Archives of Gerontology and Geriatrics 59 (2014) 195–203

Review

What are the factors associated with physical activity (PA)participation in community dwelling adults with dementia?A systematic review of PA correlates

Brendon Stubbs a,*, Laura Eggermont b, Andrew Soundy c, Michel Probst d,e,Mathieu Vandenbulcke f, Davy Vancampfort d,e

a School of Health and Social Care, University of Greenwich, Southwood Site Avery Hill Road Eltham, London SE9 2UG, UKb Department of Clinical Neuropsychology, VU University Amsterdam, Van der Boechorststraat 1, 1081 BT Amsterdam, The Netherlandsc Department of Physiotherapy, University of Birmingham, 52 Pritchatts Road, Birmingham B15 2TT, UKd University Psychiatric Centre, KU Leuven, Kortenberg, Department of Neurosciences, Leuvensesteenweg 517, B-3070 Kortenberg, Belgiume Department of Rehabilitation Sciences, KU Leuven, Tervuursevest 101, B-3001 Leuven, Belgiumf Old Age Psychiatry, University Hospitals Leuven, & Department of Neurosciences, KU Leuven, Herestraat 49, B-3000, Belgium

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196

2. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196

2.1. Data sources and searches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196

2.2. Eligibility criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196

2.3. Data collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

2.4. Selection and categorization of variables. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

2.5. Coding associations with PA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

2.6. Summary codes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

3.1. Study selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

A R T I C L E I N F O

Article history:

Received 25 March 2014

Accepted 24 June 2014

Available online 1 July 2014

Keywords:

PA

Dementia

Exercise

Alzheimer’s disease (AD)

A B S T R A C T

PA shows promise as a modifiable lifestyle intervention to benefit pathological symptoms of dementia.

However, little is known about the factors associated with participation in PA in community dwelling

adults with dementia. A systematic review was undertaken to identify PA correlates. Two independent

reviewers searched major electronic databases and extracted data on studies reporting quantitative

correlates of PA participation in community dwelling adults with dementia. PA correlates were analyzed

using the summary code approach within the socio-ecological model. Out of a potential of 118 articles,

12 met the eligibility criteria encompassing 752 participants. We conducted secondary analysis on nine

data sets. Increased energy intake, resting metabolic rate, fat free mass, gait speed, global motor function,

overall health related quality of life (HRQOL), physical HRQOL, higher levels of social functioning and

reduced apathy were positively associated with PA. Taking �four medications, dizziness, lower activities

of daily living (ADL) function, a history of falls, less waking hours in the day, more autonomic problems

and delirium were negatively associated with PA. Increasing age and lower global cognition were not

consistently associated with PA participation. It is surprising that increasing age and lower global

cognition do not appear to influence PA participation. All significant correlates should be confirmed in

prospective studies with particular focus on the relationship of PA and gait speed, ADL function, falls

history and dietary intake and the progression of frailty and nursing home admission as a priority.

� 2014 Elsevier Ireland Ltd. All rights reserved.

Contents lists available at ScienceDirect

Archives of Gerontology and Geriatrics

jo ur n al ho mep ag e: www .e lsev ier . c om / lo cate /ar c hg er

* Corresponding author. Tel.: +44 208 80300; fax: +44 1604 696 484.

E-mail addresses: b.stubbs@greenwich.ac.uk, brendonstubbs@hotmail.com (B. Stubbs).

http://dx.doi.org/10.1016/j.archger.2014.06.006

0167-4943/� 2014 Elsevier Ireland Ltd. All rights reserved.

B. Stubbs et al. / Archives of Gerontology and Geriatrics 59 (2014) 195–203196

3.2. Participant and study characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

3.3. Correlates of PA in community dwelling adults with dementia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198

3.4. Demographic correlates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198

3.5. Biological correlates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198

3.6. Behavioral attributes and skills . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200

3.7. Psychological, cognitive and emotional correlates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200

3.8. Social and cultural factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200

3.9. Physical environment and policy factor correlates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200

3.10. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200

3.11. Strengths and limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201

3.12. Clinical overview and implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202

4. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202

1. Introduction

The aging population and the number of people affected bydementia are reaching epidemic levels (Prince et al., 2013). Theinherent cognitive decline observed in dementia is associated withprofound loss of independence, increased falls risk, a reduction incapacity to undertake ADL, nursing home admission and increasedmortality (Butler & Radhakrishnan, 2012; Pitkala, Savikko, Poysti,Strandberg, & Laakkonen, 2013; Prince et al., 2013; Smith, Nielson,Woodard, Seidenberg, & Rao, 2013). Unsurprisingly, the preventionand management of dementia is an International political andhealth priority (Prince et al., 2013).

