R

Pc

Ea

b

c

T

a

ARR2AA

KPAASDL

1

1

m(pilmidlt

RT

a

1h

Ageing Research Reviews 12 (2013) 329– 338

Contents lists available at SciVerse ScienceDirect

Ageing Research Reviews

j ourna l ho mepage: www.elsev ier .com/ locate /ar r

eview

revention of onset and progression of basic ADL disability by physical activity inommunity dwelling older adults: A meta-analysis

rwin Taka,b,∗, Rebecca Kuipera, Astrid Chorusa,b, Marijke Hopman-Rocka,b,c

Netherlands Organization for Applied Scientific Research TNO, Department Lifestyle, PO Box 2215, 2301 CE Leiden, The NetherlandsBody@Work, Research Center Physical Activity, Work and Health, TNO-VU University Medical Center, Van der Boechorststraat 7, 1081 BT Amsterdam, The NetherlandsDepartment of Public and Occupational Health/EMGO Institute for Health and Care Research (EMGO+), VU University Medical Center, Van der Boechorststraat 7, 1081 BT Amsterdam,he Netherlands

r t i c l e i n f o

rticle history:eceived 1 May 2012eceived in revised form2 September 2012ccepted 1 October 2012vailable online 10 October 2012

eywords:hysical activityctivities of daily livingging

a b s t r a c t

Purpose: Physical activity (PA) is an important behavior when it comes to preventing or slowing downdisablement caused by aging and chronic diseases. It remains unclear whether PA can directly prevent orreduce disability in activities of daily living (ADL). This article presents a meta-analysis of the associationbetween PA and the incidence and progression of basic ADL disability (BADL).Methods: Electronic literature search and cross-referencing of prospective longitudinal studies of PA andBADL in community dwelling older adults (50+) with baseline and follow-up measurements, multivariateanalysis and reporting a point estimate for the association.Results: Compared with a low PA, a medium/high PA level reduced the risk of incident BADL disabilityby 0.51 (95% CI: 0.38, 0.68; p < 001), based on nine longitudinal studies involving 17,000 participantsfollowed up for 3–10 years. This result was independent of age, length of follow-up, study quality, and

ystematic reviewisablement processongitudinal studies

differences in demographics, health status, functional limitations, and lifestyle. The risk of progressionof BADL disability in older adults with a medium/high PA level compared with those with a low PA levelwas 0.55 (95% CI: 0.42, 0.71; p < 001), based on four studies involving 8500 participants.Discussion: This is the first meta-analysis to show that being physically active prevents and slows downthe disablement process in aging or diseased populations, positioning PA as the most effective preventive

d red

strategy in preventing an

. Introduction

.1. Background

Aging is often accompanied by a decline in functional perfor-ance and disability as described in the disablement process model

Nagi, 1976; Verbrugge and Jette, 1994). According to this model,athological processes specific to a disease or trauma result in

mpairments (anatomic and structural abnormalities) which in turnead to functional limitations (restrictions in basic physical and

ental actions), which ultimately result in disability (difficultyn doing activities in daily life). Disability threatens the indepen-

ence of older persons and, through demands of (long-term) care,

eads to increases in health care costs. Although there are indica-ions that the prevalence of some types of disability is declining

∗ Corresponding author<> at: Netherlands Organization for Applied Scientificesearch TNO, Department Lifestyle, PO Box 2215, 2301 CE Leiden, The Netherlands.el.: +31 88666193.

E-mail addresses: erwin.tak@tno.nl (E. Tak), rebecca.kuiper@tno.nl (R. Kuiper),strid.chorus@tno.nl (A. Chorus), marijke.hopman@tno.nl (M. Hopman-Rock).

568-1637/$ – see front matter © 2012 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.arr.2012.10.001

ucing disability, independence and health care cost in aging societies.© 2012 Elsevier B.V. All rights reserved.

(Freedman et al., 2002; Freedman, 2009; Puts et al., 2008; Schoeniet al., 2008), especially in older or frail populations prevalence is stillhigh (Fuller-Thomson et al., 2009; Picavet and Hoeymans, 2002). Ashypothesized by the disablement process model and supported byrecent studies, disability is a not a static but a dynamic process(Hardy et al., 2005) in which several factors play a role in the onset,recovery or worsening of older persons’ disability status.

Different types of factors are conceptualized to speed upand slow down the pathway from aging and disease to disabil-ity in the disablement process. These include risk factors (e.g.demographic factors), extra-individual factors (e.g. medical care)and intra-individual factors including behavioral factors such asphysical activity which are thought to be primarily interactingwith functional performance which subsequently effects disability(Lawrence and Jette, 1996; Verbrugge and Jette, 1994). Althoughsome studies confirm that part of the effect of physical activityon disability goes via improved functional performance (Lawrenceand Jette, 1996; Miller et al., 2000; Wolinsky et al., 1995) there

is evidence indicating a direct effect of physical activity on dis-ability (Keysor, 2003). There have been longitudinal studies thatunderscore the protective effect of physical activity on disability,although there are some methodological challenges in reviewing

3 rch Re

ttiia(tsatdwbwofiw

2

emagiase

3

3

vi(uild

3

ddponpe(iWp

rhraws

30 E. Tak et al. / Ageing Resea

his evidence (Keysor, 2003; Manini and Pahor, 2009). Especiallyhe diversity in used definitions and methods to measure phys-cal activity as well as disability make the synthesis of evidencento a challenging enterprise. So far in systematic reviews, physicalctivity has been identified as a risk factor for functional declineStuck et al., 1999) and as a prognostic factor of disability, althoughhe evidence was qualified as limited (Tas et al., 2007). A recentystematic review (Paterson and Warburton, 2010) reported anssociation between physical activity and a reduced risk of func-ional limitations, disability, and loss of independence, although theistinction between the entities of the disablement process modelas not always clearly followed. Also, older adults with and without

aseline disability were included and data from different studiesere not pooled because of heterogeneity in physical activity and

utcome measures. A systematic review (Vermeulen et al., 2011)ocused on physical frailty indicators including low physical activ-ty which predicted future ADL disability but also no meta-analysis

as performed on these data.

. Objective

In an effort to overcome these limitations and add to thisvidence, our objective is to report a systematic review andeta-analysis of (prospective) longitudinal studies analyzing the

ssociation between physical activity and both incidence and pro-ression of disability in (basic) activities of daily living (BADL)n older community dwelling adults. This report was written inccordance with the PRISMA and MOOSE statement for reportingystematic reviews and meta-analyses (Liberati et al., 2009; Stroupt al., 2000).

. Methods

.1. Literature search

Publications before January 2012 were electronically searchedia PubMed using the following terms: (a) older adults; (b) phys-cal activity/exercise; (c) disability, activities of daily living (ADL);d) longitudinal, prospective. A full description of the search termssed for these criteria and the results of the search can be found

n Appendix A. The search was conducted by an experiencedibrarian with broad experience in searching electronic healthcareatabases. Publications were screened for additional references.

.2. Study selection

We included studies that (i) reported prospective, longitudinalata on the association between physical activity and subsequentisability, (ii) identified and measured physical activity as a possibleredictor/risk factor at baseline, (iii) included onset or progressionf disability of basic ADL as an outcome (iv) included older commu-ity dwelling adults (50+ at baseline; studies focusing on specificopulations (i.e. specific diseases, athletes) were excluded with thexception of studies with a focus on gender (i.e. only women)), andv) used multivariate analysis and reported the association explic-tly in a point estimate preferably as a risk, hazard or odds ratio.

e included reports that were published as full-text articles ineer-reviewed journals.

