of donepezil, galantamine, rivastigmine, and memantine for the treatment of Alzheimer's disease: a systematic review and meta-analysis. J Alzheimers Dis 2014;41:615-631.

133. Birks J, McGuinness B, Craig D. Rivastigmine for vascular cognitive impairment. Cochrane Database Syst Rev 2013:CD004744.

134. Christofoletti G, Oliani MM, Gobbi S, et al. A controlled clinical trial on the effects of motor intervention on balance and cognition in institutionalized elderly patients with dementia. Clin Rehabil 2008;22:618-626.

135. Steinberg M, Leoutsakos JM, Podewils LJ, et al. Evaluation of a home-based exercise program in the treatment of Alzheimer's disease: the Maximizing Independence in Dementia (MIND) study. Int J Geriatr Psychiatry 2009;24:680-685.

136. Eggermont LH, Swaab DF, Hol EM, et al. Walking the line: a randomised trial on the effects of a short term walking programme on cognition in dementia. J Neurol Neurosurg Psychiatry 2009;80:802-804.

137. Vreugdenhil A, Cannell J, Davies A, et al. A community-based exercise programme to improve functional ability in people with Alzheimer's disease: a randomized controlled trial. Scand J Caring Sci 2012;26:12-19.

138. Kemoun G, Thibaud M, Roumagne N, et al. Effects of a physical training programme on cognitive function and walking efficiency in elderly persons with dementia. Dement Geriatr Cogn Disord 2010;29:109-114.

139. Venturelli M, Scarsini R, Schena F. Six-month walking program changes cognitive and ADL performance in patients with Alzheimer. Am J Alzheimers Dis Other Demen 2011;26:381-388.

140. Arcoverde C, Deslandes A, Moraes H, et al. Treadmill training as an augmentation treatment for Alzheimer's disease: a pilot randomized controlled study. Arquivos De Neuro-Psiquiatria 2014;72:190-196.

141. Öhman H, Savikko N, Strandberg TE, et al. Effects of Exercise on Cognition: The Finnish Alzheimer Disease Exercise Trial: A Randomized, Controlled Trial. J Am Geriatr Soc 2016;64:731-738.

67

142. Cancela JM, Ayan C, Varela S, et al. Effects of a long-term aerobic exercise intervention on institutionalized patients with dementia. J Sci Med Sport 2016;19:293-298.

143. Hoffmann K, Sobol NA, Frederiksen KS, et al. Moderate-to-High Intensity Physical Exercise in Patients with Alzheimer's Disease: A Randomized Controlled Trial. J Alzheimers Dis 2015;50:443-453.

144. Bossers WJ, van der Woude LH, Boersma F, et al. A 9-Week Aerobic and Strength Training Program Improves Cognitive and Motor Function in Patients with Dementia: A Randomized, Controlled Trial. Am J Geriatr Psychiatry 2015;23:1106-1116.

145. McCarney R, Warner J, Iliffe S, et al. The Hawthorne Effect: a randomised, controlled trial. BMC Med Res Methodol 2007;7:30.

146. Kolanowski A, Litaker M. Social interaction, premorbid personality, and agitation in nursing home residents with dementia. Arch Psychiatr Nurs 2006;20:12-20.

147. Katz S, Ford AB, Moskowitz RW, et al. Studies of Illness in the Aged. The Index of Adl: A Standardized Measure of Biological and Psychosocial Function. JAMA 1963;185:914-919.

148. Folstein MF, Folstein SE, McHugh PR. "Mini-mental state". A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189-198.

149. Rydwik E, Bergland A, Forsen L, et al. Investigation into the reliability and validity of the measurement of elderly people's clinical walking speed: a systematic review. Physiother Theory Pract 2012;28:238-256.

150. Collin C, Wade DT, Davies S, et al. The Barthel ADL Index: a reliability study. Int Disabil Stud 1988;10:61-63.

151. Ranhoff AH. Reliability of nursing assistants' observations of functioning and clinical symptoms and signs. Aging (Milano) 1997;9:378-380.