Within recent years interest has risen in PA as an effective non-pharmacological intervention in the prevention and managementof dementia (Ahlskog, Geda, Graff-Radford, & Petersen, 2011). Forinstance, Erickson, Weinstein, and Lopez (2012) recently conclud-ed that there is convincing evidence that PA reduces the risk of AD,improves brain health and decreases the risk of pathologicalsymptoms associated with AD. Other recent reviews (Blankevoortet al., 2010; Pitkala, Savikko, Poysti, Strandberg, & Laakkonen,2013; Potter, Ellard, Rees, & Thorogood, 2011) have establishedthat PA interventions can improve mobility and functionallimitations in people with dementia. This is exemplified in arecent randomized control trial (Hauer et al., 2012) involving PAwhich resulted in significant improvements in strength (p < 0.001)and functional performance (p < 0.001) in community dwellingadults with mild and moderate dementia. In a recent updatedCochrane review, Forbes et al. (2013) concluded that structured PA(exercise) can have a positive influence on cognition and a personwith dementia’s ability to undertake their ADL. In addition, arecent randomized controlled (RCT) (Lowery et al., 2013)demonstrated that PA program for people with dementia resultedin a reduction of caregiver burden from 23% to 17% (p = 0.01) whilethe burden doubled in the control arm.

Despite these positive findings, there remain concerns thatcommunity-dwelling adults with dementia are physically inactive(James, Boyle, Bennett, & Buchman, 2012). A recent randomizedcontrol trial found that only 30.7% achieved the prescribedfrequency of the PA intervention and the authors concluded thatit is essential research is undertaken to identify factors influencingparticipation in PA in community dwelling adults with dementia(Lowery et al., 2013). A systematic review of quantitative researchwould identify potential factors associated with PA and thisinformation can be used to target future PA interventions forpersons with dementia. One approach that is potentially useful tounderstand the factors associated with PA uptake is the socio-ecological model (Sallis et al., 2006). This model posits thatmultiple levels may impact upon an individual’s participation inPA, including intrapersonal factors (e.g. socio-demographic orpsychological) factors, interpersonal factors (e.g. social support);

environment factors (e.g. access to facilities) and policy factors (e.g.regulations) (Bauman et al., 2012; Sallis et al., 2006). The socio-ecological model has been successfully used to identify influentialfactors in the uptake of PA within the general population (Baumanet al., 2012; Sallis et al., 2006), persons with mental illness(Vancampfort et al., 2014). Given the promising benefits associatedwith PA for community dwelling adults with dementia, it isimportant to understand the multiple levels over which PAparticipation may be affected.

To the best of our knowledge, no author has systematicallyreviewed the literature to establish the factors associated with PAparticipation in community dwelling adults with dementia. Thisinformation would provide essential information that can beutilized by clinicians, researchers and policy makers. With this inmind we conducted a systematic review to establish quantitativecorrelates of PA participation in community dwelling adults withdementia utilizing the socio-ecological approach.

2. Methods

2.1. Data sources and searches

Two independent reviewers conducted an electronic search ofPsycINFO, CINAHL, PubMed and Embase from the inception ofthese databases until October 2013. The medical subject headingsused were ‘PA’ or ‘exercise’ or ‘physical inactivity’ or ‘sedentary’and ‘dementia’, or ‘AD’ or ‘vascular dementia (VAD)’ or ‘fronto-temporal dementia’ or ‘Lewy Body dementia’ in the title, abstractor index term fields. We also conducted manual searches of thereference lists of identified articles.

2.2. Eligibility criteria

Inclusion criteria were: (a) studies published in a peer reviewjournal providing quantitative data, (b) including communitydwelling adults with a confirmed diagnosis of dementia (assessedwith a structured clinical diagnostic interview, or the Diagnosticand Statistical Manual of Mental Disorders (DSM-IV; APA, 1994),or the National Institute of Neurological and CommunicativeDisorders and Stroke and the AD and Related Disorders Associationcriteria (NINCDS-ADRDA, McKhann et al., 1984) and (c) thedependent variable was a measure of PA participation. We did notplace a restriction on the selection of the outcome measure or thelanguage of manuscripts that were considered within the review.When we encountered cohort or interventional studies, we utilizedthe PA outcome measurement and correlates from the baselinedata. Authors of all potential and included studies were contactedup to three times in order to obtain correlates if this informationnot given and in order to identify any potential additionalcorrelates not reported in the paper.