To distinguish disability from functional limitations, defined asestrictions in basic and mental actions, we defined disability asaving any difficulty in performing basic ADL, i.e. those activities

elated to personal care and hygiene, such as dressing/grooming,rising, eating, bathing, and using the toilet. Physical activityas defined as any bodily movement produced by contraction of

keletal muscle that substantially increases energy expenditure

views 12 (2013) 329– 338

above the basal level (Caspersen et al., 1991). No restrictionswere imposed on the literature search, with the exception thatstudies referring to specific exercise programs (i.e. planned or struc-tured types of physical activity) not including habitual activities(i.e. walking, cycling, sports, leisure-time activities, and house-hold) were excluded. Titles and abstracts of identified reportswere reviewed to identify relevant articles. Two authors (ET, MH)scanned all the titles and abstracts independently. If necessary, thefull-text article was scanned by one author (ET) to extract furtherinformation on eligibility criteria. Disagreement between review-ers was resolved by consensus.

3.3. Data extraction

Next, two authors (ET, MH) independently reviewed all selectedfull-text articles using a structured data extraction form, thatincluded, the origin of the study, sample size, follow-up period,population description, used definition and measurement of phys-ical activity and disability, included predictors, analytic method,parameter estimates, and factors for which the reported associa-tion was adjusted. Selected cohort studies were searched to extractrelevant information referred to, but not reported in the selectedpublications. Articles reporting on the same original cohort wereincluded at first, but were excluded if there was data overlap thatresulted in duplication of results. All studies were assessed for riskof bias using a 14-item checklist for assessing the methodologicalquality of prospective studies (based on Singh et al. (2008)). Thechecklist covered four dimensions: (i) study population and partic-ipation rate, (ii) study attrition, (iii) data collection, and (iv) dataanalysis. Half of the items dealt with information and half withvalidity. The two non-blinded reviewers (ET, MH) rated each cri-terion as positive, negative, or insufficient. Disagreement betweenthe reviewers was discussed until a consensus was reached. A sumscore was assigned to each study based on the number of itemsthat were scored positively (range for total 0–14, information andvalidity scores ranged from 0 to 7).

Reported levels of physical activity were reduced to a maximumof three levels: none/low, medium, high/vigorous as presented inTable 1 Table 1. The outcome was defined as new onset of basic ADLdisability (incidence) or increase in disability (progression, eitherdefined as a change score between measurements or increase onthe respective ADL scale score). Summary measures recorded werepoint estimates of the (adjusted) association between the specifiedphysical activity levels and the incidence or progression of basicADL disability.

3.4. Analysis

Studies were included in the meta-analysis if they reportedodds ratios (OR) or relative risk (RR), comparing medium,medium/high+, or high versus low physical activity levels (seeTable 1 for details). For those studies that reported ratios formedium versus low and high versus low comparisons, the ratioswere averaged to compare medium/high versus low.

In each analysis, we incorporated the ratio and its 95% confi-dence interval for each study. To allow generalization we used arandom-effects model and present results in a forest plot, whichgives an overview of the studies, their ratios (treatments effects),and the relative strengths of the treatment effects (weights). Sincewe used the random-effects model, the weights incorporate thewithin-study variance (including sample size) and the between-study variance. Additionally, we calculated three measures of

heterogeneity, namely Q, T2, and I2. To obtain insight into bias,such as publication bias and selective reporting within studies, weinspected the funnel plot visually and used Egger’s test, Duval andTweedie’s trim and fill, and the Rosenthals’ fail-safe N. In addition,

E. Tak et al. / Ageing Research Reviews 12 (2013) 329– 338 331

Table 1Overview of reported PA levels and reduction to (a maximum of) three levels.

Categoriesa Low Moderate High

Odds Ratiosb

Avlund Sedentary Active (at least a weekly activity)Haveman-Nies Inactive (1st tertile) Active (2nd and 3rd tertile)Ishizaki No habit walk Habit to walkJenkins <3 times/week ≥3 times/week vigorousLeveille Rarely, never,1 month 1–2 on PA scale 3–6 on PA scaleMiller Not Walking at least 1 mile/weekOstbye Never

Other1–2 p/w heavy or 3 p/w light ≥3 p/w heavy

Puts Low (<76 min/day) Higher (>76 min/day)Stessman (2009) Sedentary Vigorous PA or walking 1 h/dayStessman (2002) Less Exercise/PA ≥ 4 day/weekVan den Brink 1st tertile EEc 2nd tertile EE 3rd tertile EEYoung 1st tertiel EEd 2nd tertile EE 3rd tertile EE

Relative RisksWu No routine Routine exerciser, at least twice a week

a The categories mentioned after the separate authors refer to the names reported in their respective publication.

–6360

wgmetaoft

4

4

sepIdotcam

sW(oo(aib2atsqBoa

association between physical activity and new onset of disability,and 4 publications reported on progression of disability as the out-come measure. All studies adjusted for either demographic factors,

Literature databas e search:Databases : pubmedLimits: Engl ish, Dutch , German l ang uag e articles only

Search resu lts combin ed (n=37 3)

Articl es scree ned on basis of title and abstract

Excluded 304

104 (69 + 35) full -te xt articl es assessed fo r eligibility

Excluded 80 (59 + 2 1):- PA not includ ed as p red ictor (n =21+3)- No AD L di sab ility as outcome (n=1 7+10)- No associa tion repor ted fo r PA d isab ility

(n=9+2)- No longitu dinal d esign (n=11+4)- Other: (su b)popu latio n (n=1+2)

24 studies included

35 ad ditio nal records identified through other sources

Excluded 11- no odds ratio reported (n=10 )- no compariso n hig h/med ium vs.

low (n= 1)

b Reports are further categorized based on the point-estimate used.c EE energy expenditure; values for tertiles not reportedd Low = 914.6–2110.2 kcal/day, Moderate = 2111.1–2533.9 kcal/day, High = 2534.1

e performed subgroup analysis and meta-regression. In the sub-roup analysis, we compared two or more subgroups and, in theeta regression, we examined the association between the mean

ffect size with a continuous variable. This was done for age olderhan 75 years (yes/no), follow-up longer than 72 months (yes/no),nalyses adjusted for demographics, health, functional limitationsr lifestyle (yes/no), and quality of the study (continuous). We per-ormed a fixed-effects analysis to examine the differences betweenhe subgroups.

. Results

.1. Description of studies

One hundred and four full-text articles were retrieved andcreened for eligibility (Fig. 1 Fig. 1). Of these articles, 80 werexcluded mostly because physical activity was not studied as aredictor or basic ADL disability was not studied as an outcome.

n most of these cases, a specific type of disability was studied thatid not involve personal care (such as instrumental ADL disabilityr mobility disability), or measures were used that resembled func-ional limitations (i.e., walking, climbing stairs, crouching, kneeling,arrying ¼lb, etc.), although these were often referred to as dis-bility. Of the remaining 24 publications, 13 were eligible for theeta-analysis.Table 2 Table 2 reviews the characteristics of the 13 included

tudies. Most studies were conducted in the USA or Europe.ith the exception of the Jerusalem Longitudinal Cohort Study

Stessman et al., 2002, 2009), no study was reported in more thanne publication. Since one publication from this group reportedn incidence (Stessman et al., 2009) and the other on progressionStessman et al., 2002), there was no overlap of participants in thenalyses. With respect to the other characteristics of included stud-es, most involved participants aged between 70 and 80 years ataseline, with only 1 study reporting on 50 year (Ostbye et al.,002). The follow-up ranged between 36 and 42 months in 5 studiesnd 60–72 months in 5 studies; 3 studies had a follow-up longerhan 100 months. Physical activity was usually measured with aingle question, but in 3 studies it was measured with a validated

uestionnaire (Haveman-Nies et al., 2003; Puts et al., 2005; Van Denrink et al., 2005). Most of these measures registered frequencyr intensity of physical activity, but 2 studies converted physicalctivity data into a measure of energy expenditure. Disability in

.6 kcal/day

basic ADL was usually measured with questions about differentactivities or with a validated scale, mainly the Katz scale (Katz et al.,1963). Several levels of disability were used, ranging from difficultyin performing basic ADL, needing help to perform, the inability toperform one or more activities. Nine publications reported on the

13 studies incl uded in meta-an alysis

Fig. 1. Flowchart: results of the literature search and selection.