152. McDowell I. Measuring Health 3rd Ed. New York, United States of America: Oxford University Press, 2006.

153. Cotter EM, Burgio LD, Stevens AB, et al. Correspondence of the functional independence measure (FIM) self-care subscale with real-time observations of dementia patients' ADL performance in the home. Clin Rehabil 2002;16:36-45.

154. Ottenbacher KJ, Hsu Y, Granger CV, et al. The reliability of the functional independence measure: a quantitative review. Arch Phys Med Rehabil 1996;77:1226-1232.

155. Berg KO, Wood-Dauphinee SL, Williams JI, et al. Measuring balance in the elderly: validation of an instrument. Can J Public Health 1992;83 Suppl 2:S7-11.

156. Conradsson M, Lundin-Olsson L, Lindelöf N, et al. Berg balance scale: intrarater test-retest reliability among older people dependent in activities of

68

daily living and living in residential care facilities. Phys Ther 2007;87:1155-1163.

157. Sibley KM, Howe T, Lamb SE, et al. Recommendations for a core outcome set for measuring standing balance in adult populations: a consensus-based approach. PLoS One 2015;10:e0120568.

158. Rosen WG, Mohs RC, Davis KL. A new rating scale for Alzheimer's disease. Am J Psychiatry 1984;141:1356-1364.

159. Cano SJ, Posner HB, Moline ML, et al. The ADAS-cog in Alzheimer's disease clinical trials: psychometric evaluation of the sum and its parts. J Neurol Neurosurg Psychiatry 2010;81:1363-1368.

160. Morris JC, Heyman A, Mohs RC, et al. The Consortium to Establish a Registry for Alzheimer's Disease (CERAD). Part I. Clinical and neuropsychological assessment of Alzheimer's disease. Neurology 1989;39:1159-1165.

161. Sheikh JI, Yesavage JA. Geriatric Assessment Scale (GDS): Recent evidence and development of a shorter version. Clinical Gerontologist 1986;5:165-172.

162. Shrive FM, Stuart H, Quan H, et al. Dealing with missing data in a multi-question depression scale: a comparison of imputation methods. BMC Med Res Methodol 2006;6:57.

163. Cummings JL, Mega M, Gray K, et al. The Neuropsychiatric Inventory: comprehensive assessment of psychopathology in dementia. Neurology 1994;44:2308-2314.

164. Littbrand H, Lindelöf N, Rosendahl E. The HIFE Program: The High-Intensity Functional Exercise Program, 2nd ed. Umeå. ISBN 978-91-7601-166-9.: Umeå University, Department of Community Medicine and Rehabilitation, Geriatric Medicine, 2014.

165. Littbrand H, Rosendahl E, Lindelöf N, et al. A high-intensity functional weight-bearing exercise program for older people dependent in activities of daily living and living in residential care facilities: evaluation of the applicability with focus on cognitive function. Phys Ther 2006;86:489-498.

166. Kleinbaum DG, Klein M. Survival Analysis: A Self-Learning Text, 3rd Ed. New York, NY: Springer New York, 2012.

167. Pocock SJ, Assmann SE, Enos LE, et al. Subgroup analysis, covariate adjustment and baseline comparisons in clinical trial reporting: current practice and problems. Stat Med 2002;21:2917-2930.

168. Assmann SF, Pocock SJ, Enos LE, et al. Subgroup analysis and other (mis)uses of baseline data in clinical trials. Lancet 2000;355:1064-1069.

169. Rosendahl E. Effect size underestimates the effects of interventions among older people with severe physical or cognitive impairments? Journal of the American Geriatrics Society 2007;55:1315-1316.

170. Buuren Sv. Flexible imputation of missing data. Boca Raton, FL: CRC Press, 2012.

69

171. Ostir GV, Kuo YF, Berges IM, et al. Measures of lower body function and risk of mortality over 7 years of follow-up. Am J Epidemiol 2007;166:599-605.

172. Rolland Y, Lauwers-Cances V, Cesari M, et al. Physical performance measures as predictors of mortality in a cohort of community-dwelling older French women. Eur J Epidemiol 2006;21:113-122.