B. Stubbs et al. / Archives of Gerontology and Geriatrics 59 (2014) 195–203 197

Articles were excluded if the dependent variable was aerobicfitness, PA intention, motivation, self-efficacy or other non-behavioral measures because we aimed for variables that were adirect indicator of actual participation in PA. In addition, weexcluded case reports, conference abstracts, qualitative research,reviews and expert opinion pieces. If studies were encounteredthat included a population with multiple diagnoses (for examplemild cognitive impairment and dementia) or participants residingin mixed settings (e.g. nursing home residents and communitydwelling participants) we contacted the primary author in order toprovide us with the dementia specific information. If we did notreceive a response after three requests and it was not possible toextract the information from the paper, we subsequently excludedthe paper.

2.3. Data collection

Two reviewers independently extracted data from the includedstudies using a predetermined form. The data extracted includedsocio-demographic information (mean age, % male), details ofdementia diagnosis (including information on mini mental statescores if available) and the type and quality of the PA measure.

In accordance with previous reviews (Vancampfort et al., 2014)we coded the quality of the self-reported PA measures as follows:(a) self-report with poor, unknown or not reported reliability/validity in dementia, (b) self-report with reported and acceptablereliability/validity in dementia, and (c) acceptable objectivemeasurements for dementia. The categorization of the quality ofPA measure was conducted by two reviewers and a third reviewerwas available for guidance. We did not utilize a code foracceptability of objective based measures, rather we identifiedobjective measures as motion sensors such as accelerometers,pedometers, actigraphs and the doubly labeled water method(Warren et al., 2010). In order to establish if the quality of PAmeasure had any effect upon the proportion of significantcorrelates in each study, we conducted a Chi squared test.

We classified all variables as ‘related’ or ‘not related’ to PA basedupon statistical significance and the direction of the association foreach related variable was coded. The data was summarized foreach variable to provide an overview of the literature on eachdifferent variable (see Table 1).

2.4. Selection and categorization of variables

When we encountered studies based upon the same sampleexamining the same correlates, we included the most recent dataand/or those with the largest sample size. This approach enablesthe identification of domains which have been explored in theliterature and to elucidate the multidimensional perspective ofpotential influencing factors upon PA participation in older adultswith dementia (Bauman et al., 2012). In accordance with previousreviews (Vancampfort et al., 2014) and following the socio-ecological model, we examined and grouped PA correlates withinthe following categories: (a) demographic, (b) biological, (c)psychological/cognitive/emotional, (d) behavioral attributes/skills,(e) social/cultural factors, (f) physical environment, and (g) policyfactors.

2.5. Coding associations with PA

Within the published papers we encountered a variety ofstatistical techniques to evaluate correlates of PA, including uni-/bivariate analyses such as correlations, t-tests, and analysis ofvariance (ANOVA). Sometimes, only multivariate analyses werereported, including linear regression or logistic regression. In orderto gain consistency whenever possible we used the uni/bivariate

analysis available in the paper or those authors had provided us inresponse to our request for additional data. We contacted thecorresponding author of all included studies requesting either (a)uni/bivariate correlation with PA as the dependent variable or (b)the raw data so that we could conduct the analysis. If the author didnot respond after three requests we included the variablesavailable in the paper. If these were not available in the paperor if authors did not respond to our email requests, we excludedthe paper.

The information regarding the correlates of PA were collatedand grouped under the socio-ecological themes in Table 2. WithinTable 2 the column ‘related to PA’ denotes that there was asignificant association with PA participation and the column‘unrelated to PA’ denotes that there was a non-significantcorrelation to PA participation. For those correlates identified asbeing significantly associated with PA we employed the symbol ‘+’or ‘�’ based upon whether or not it was positively or negativelyassociated with PA participation.

2.6. Summary codes

A summary code for each correlate was calculated andpresented in accordance with the established method of Salliset al. (2006). This summary code (see Table 2) provides an overallreflection of the findings from the literature for the association ofthat variable with PA participation. We calculated total percentageof studies supporting or negating each correlate by dividing thenumber supporting an association by the overall numberinvestigating each correlate. In accordance with previous reviews(Vancampfort et al., 2014) associations were coded with: ‘00 (0–33% of studies supporting association); ‘?’ (34–59% of studiessupporting an association); or ‘+’ or ‘�’ (60–100% of studiessupporting an association). When we encountered four or morestudies supporting or refuting an association, we coded thesummary as ‘000, ‘��’, or ‘++’. The ‘??’ code indicated a variable thatwas studied four (or more) times but that there was a lack ofconsistency in the findings.