332E.

Tak et

al. /

Ageing

Research

Review

s 12 (2013) 329– 338

Table 2Characteristics of included longitudinal studies.

First author (year) Study (country) Mean age (SD)at baseline

Numberincluded inanalysis

FU-period(months)

PA measurement(mode)

PA types PA predictorstatus

ADL-disabilitymeasurement

Disability outcomemeasure (criterion)

Quality score,range 0–14(validity items,0–7)

Avlund (2002) NORAa (DK, FI) NR (≥74) 429 60 Single question(frequency)

Sports,walking,gardening

Covariate PADL-H scale Incidence (help in oneof more activities)

10 (4)

Haveman-Nies(2003)

SENECAb (EU) m 72.6 (1.6) w72.7 (1.7)

381 120 Voorripsquestionnaire(Voorrips et al.,1991) (frequency)

Sports,household,leisure-time

One of 3 6 self-reportquestions

Incidence (no difficultyor difficulty in only 1activity)

9 (3)

Ishizaki (2000) LISAc (JP) 70.9 (4.9) 583 36 Single question(frequency)

Walking One of 28 5 self-reportquestions

Incidence (loss ofindependence in eachactivity)

11 (5)

Jenkins (2004) AHEADd (US) 78.0 (4.9) 3373 36 Single question(intensity)

Vigorousexercise

Covariate 4 self-reportquestions

Incidence (difficulty inat least 1 activity)

7 (3)

Leveille (1999) EPESEe (US) NR (≥65) 1097 73 (range24–96)

Several questions(frequency)

Vigorousexercise,walking,gardening,

One of 7 6 self-reportquestions

Dying withoutdisability (need helpwith activities)

9 (3)

Miller (2000) LSOA f(US) 78.2 (6.0) 3589 72 Single question(frequency)

Walking Principal ADL-scale(Johnson andWolinsky,1993;Wolinsky andJohnson, 1991)

Progression (none, 1–2or 3–5 disabilities)

10 (4)

Ostbye (2002) HRSg (US) NR (51–64) 7845 72 Two questions(frequency,intensity)

Sports,walking,dancing,gardening

One of 8 4 self-reportquestions

Incidence (need of helpin at least one of 4)

7 (3)

Puts (2005) LASAh (NL) Between 74.1(6.1) and 78.3(6.1)p

1321 72 LAPAQquestionnaire (Stelet al., 2004)(frequency)

Sports, cycling,walking,gardening,household,

One of 9 frailtymarkers

OECDquestionnaire

Progression (declinebased on changescores)

10 (4)

Stessman (2002) JLCSi (IS) m 77.1 (0.8) w77.1 (0.7)

287 96–108 NR Exercise Principal ADL scale (Katzet al., 1963)

Progression (carry outat least 3 activitieswith ease)

9 (4)

Stessman (2009)m JLCSi (IS) NR (≥70) (1861)n 42 Single question(frequency)

Vigoroussports,physicallyactive

Principal ADL scale (Katzet al., 1963)

Incidence (dependencein at least one activity)

10 (3)

Van den Brink(2005)

FINEj (FI, IT, NL) FI 74.7 (4.0) IT76.1 (3.4) NL74.2 (4.1)

286 120 Zutphenquestionnaire(Caspersen et al.,1991) (frequency,intensity)

Sports, cycling,walking,gardening,hobbies, oddjobs

Principal Questionnaire(Hoeymanset al., 1996)

Incidence (needinghelp on at least oneitem)

11 (4)

E. Tak et al. / Ageing Research Re

Wu

(199

9)

LSk

65+T

apei

(TW

)

NR

(≥65

)13

21

36

NR

(fre

quen

cy)

Hik

ing,

jogg

ing,

wal

kin

g,d

anci

ng

On

e

of

11

AD

L

scal

e

(Kat

zet

al.,

1963

)In

cid

ence

(in

cap

able

ofp

erfo

rmin

g

one

orm

ore

of

acti

viti

es

for

3m

onth

s)

11

(4)

You

ng

(199

5)

HH

Pl(U

S)

77.8

(4.7

)

3428

36

18

acti

viti

es

scor

ed(f

requ

ency

)

and

tran

sfer

red

into

kcal

o

Spor

ts,

wal

kin

g,ga

rden

ing,

leis

ure

-tim

e,w

ork

Prin

cip

al

AD

L

scal

e

(Kat

zet

al.,

1963

)Pr

ogre

ssio

n

(dif

ficu

lty

in

per

form

ing

at

leas

ton

e

acti

vity

)

8

(4)

PA

=

Phys

ical

Act

ivit

y;

AD

L

=

Act

ivit

ies

of

Dai

ly

Livi

ng;

NR

=

not

rep

orte

d;

SD

=

stan

dar

d

dev

iati

on;

FU

=

Foll

ow-u

pa

Nor

dic

Res

earc

h

on

Agi

ng.

bSu

rvey

in

Euro

pe

on

Nu

trit

ion

and

the

Eld

erly

.c

Lon

gitu

din

al

inte

rdis

cip

lin

ary

stu

dy

on

agin

g.d

Ass

et

and

hea

lth

dyn

amic

s

amon

g

the

old

est

old

surv

ey.

eEs

tabl

ish

ed

pop

ula

tion

s

for

epid

emio

logi

c

stu

die

s

of

the

eld

erly

.f

Lon

gitu

din

al

stu

dy

on

agin

g;

incl

ud

es

inst

itu

tion

aliz

ed

par

tici

pan

ts

and

only

men

.g

Hea

lth

reti

rem

ent

stu

dy.

hLo

ngi

tud

inal

agin

g

stu

dy

amst

erd

am;

incl

ud

es

inst

itu

tion

aliz

ed

par

tici

pan

ts.

iJe

rusa

lem

lon

gitu

din

al

coh

ort

stu

dy.

jFi

nla

nd

, Ita

ly

and

the

Net

her

lan

ds

Eld

erly

Stu

dy.

kLo

ngi

tud

inal

Stu

dy.

lH

onol

ulu

Hea

rt

Prog

ram

.m

Two

outc

omes

rep

orte

d:

(1)

mai

nta

inin

g

ease

in

per

form

ing

AD

L

at

78, (

2)

rem

ain

ing

ind

epen

den

t

wh

ile

per

form

ing

AD

L

at

85;

only

the

last

one

was

chos

en

beca

use

of

its

stri

cter

defi

nit

ion

and

clos

er

rese

mbl

ance

to

oth

erou

tcom

e

mea

sure

s.n

Not

spec

ified

for

LR

anal

ysis

.o

Kil

o-ca

lori

es.

pD

isti

ngu

ish

es

fou

r

subg

rou

ps

(no

dec

lin

e

per

form

ance

, dec

lin

e

per

form

ance

, no

self

-rep

orte

d

dec

lin

e

and

self

-rep

orte

d

dec

lin

e)

views 12 (2013) 329– 338 333

life style, or health status; 6 studies for all three; about half of thestudies adjusted for functional limitations (Table 3 Table 3).