173. Cesari M, Onder G, Zamboni V, et al. Physical function and self-rated health status as predictors of mortality: results from longitudinal analysis in the ilSIRENTE study. BMC Geriatr 2008;8:34.

174. Taekema DG, Gussekloo J, Westendorp RG, et al. Predicting survival in oldest old people. Am J Med 2012;125:1188-1194 e1181.

175. Schnelle JF, Alessi CA, Simmons SF, et al. Translating clinical research into practice: a randomized controlled trial of exercise and incontinence care with nursing home residents. J Am Geriatr Soc 2002;50:1476-1483.

176. Schnelle JF, Kapur K, Alessi C, et al. Does an exercise and incontinence intervention save healthcare costs in a nursing home population? J Am Geriatr Soc 2003;51:161-168.

177. Ouslander JG, Griffiths PC, McConnell E, et al. Functional incidental training: a randomized, controlled, crossover trial in Veterans Affairs nursing homes. J Am Geriatr Soc 2005;53:1091-1100.

178. Ames D, Burns A, O'Brien J, et al. Dementia. 4th ed. London, United Kingdom: Hodder Arnold, 2010.

179. Nudo RJ. Recovery after brain injury: mechanisms and principles. Frontiers in Human Neuroscience 2013;7.

180. Veerbeek JM, van Wegen E, van Peppen R, et al. What Is the Evidence for Physical Therapy Poststroke? A Systematic Review and Meta-Analysis. Plos One 2014;9.

181. Sobol NA, Hoffmann K, Frederiksen KS, et al. Effect of aerobic exercise on physical performance in patients with Alzheimer's disease. Alzheimers Dement 2016.

182. Erickson KI, Voss MW, Prakash RS, et al. Exercise training increases size of hippocampus and improves memory. Proc Natl Acad Sci U S A 2011;108:3017-3022.

183. Kudielka BM, Hellhammer DH, Wust S. Why do we respond so differently? Reviewing determinants of human salivary cortisol responses to challenge. Psychoneuroendocrinology 2009;34:2-18.

184. Donders ART, van der Heijden GJMG, Stijnen T, et al. Review: A gentle introduction to imputation of missing values. Journal of Clinical Epidemiology 2006;59:1087-1091.

185. Atkins D, Best D, Briss PA, et al. Grading quality of evidence and strength of recommendations. British Medical Journal 2004;328:1490-1494.

70

186. von Elm. Strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies (vol 335, pg 806, 2007). British Medical Journal 2008;336:35-35.

187. Campbell MK, Piaggio G, Elbourne DR, et al. Consort 2010 statement: extension to cluster randomised trials. British Medical Journal 2012;345.

188. Moher D, Hopewell S, Schulz KF, et al. CONSORT 2010 Explanation and Elaboration: updated guidelines for reporting parallel group randomised trials. Bmj-British Medical Journal 2010;340.

189. Pocock SJ, Assmann SE, Enos LE, et al. Subgroup analysis, covariate adjustment and baseline comparisons in clinical trial reporting: current practice and problems. Statistics in Medicine 2002;21:2917-2930.

190. Pasma JH, Stijntjes M, Ou SS, et al. Walking speed in elderly outpatients depends on the assessment method. Age (Dordr) 2014;36:9736.

191. Woods B, Aguirre E, Spector AE, et al. Cognitive stimulation to improve cognitive functioning in people with dementia. Cochrane Database of Systematic Reviews 2012.

192. Hamilton JM, Salmon DP, Raman R, et al. Accounting for functional loss in Alzheimer's disease and dementia with Lewy bodies: beyond cognition. Alzheimers Dement 2014;10:171-178.

193. Granger CV, Cotter AC, Hamilton BB, et al. Functional assessment scales: a study of persons after stroke. Arch Phys Med Rehabil 1993;74:133-138.

194. McDowell I. Measuring Health 3rd Ed. New York, United States of America: Oxford University Press 2006.

195. Perera S, Mody SH, Woodman RC, et al. Meaningful change and responsiveness in common physical performance measures in older adults. J Am Geriatr Soc 2006;54:743-749.