The included studies had total quality scores between 7 and 11(out of 14) and a validity score of 3 or 4 (out of 7), with the excep-tion of 1 study which scored 5 (Ishizaki et al., 2000). The averagequality score on quality for all reports was 9.4 out of 14 (67%). Allstudies scored negatively on the measurement of physical activ-ity and disability because self-report measures were used. Lack ofinformation was the main reason for poor scores on the qualityitem ‘selective dropout’, and ‘baseline response levels’. Most stud-ies did not mention reasons for dropout, but when it was reportedby the authors, selective dropout occurred. Response levels werescored negatively in 8 studies on short (80% of participants up to12 months) or a longer follow-up (70% of participants at baseline);response data were missing in 5 studies. For all other quality aspects79% of the publications scored positively.

4.2. Association between physical activity and incident disability

Nine studies reporting incident disability were entered into themeta-analysis (Avlund et al., 2002; Haveman-Nies et al., 2003;Ishizaki et al., 2000; Jenkins, 2004; Leveille et al., 1999; Ostbye et al.,2002; Stessman et al., 2009; Van Den Brink et al., 2005; Wu et al.,1999). Table 4 Table 4 and Fig. 2 Fig. 2 provide information aboutthe type of comparison, outcome and effect estimates, forest plot,and study weights for these studies. The pooled OR estimate for theassociation between medium/high physical activity levels versuslow physical activity levels on the incidence of basic ADL disabilitywas 0.51 (95% CI: 0.38, 0.68; p < 0.001), indicating a lower risk. Withthe exception of 1 study, all studies reported that medium/highphysical activity levels significantly reduced the risk of basic ADLdisability. In this study by Jenkins (2004) the association betweenphysical activity and the onset of disability only lost significancewhen health conditions such as symptoms and functional limita-tions were added to the model.

There was statistical heterogeneity (Q = 35.21, df = 8, p < 0.001;I2 = 77.28%; T2 = 0.14). There was no significant publication biasaccording to either Egger’s Test (p = 0.58), Duval and Tweedie’s trimand fill (0 studies are trimmed both to the right and to the left ofthe mean), and Rosenthals’ fail safe (N = 220). In subgroup analysesand meta-regression, age (older than 75 years), follow-up period,and factors controlled for in the analysis did not significantly alterthe relationship between low levels of physical activity and theincidence of basic ADL disability (for details see Table 5 Table 5).However, as there were only 9 studies, there might not have beenenough power to detect whether these factors had a significanteffect.

The meta-regression on quality yielded the following estimatedmodel:

ln(OR) = −0.83 + 0.02 × QUALITY,where both the interceptand slope cannot be rejected as differing from zero. Because of this,we do not report further meta-regression results on quality. Weconclude that the quality of the study was not linearly related tothe association between physical activity and disability.

4.3. Association between physical activity and progression ofbasic ADL disability

Four studies reported on the progression of disability (Milleret al., 2000; Puts et al., 2005; Stessman et al., 2002; Younget al., 1995). The pooled OR estimate for the associations betweenmedium/high versus low physical activity levels on progression of

basic ADL disability was 0.55 (95% CI: 0.42–0.71; p < 0.001) (Table 6Table 6 and Fig. 3 Fig. 3).

Different studies used different models in their analyses. Milleret al. (2000) differentiated between a model that included or

334 E. Tak et al. / Ageing Research Reviews 12 (2013) 329– 338

Table 3Number of reported adjusted factors per category and per study on incidence and progression.

First author (year) Demographica Lifestyleb Functional limitationsc Health statusd

Studies on incidence of basic ADL disabilityAvlund (2002) 2Haveman-nies (2003) 2 2Ishizaki 2000 2 1 1Jenkins (2004) 6 3 1 1

Leveille (1999) 3 1 3Ostbye (2002) 5 3Stessman (2009) 1 1 1 3Van den Brink (2005) 2 3Wu (1999) 4 3 1 2

Studies on progression of disabilityMiller (2000) 2Puts (2005) 3Stessman (2002) 2 1 1 4Young (1995) 1 1

a Demographic includes: age, sex, race, education, marital status, income, country/study siteb Lifestyle includes: smoking, alcohol, diet, BMI/weight, sleep, social networkc Functional performance includes: cognitive function, functional limitations, gait speed, hand gripd Health status includes: (chronic) diseases (vascular diseases, cancer, stroke, urinary incontinence, hypertension, psychiatric, joint pain, neurological, diabetes mellitus,

back, arthritis, COPD, depressive symptoms, tiredness), self-rated health and health care utilization (hospitalization, service use, preventive tests, home modification)

Table 4Results of the meta-analysis of the level of physical activity on the incidence of basic ADL disability.

First Author (year) Comparisona OR Lower limit Upper limit Relative weight

Avlund (2002) Hm vs Low 0.091 0.091 0.715 21.45Haveman-Nies (2003) Hm vs Low 0.227 0.227 0.887 8.17Ishizaki (2000) Med vs Low 0.230 0.230 0.960 5.49Jenkins (1999) Hm vs Low 0.616 0.616 1.173 27.33Leveille (1999) Hm vs Lowb 0.445 0.445 0.970 10.92Ostbye (2002) High vs Low 0.219 0.219 0.358 16.01Stessman (2009) Hm vs Low 0.300 0.300 0.901 3.25Van den Brink (2005) Hm vs Lowb 0.392 0.392 0.965 3.57Wu (1999) Hm vs Low 0.394 0.394 0.687 3.81

Summary 0.507 0.379 0.679

a Hm = high/medium PA level, High = high PA level, Med = medium PA level, Low = low PA levelb Combined with Med vs. Low (mean of the two comparisons was used)

Table 5Results subgroup analysis and meta-regression.

Factor N no/0 N yes/1 Est. no/0 Est. yes/1 Qbetween df p-value

Age above 75a 2 4 0.701 0.421 2.36 1 .12Follow-up time larger than 72 months 3 6 0.604 0.475 0.95 1 .33Control for demographic factors 1 8 0.370 0.523 0.86 1 .35Control for health factors 3 6 0.413 0.586 1.30 1 .26Control for function limitations 6 3 0.451 0.623 1.55 1 .21Control for factors of lifestyle 1 8 0.370 0.523 0.86 1 .35

N

emPia

TR

= number of studies per group (i.e. for no/0 and for yes/1)a Note that (the range of) age was not available for all studies.

xcluded functional limitations and found the latter to reduce risk

ore (OR = 0.77, 95% CI: 0.60–0.98, vs. OR = 0.58, 95% CI: 0.46–0.73).

uts et al. (2005) differentiated between static and dynamic phys-cal activity (decline in physical activity levels during follow-up)nd reported lower risk reduction in older adults with declined

able 6esults of the meta-analysis of the level of physical activity on the progression of basic A

First Author (year) Comparisona OR

Young (1995) Hm vs. Lowb 0.543

Stessman (2002) Hm vs. Low 0.230

Puts (2005) Hm vs. Low 0.538

Miller (2000) Hm vs. Low 0.668

Summary 0.550

a Hm = high/medium PA level, Low = low PA levelb Combined with Med vs. Low (mean of the two comparisons was used)

physical activity levels (OR = 0.49, 95% CI: 0.36–0.67) vs. 0.59, 95%

CI: 0.41–0.85). One study compared medium as well as high levelsof physical activity with low levels and found similar reductions inrisk (Young et al., 1995). This association was also seen in subsam-ples of healthy or chronically diseased older adults.

DL disability.

Lower limit Upper limit Relative weight

0.386 0.764 48.580.097 0.544 24.040.383 0.754 3.720.527 0.848 23.67

0.420 0.710

E. Tak et al. / Ageing Research Reviews 12 (2013) 329– 338 335

lts on

dseTteabs

5

nflyedid

veoiiia

F

Fig. 2. Forest plot meta-analysis resu

There was no statistical evidence of heterogeneity (Q = 6.12,f = 3, p = .10; I2 = 51.00; T2 = 0.04); however as there were only fourtudies the precision of T2 is poor and the insignificance of the het-rogeneity test might be due to the low power. According to Egger’sest (p = 0.004), Duval and Tweedie’s trim and fill (0 studies arerimmed to the right and 2 to the left of the mean, changing thestimate and Q from 0.54 and 6.12 to 0.61 and 12.77, respectively)nd Rosenthals’ fail safe (N = 46), there was significant publicationias. For these reasons, we did not perform additional analyses (i.e.ubgroup analyses and meta-regression).

. Discussion

This is the first meta-analysis to show that physical activity canot only prevent but also slows down the disease and age-related

unctional decline that leads to basic ADL disability up to 10 yearsater. The preventive effect was found in both older (≥75 years) andounger (<74 years) older adults, in individuals with or without dis-ases, and in older adults who already had functional limitations orisability. Studies were consistent in reporting that physical activ-

ty positively influences pathways between aging or disease andisability.

These results emphasize increasing physical activity levels areital for reducing disability and healthcare costs in an aging soci-ty. Most older adults in the included studies were over 70 yearsf age at baseline, indicating that also for these age groups hav-ng an active lifestyle offers a possibility to actively influence their

ndependence. The prevention of new-onset basic ADL disabilitys important because this type of disability in particular is stronglyssociated with receiving home-care services (LaPlante et al., 2002),

ig. 3. Forest plot meta-analysis results on progression of disability in basic ADL.

incidence of disability in basic ADL.

the risk of long-term nursing home admission (Gill et al., 2006), andhealthcare costs (Fried et al., 2001; Wang et al., 2002). Despite thesebeneficial effects of physical activity, older adults are the least phys-ically active age group (Davis and Fox, 2007; Hansen et al., 2011;Troiano et al., 2008), and especially when they suffer from chronicdiseases (Harris et al., 2009). It is not easy to get older adults tobecome and stay physically active (Stiggelbout et al., 2005; Taket al., 2012). Although the reviewed studies measured differenttypes of activities, almost all of them included walking. Promot-ing accessible, popular, and everyday activities such as walking andactive form of transport provides opportunities to increase habit-ual physical activity levels (Belanger et al., 2011; Davis et al., 2011;Michael et al., 2010; Simonsick et al., 2005).

On the basis of the studies reviewed, we may conclude that anyactivity above a low level is beneficial. In most studies a low levelwas defined as being sedentary or a lack of habit of being physicallyactive. A moderate level would constitute a regular habit of beingphysically at least 1 time a week (Avlund et al., 2002) to 3 days aweek (Ostbye et al., 2002) while a vigorous level would mean atleast 3 days a week being active with a vigorous intensity (Jenkins,2004; Ostbye et al., 2002). However, these findings need to be takenwith caution as the selected studies used different ways to measureand define physical activity levels. Also there was a lack in reportingof the different aspects of physical activity such as types, frequency,intensity and duration. More research and probably consensus isrequired in defining and measuring physical activity to provide aclear advice on what constitutes medium or high levels of physi-cal activity in older adults. Current guidelines advice older adultsto be physically active at a moderate level, preferably daily, butat least 150 min per week spread out over 5 days of 30 min each(Department of Health, Physical Activity, Health Improvement andProtection, 2011; Nelson et al., 2007).

Earlier reviews concluded that the evidence for an effect of phys-ical activity on basic ADL disability was limited and difficult tosynthesize (Keysor, 2003; Paterson and Warburton, 2010; Tas et al.,2007). Our study clearly fills this gap in knowledge since it is thefirst meta-analysis to report an association between physical activ-ity and basic ADL disability and distinguished between the onset ofdisability and the progression of disability.

To fully use the potential of physical activity in preventingor reducing disability a further understanding of the underlyingmechanisms is needed. It is likely that the major influence ofregular physical activity on disability is mainly through delayingthe start of the disablement process as it prevents (chronic) dis-

eases (Brown et al., 2007; Chodzko-Zajko et al., 2009; Warburtonet al., 2006), slows down the aging process and prevents disuse.Next, there is also evidence that physical activity has direct bene-ficial effects, on impairments such as muscle strength and aerobic

3 rch Re

cltlmPsaimppbmpcpaiiouif

tsrd

twpsd(T

6

atspaSsdoestwc

bcaVotcto

36 E. Tak et al. / Ageing Resea

apacity (Latham et al., 2003; Lemura et al., 2000) and on functionalimitations (Keysor, 2003; Paterson and Warburton, 2010), Func-ional limitations (such as lower extremity limitations, mobilityimitations) have been shown to act as a mediator between impair-

ents and disability (Lawrence and Jette, 1996; Miller et al., 2000;opa et al., 2009). Our results show that even in those longitudinaltudies that adjusted for functional limitations, being physicallyctive still resulted in a reduced risk of basic ADL disability. Thenterplay between different determinants of disability is probably

ore complex than the chronological pathway of the disablementrocess projects. Therefore alternative pathways of the effect ofhysical activity on disability cannot be ruled out and should alsoe part of further investigation. For instance, being physically activeay serve as a proxy for a broader healthy lifestyle which leads to

revention of negative consequences. A lifelong history of physi-al activity may have a stronger effect on delaying the disablementrocess than current physical activity status. The effect of physicalctivity has also been shown to work through other, psycholog-cal, pathways such as depression and self-efficacy. Depressionnfluences the level of physical activity and the subsequent riskf disability (Penninx et al., 1999). And finally, a lack of completenderstanding of which confounding factors are related to disabil-

ty may lead to overestimation of the contribution of current knownactors related to disability.

The reduced risk of progression of disability is clinically impor-ant. Older adults who are disabled can recover from a disabledtate (Hardy et al., 2005; Nusselder et al., 2008) but the risk ofemaining disabled increases with time and chances of recoveryecrease (Nikolova et al., 2011; Wolinsky et al., 2011).

In order to sustain these effects, effective strategies are neededo prevent disability from becoming persistent in older adults,hich results in dependence and a need for care. In addition tohysical activity promotion (Cress et al., 2005), clinical trials havehown that exercise programs are effective in slowing down theisablement process in older adults with disabling chronic diseasesBinder et al., 2002; Chodzko-Zajko et al., 2009; Penninx et al., 2001;ak et al., 2005) as well as in frail older adults (Binder et al., 2002).

. Limitations

The electronic database search identified relatively few articles,s reported by others (Paterson and Warburton, 2010) and morehan half of the selected studies were recovered by the additionalearch strategy of cross-referencing. Search strategies are ham-ered by the lack of standardized definitions of key terms, suchs basic ADL disability, and liberal use of these terms by authors.ome of the studies included in this analysis may have suffered fromelective sampling. For instance, only participants without missingata on key variables were included in the studied cohorts. Drop-uts or non-selected cases were rarely discussed, so we could notstablish whether selective non-response or dropout occurred. Ifelective dropout was reported and corrected for in the analysis,he association between physical activity and basic ADL disabilityas weaker (Van Den Brink et al., 2005). This might have been the

ase in more included studies.All studies measured physical activity and basic ADL disability

y means of self-report information. Although there seems to beonsensus that self-reported information about disability is reli-ble, objective tests have been developed (Cress et al., 1996; dereede et al., 2006; Kuriansky et al., 1976). While reliable andbjective measures of physical activity do exist (Ainsworth, 2009),

hey may be ineffective when used in large cohorts. It was diffi-ult to compare the low, medium, or high classifications used inhe different studies, which may have resulted in overestimationr underestimation of the physical activity level in some studies.

views 12 (2013) 329– 338

In the future, continuous measures of physical activity could beused (Boyle et al., 2007), standard validated instruments (eitherquestionnaires or objective measures) should be used, or the com-parability of physical activity data could be increased by usingresponse conversion techniques (Hopman-Rock et al., 2012).

In most prospective longitudinal studies, baseline physicalactivity levels were compared with disability status 3–10 yearslater. Physical activity behaviors (Picavet et al., 2011) and disabil-ity (Hardy and Gill, 2004; Hardy et al., 2005) change over time,especially in older populations that suffer from newly acquired dis-eases or the general consequences of aging. Although some of theincluded studies took these changes into account (Nusselder et al.,2008) or used multiple measurements (Liao et al., 2011; Unger et al.,1997), most used only a single point in time to define disability sta-tus. Also as most studies did not adjust for the influence of lifelongphysical activity, the effect we found might be overestimated.

Lastly, was the heterogeneity of the included studies in themeta-analysis too large to pool the data, as discussed by Patersonand Warburton (2010) and by Keysor (2003). In order to reduceheterogeneity, we used only one outcome, basic ADL disability,distinguished between the incidence and progression of basic ADLdisability, and we only included studies that presented odds/riskratios, so that the data could be pooled. We used a random-effectsmodel and did not assume a true effect as in the fixed-effect model.We also evaluated the differences in mean age of the study popu-lations, follow-up period, and confounding in subgroup analyses(and meta-regression) and found no significant effect on the onsetof disability. That study quality did not have an effect is probablybecause of the low number of studies that was included.

7. Conclusion

This is the first meta-analysis to show that being physicallyactive at a medium/high level reduces not only the risk of becom-ing disabled by approximately one half (compared to a low physicalactivity level), but also the risk of progression of basic ADL disabilityin community dwelling adults. The preventive effect on incidencewas irrespective of age group, follow-up period, and other factorsrelated to disability. This result adds to the body of evidence show-ing that sufficient physical activity may slow down the disablementprocess in aging or diseased populations, making increasing phys-ical activity an effective strategy to reduce disability, and thereby,independence and healthcare costs in aging societies. Efforts needto be intensified to decrease inactivity and to increase habitualactivities, such as walking in older adults with and without dis-ability.

Conflict of interest

No conflict of interest reported. Erwin Tak had full access to allthe data in the study and takes responsibility for the integrity ofthe data and the accuracy of the data analysis.

Acknowledgements

Erwin Tak initiated, designed and conducted the systematicreview and prepared the manuscript. Rebecca Kuiper performedthe meta-analysis and co-wrote the analysis and results section ofthe manuscript. Astrid Chorus helped in interpretation of the dataand preparing the manuscript. Marijke Hopman-Rock conductedthe systematic review (selection and reviewing of studies) and

helped in preparing the manuscript. Lidy-Marie Ouwehand, librar-ian at the Netherlands organization for applied scientific researchTNO, Department Lifestyle, Leiden, the Netherlands, performed theelectronic literature search.

rch Re

A

t

R

A

A

B

B

B

B

C

C

C

C

D

D

d

D

F

F

F

F

G

H

H

H

H

E. Tak et al. / Ageing Resea

ppendix A. Supplementary data

Supplementary data associated with this article can be found, inhe online version, at http://dx.doi.org/10.1016/j.arr.2012.10.001.

eferences

insworth, B.E., 2009. How do I measure physical activity in my patients? Question-naires and objective methods. British Journal of Sports and Medicine 43 (1), 6–9,http://dx.doi.org/10.1136/bjsm.2008.052449.

vlund, K., Damsgaard, M.T., Sakari-Rantala, R., Laukkanen, P., Schroll, M.,2002. Tiredness in daily activities among nondisabled old people as deter-minant of onset of disability. Journal of Clinical Epidemiology 55 (10),965–973.

elanger, M., Townsend, N., Foster, C., 2011. Age-related differences in physicalactivity profiles of English adults. Preventive Medicine 52 (3–4), 247–249.

inder, E.F., Schechtman, K.B., Ehsani, A.A., Steger-May, K., Brown, M., Sinacore,D.R., Yarasheski, K.E., Holloszy, J.O., 2002. Effects of exercise training on frailtyin community-dwelling older adults: results of a randomized, controlled trial.Journal of American Geriatrics Society 50 (12), 1921–1928.

oyle, P.A., Buchman, A.S., Wilson, R.S., Bienias, J.L., Bennett, D.A., 2007. Physicalactivity is associated with incident disability in community-based older persons.Journal of American Geriatrics Society 55 (2), 195–201.

rown, W.J., Burton, N.W., Rowan, P.J., 2007. Updating the evidence on physicalactivity and health in women. American Journal of Preventive Medicine 33 (5),404–411.

aspersen, C.J., Bloemberg, B.P., Saris, W.H., Merritt, R.K., Kromhout, D., 1991. Theprevalence of selected physical activities and their relation with coronary heartdisease risk factors in elderly men: the Zutphen Study, 1985. American Journalof Epidemiology 133 (11), 1078–1092.

hodzko-Zajko, W.J., Proctor, D.N., Fiatarone Singh, M.A., Minson, C.T., Nigg, C.R.,Salem, G.J., Skinner, J.S., 2009. American College of Sports Medicine positionstand. Exercise and physical activity for older adults. Medicine and Science inSports and Exercise 41 (7), 1510–1530.

ress, M.E., Buchner, D.M., Prohaska, T., Rimmer, J., Brown, M., Macera, C., Dipietro,L., Chodzko-Zajko, W., 2005. Best practices for physical activity programs andbehavior counseling in older adult populations. Journal of Aging and PhysicalActivity 13 (1), 61–74.

ress, M.E., Buchner, D.M., Questad, K.A., Esselman, P.C., deLateur, B.J., Schwartz,R.S., 1996. Continuous-scale physical functional performance in healthy olderadults: a validation study. Archives of Physical Medicine and Rehabilitation 77(12), 1243–1250.

avis, M.G., Fox, K.R., 2007. Physical activity patterns assessed by accelerometry inolder people. European Journal of Applied Physiology 100 (5), 581–589.

avis, M.G., Fox, K.R., Hillsdon, M., Coulson, J.C., Sharp, D.J., Stathi, A., Thompson, J.L.,2011. Getting out and about in older adults: the nature of daily trips and theirassociation with objectively assessed physical activity. International Journal ofBehavioral Nutrition and Physical Activity 8, 116.

e Vreede, P.L., Samson, M.M., van Meeteren, N.L., Duursma, S.A., Verhaar, H.J., 2006.Reliability and validity of the Assessment of Daily Activity Performance (ADAP)in community-dwelling older women. Aging Clinical and Experimental Research18 (4), 325–333.

epartment of Health, Physical Activity, Health Improvement and Protection, 2011.Start active, Stay active: a report on physical activity from the four homecountries’ Chief Medical Officers, 16306.

reedman, V.A., 2009. Adopting the ICF language for studying late-life disability: afield of dreams? Journal of Gerontology: Series A Biological Sciences and MedicalSciences 64 (11), 1172–1174.

reedman, V.A., Martin, L.G., Schoeni, R.F., 2002. Recent trends in disability and func-tioning among older adults in the United States: a systematic review. JAMA 288(24), 3137–3146.

ried, T.R., Bradley, E.H., Williams, C.S., Tinetti, M.E., 2001. Functional disability andhealth care expenditures for older persons. Archives of Internal Medicine 161(21), 2602–2607.

uller-Thomson, E., Yu, B., Nuru-Jeter, A., Guralnik, J.M., Minkler, M., 2009. Basic ADLdisability and functional limitation rates among older AMERICANS from 2000-2005: the end of the decline? Journal of Gerontology: Series A Biological Sciencesand Medical Sciences 64 (12), 1333–1336.

ill, T.M., Allore, H.G., Han, L., 2006. Bathing disability and the risk of long-termadmission to a nursing home. Journal of Gerontology: Series A Biological Sci-ences and Medical Sciences 61 (8), 821–825.

ansen, B.H., Kolle, E., Dyrstad, S.M., Holme, I., Anderssen, S.A., 2011. Accelerometer-Determined Physical Activity in Adults and Older People. Medicine and Sciencein Sports and Exercise 44 (2), 266–272.

ardy, S.E., Dubin, J.A., Holford, T.R., Gill, T.M., 2005. Transitions between states ofdisability and independence among older persons. American Journal of Epidemi-ology 161 (6), 575–584.

ardy, S.E., Gill, T.M., 2004. Recovery from disability among community-dwelling

older persons. JAMA 291 (13), 1596–1602.

arris, T.J., Owen, C.G., Victor, C.R., Adams, R., Ekelund, U., Cook, D.G., 2009.A comparison of questionnaire, accelerometer, and pedometer: measures inolder people. Medicine and Science in Sports and Exercise 41 (7), 1392–1402,http://dx.doi.org/10.1249/MSS.0b013e31819b3533.

views 12 (2013) 329– 338 337

Haveman-Nies, A., De Groot, L.C., Van Staveren, W.A., 2003. Relation of dietary qual-ity, physical activity, and smoking habits to 10-year changes in health status inolder Europeans in the SENECA study. American Journal of Public Health 93 (2),318–323.

Hoeymans, N., Feskens, E.J., van den Bos, G.A., Kromhout, D., 1996. Measuringfunctional status: cross-sectional and longitudinal associations between perfor-mance and self-report (Zutphen Elderly Study 1990–1993). Journal of ClinicalEpidemiology 49 (10), 1103–1110.

Hopman-Rock, M., Dusseldorp, E., Chorus, A., Jacobusse, G., Ruetten, A., van Buuren,S., 2012. Response conversion for improving comparability of internationalphysical activity data. Journal of Physical Activity and Health 9 (1), 29–38.

Ishizaki, T., Watanabe, S., Suzuki, T., Shibata, H., Haga, H., 2000. Predictors for func-tional decline among nondisabled older Japanese living in a community during a3-year follow-up. Journal of the American Geriatrics Society 48 (11), 1424–1429.

Jenkins, K.R., 2004. Obesity’s effects on the onset of functional impairment amongolder adults. Gerontologist 44 (2), 206–216.

Johnson, R.J., Wolinsky, F.D., 1993. The structure of health status among older adults:disease, disability, functional limitation, and perceived health. Journal of Healthand Social Behavior 34 (2), 105–121.

Katz, S., Ford, A.B., Moskowitz, R.W., Jackson, B.A., Jaffe, M.W., 1963. Studies of ill-ness in the aged. The index of ADL: a standardized measure of biological andphysiological function. JAMA 185, 914–919.

Keysor, J.J., 2003. Does late-life physical activity or exercise prevent or minimizedisablement? A critical review of the scientific evidence. American Journal ofPreventive Medicine 25 (3), 129–136.

Kuriansky, J.B., Gurland, B.J., Fleiss, J.L., 1976. The assessment of self-care capacityin geriatric psychiatric patients by objective and subjective methods. Journal ofClinical Psychology 32 (1), 95–102.

LaPlante, M.P., Harrington, C., Kang, T., 2002. Estimating paid and unpaid hours ofpersonal assistance services in activities of daily living provided to adults livingat home. Health Services Research 37 (2), 397–415.

Latham, N., Anderson, C., Bennett, D., Stretton, C., 2003. Progressive resistancestrength training for physical disability in older people. Cochrane Database ofSystematic Reviews (2), CD002759.

Lawrence, R.H., Jette, A.M., 1996. Disentangling the disablement process. Journalof Gerontology: Series B Psychological Sciences and Social Sciences 51 (4),S173–S182.

Lemura, L.M., von Duvillard, S.P., Mookerjee, S., 2000. The effects of physical trainingof functional capacity in adults,ages 46 to 90: a meta-analysis. Journal of SportsMedicine and Physical Fitness 40 (1), 1–10.

Leveille, S.G., Guralnik, J.M., Ferrucci, L., Langlois, J.A., 1999. Aging successfully untildeath in old age: opportunities for increasing active life expectancy. AmericanJournal of Epidemiology 149 (7), 654–664.

Liao, W.C., Li, C.R., Lin, Y.C., Wang, C.C., Chen, Y.J., Yen, C.H., Lin, H.S., Lee, M.C.,2011. Healthy behaviors and onset of functional disability in older adults: resultsof a national longitudinal study. Journal of American Geriatrics Society 59 (2),200–206.

Liberati, A., Altman, D.G., Tetzlaff, J., Mulrow, C., Gotzsche, P.C., Ioannidis, J.P., Clarke,M., Devereaux, P.J., Kleijnen, J., Moher, D., 2009. The PRISMA statement forreporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration. Journal of Clinical Epidemiol-ogy 62 (10), e1–e34.

Manini, T.M., Pahor, M., 2009. Physical activity and maintaining physical func-tion in older adults. British Journal of Sports and Medicine 43 (1), 28–31,http://dx.doi.org/10.1136/bjsm.2008.053736.

Michael, Y.L., Perdue, L.A., Orwoll, E.S., Stefanick, M.L., Marshall, L.M., 2010. Physicalactivity resources and changes in walking in a cohort of older men. AmericanJournal of Public Health 100 (4), 654–660.

Miller, M.E., Rejeski, W.J., Reboussin, B.A., Ten Have, T.R., Ettinger, W.H., 2000. Phys-ical activity, functional limitations, and disability in older adults. Journal ofAmerican Geriatrics Society 48 (10), 1264–1272.

Nagi, S.Z., 1976. An epidemiology of disability among adults in the United States.Milbank Memorial Fund Quarterly Health and Society 54 (4), 439–467.

Nelson, M.E., Rejeski, W.J., Blair, S.N., Duncan, P.W., Judge, J.O., King, A.C., Macera,C.A., Castaneda-Sceppa, C., 2007. Physical activity and public health in olderadults: recommendation from the American College of Sports Medicine and theAmerican Heart Association. Medicine and Science in Sports and Exercise 39 (8),1435–1445.

Nikolova, R., Demers, L., Beland, F., Giroux, F., 2011. Transitions in the functionalstatus of disabled community-living older adults over a 3-year follow-up period.Archives of Gerontology and Geriatrics 52 (1), 12–17.

Nusselder, W.J., Looman, C.W., Franco, O.H., Peeters, A., Slingerland, A.S., Macken-bach, J.P., 2008. The relation between non-occupational physical activity andyears lived with and without disability. Journal of Epidemiology and CommunityHealth 62 (9), 823–828.

Ostbye, T., Taylor, D.H., Jung, S.H., 2002. A longitudinal study of the effects of tobaccosmoking and other modifiable risk factors on ill health in middle-aged and oldAmericans: results from the Health and Retirement Study and Asset and HealthDynamics among the Oldest Old survey. Preventive Medicine 34 (3), 334–345.

Paterson, D.H., Warburton, D.E., 2010. Physical activity and functional limitations inolder adults: a systematic review related to Canada’s Physical Activity Guide-

lines. International Journal of Behavioral Nutrition and Physical Activity 7, 38.

Penninx, B.W., Leveille, S., Ferrucci, L., van Eijk, J.T., Guralnik, J.M., 1999. Explor-ing the effect of depression on physical disability: longitudinal evidence fromthe established populations for epidemiologic studies of the elderly. AmericanJournal of Public Health 89 (9), 1346–1352.

3 rch Re

P

P

P

P

P

P

S

S

S

S

S

S

S

S

38 E. Tak et al. / Ageing Resea

enninx, B.W., Messier, S.P., Rejeski, W.J., Williamson, J.D., DiBari, M., Cavazzini,C., Applegate, W.B., Pahor, M., 2001. Physical exercise and the prevention ofdisability in activities of daily living in older persons with osteoarthritis. Archivesof Internal Medicine 161 (19), 2309–2316.

icavet, H.S., Hoeymans, N., 2002. Physical disability in The Netherlands: prevalence,risk groups and time trends. Public Health 116 (4), 231–237.

icavet, H.S., Wendel-vos, G.C., Vreeken, H.L., Schuit, A.J., Verschuren, W.M.,2011. How stable are physical activity habits among adults? The Doet-inchem Cohort Study. Medicine and Science in Sports and Exercise 43 (1),74–79.

opa, M.A., Reynolds, S.L., Small, B.J., 2009. Is the effect of reported physical activityon disability mediated by cognitive performance in white and african americanolder adults? Journal of Gerontology A Biological Sciences and Medical Sciences64 (1), 4–13.

uts, M.T., Deeg, D.J., Hoeymans, N., Nusselder, W.J., Schellevis, F.G., 2008. Changesin the prevalence of chronic disease and the association with disability inthe older Dutch population between 1987 and 2001. Age and Ageing 37 (2),187–193.

uts, M.T., Lips, P., Deeg, D.J., 2005. Static and dynamic measures of frailty predicteddecline in performance-based and self-reported physical functioning. Journal ofClinical Epidemiology 58 (11), 1188–1198.

choeni, R.F., Freedman, V.A., Martin, L.G., 2008. Why is late-life disability declining?Milbank Quarterly 86 (1), 47–89.

imonsick, E.M., Guralnik, J.M., Volpato, S., Balfour, J., Fried, L.P., 2005. Just get out thedoor! Importance of walking outside the home for maintaining mobility: find-ings from the women’s health and aging study. Journal of American GeriatricsSociety 53 (2), 198–203.

ingh, A.S., Mulder, C., Twisk, J.W., van Mechelen, W., Chinapaw, M.J., 2008. Trackingof childhood overweight into adulthood: a systematic review of the literature.Obesity Reviews 9 (5), 474–488.

tel, V.S., Smit, J.H., Pluijm, S.M., Visser, M., Deeg, D.J., Lips, P., 2004.Comparison of the LASA Physical Activity Questionnaire with a 7-day diary and pedometer. Journal of Clinical Epidemiology 57 (3),252–258.

tessman, J., Hammerman-Rozenberg, R., Cohen, A., Ein-Mor, E., Jacobs, J.M., 2009.Physical activity, function, and longevity among the very old. Archives of InternalMedicine 169 (16), 1476–1483.

tessman, J., Hammerman-Rozenberg, R., Maaravi, Y., Cohen, A., 2002. Effect of exer-cise on ease in performing activities of daily living and instrumental activitiesof daily living from age 70 to 77: the Jerusalem longitudinal study. Journal ofAmerican Geriatrics Society 50 (12), 1934–1938.

tiggelbout, M., Hopman-Rock, M., Tak, E., Lechner, L., van Mechelen, W., 2005.Dropout from exercise programs for seniors: a prospective cohort study. Journal

of Aging and Physical Activity 13 (4), 406–421.

troup, D.F., Berlin, J.A., Morton, S.C., Olkin, I., Williamson, G.D., Rennie, D., Moher,D., Becker, B.J., Sipe, T.A., Thacker, S.B., 2000. Meta-analysis of observationalstudies in epidemiology, (MOOSE) group: a proposal for reporting. JAMA 283(15), 2008–2012.

views 12 (2013) 329– 338

Stuck, A.E., Walthert, J.M., Nikolaus, T., Bula, C.J., Hohmann, C., Beck, J.C., 1999. Riskfactors for functional status decline in community-living elderly people: a sys-tematic literature review. Social Science and Medicine 48 (4), 445–469.

Tak, E., Staats, P., Van Hespen, A., Hopman-Rock, M., 2005. The effects of an exerciseprogram for older adults with osteoarthritis of the hip. Journal of Rheumatology32 (6), 1106–1113.

Tak, E.C.P., van Uffelen, J.G.Z., Chin, A., Paw, M.J., van Mechelen, W., Hopman-Rock,M., 2012. Adherence to exercise programs and determinants of maintenancein older adults with mild cognitive impairment. Journal of Aging and PhysicalActivity 20 (1), 32–46.

Tas, U., Verhagen, A.P., Bierma-Zeinstra, S.M., Odding, E., Koes, B.W., 2007. Progno-stic factors of disability in older people: a systematic review. British Journal ofGeneral Practice 57 (537), 319–323.

Troiano, R.P., Berrigan, D., Dodd, K.W., Masse, L.C., Tilert, T., McDowell, M., 2008.Physical activity in the United States measured by accelerometer. Medicine andScience in Sports and Exercise 40 (1), 181–188.

Unger, J.B., Johnson, C.A., Marks, G., 1997. Functional decline in the elderly: evi-dence for direct and stress-buffering protective effects of social interactions andphysical activity. Annals of Behavioral Medicine 19 (2), 152–160.

Van Den Brink, C.L., Picavet, H., Van Den Bos, G.A., Giampaoli, S., Nissinen, A.,Kromhout, D., 2005. Duration and intensity of physical activity and disabilityamong European elderly men. Disability and Rehabilitation 27 (6), 341–347.

Verbrugge, L.M., Jette, A.M., 1994. The disablement process. Social Science andMedicine 38 (1), 1–14.

Vermeulen, J., Neyens, J.C., van, R.E., Spreeuwenberg, M.D., de Witte, L.P., 2011.Predicting ADL disability in community-dwelling elderly people using physicalfrailty indicators: a systematic review. BMC Geriatrics 11, 33.

Voorrips, L.E., Ravelli, A.C., Dongelmans, P.C., Deurenberg, P., Van Staveren, W.A.,1991. A physical activity questionnaire for the elderly. Medicine and Science inSports and Exercise 23 (8), 974–979.

Wang, B.W., Ramey, D.R., Schettler, J.D., Hubert, H.B., Fries, J.F., 2002. Postponeddevelopment of disability in elderly runners: a 13-year longitudinal study.Archives of Internal Medicine 162 (20), 2285–2294.

Warburton, D.E., Nicol, C.W., Bredin, S.S., 2006. Health benefits of physical activity:the evidence. CMAJ 174 (6), 801–809.

Wolinsky, F.D., Bentler, S.E., Hockenberry, J., Jones, M.P., Obrizan, M., Weigel, P.A.,Kaskie, B., Wallace, R.B., 2011. Long-term declines in ADLs, IADLs, and mobilityamong older Medicare beneficiaries. BMC Geriatrics 11, 43.

Wolinsky, F.D., Johnson, R.J., 1991. The use of health services by older adults. Journalof Gerontology 46 (6), S345–S357.

Wolinsky, F.D., Stump, T.E., Clark, D.O., 1995. Antecedents and consequences ofphysical activity and exercise among older adults. Gerontologist 35 (4), 451–462.

Wu, S.C., Leu, S.Y., Li, C.Y., 1999. Incidence of and predictors for chronic disability

in activities of daily living among older people in Taiwan. Journal of AmericanGeriatrics Society 47 (9), 1082–1086.

Young, D.R., Masaki, K.H., Curb, J.D., 1995. Associations of physical activity withperformance-based and self-reported physical functioning in older men: theHonolulu Heart Program. Journal of American Geriatrics Society 43 (8), 845–854.