3. Results

3.1. Study selection

We identified 2603 articles from the searches and followingscreening of these we obtained the full text of 118 articles to assesstheir eligibility. At this stage, we contacted 19 research groupsrequesting additional information on the correlates with PA andnine provided us with additional analysis and/or provided theirdata set so we could conduct secondary analysis of their data set(see acknowledgements). Seven groups were not contactable orable/willing to provide additional data and were subsequentlyexcluded as there was not any information on the correlates withPA in the published paper. Three papers were still included as therewas at least one correlate of PA reported in the published paper. Intotal, 106 articles were excluded with reasons and the completesearch strategy is outlined in Fig. 1. Twelve articles met theeligibility criteria and were included in the review.

3.2. Participant and study characteristics

In total 752 unique adults with dementia were included withinour analysis and most (n = 671, 89%) had a diagnosis of AD. It waspossible to establish the percentage of males in 11 of the 12 studiesand on average 48% of the samples were male (�13%). The largestsample was the Allen et al. (2006) study (n = 156) while the smallest(n = 24) was Cedervall, Kilander, and Aberg (2012). All of the studiesconfirmed a diagnosis of dementia with the NINCDS-ADRDA criteria

Table 1Characteristics of included studies.

Study [Ref nr] Participants Quality of PA

measurement

Statistical tests

N Diagnosis Age (mean � SD) % Male

Allan et al. (2006) 156 46 PDD

32 DLB 40 AD

38 VAD

Diagnosis confirmed by 2,

or if necessary, 3 clinicians

76.2 � 6.5 y 59 A Pearson product moment

correlation coefficient

Auyeung et al. (2008) 28 Diagnosed by psychiatrist

in accordance to DSM-IV

All >70 y 54 A Pearson product moment

correlation coefficient

Cedervall et al. (2012) 24 AD

MMSE 25 (21–30)

Diagnosis according to

DSM-IV and NINCDS-

ADRDA

73 years (55–79 range) 56 A Spearman rank

correlation coefficient

Christofoletti et al. (2011) 59 23 AD

17 mixed dementia

19 VAD

DSM-IV and according to

NINCDS-ADRDA/NINCDS-

AIREN criteria

76 y NA A Kruskal–Wallis tests and

Mann Whitney U test.

Stepwise linear regression

David et al. (2012) 80 AD*

NINCDS-ADRDA criteria

MMSE 21.4 � 4.6

77.5 � 6.7 y 53 C Pearson product moment

correlation coefficient

Dvorak and Poehlman (1998) 30 AD

NINCDS-ADRDA criteria

MMSE 17 � 8

74 � 8 y 43 C Pearson product moment

correlation coefficient

Erickson et al. (2013) 26 AD

NINCDS-ADRDA criteria

76.8 � 9.3 y 69 C Pearson product moment

correlation coefficient

James et al. (2012) 70 Clinical diagnosis with

three stage process,

NINCDS-ADRDA

85.6 � 6.5 y 36 C Pearson product moment

correlation coefficient

McCurry et al. (2010) 66 AD

NINCDS-ADRDA criteria

MMSE 19.2 � 6.7

81.2 � 8.3 y 46 A Bivariate correlations

Vital et al. (2012) 37 AD

DSM-IV and CDR

MMSE 17 � 4.4

78.8 � 7 y 22 A Spearman correlation

coefficient

Watts et al. (2013) 72 Early stage AD

NINCDS-ADRDA criteria

74.9 � 6.4 y 37 A Bivariate correlations

Winchester et al. (2013) 104 AD

NINCDS-ADRDA criteria

81 � 6.54 y 56 A ANCOVA

Key: PA measurement grading: A = self-report of poor or unknown reliability/validity in people with dementia, B = self-report with acceptable reliability/validity in people

with dementia, C = objective measure of PA.

ANCOVA, analysis of covariance; CDR, clinical dementia rating scale; DLB, dementia with Lewy bodies; MDT, multidisciplinary team; MMSE, mini mental state examination;

NA, not available; PDD, Parkinson’s disease dementia; y, years.

B. Stubbs et al. / Archives of Gerontology and Geriatrics 59 (2014) 195–203198

and/or the DSM-IV. Seven studies adopted a cross sectional design(Allan, McKeith, Ballard, & Kenny, 2006; Christofoletti et al., 2011;David et al., 2012; Dvorak & Poehlman, 1998; Erickson et al., 2013;Vital et al., 2012; Watts, Vidoni, Loskutova, Johnson, & Burns, 2013),four were cohort studies (Auyeung et al., 2008; Cedervall et al., 2012;James et al., 2012; Winchester et al., 2013) and one was a RCT(McCurry et al., 2010). However all studies adopted cross-sectionalmeasurement of PA and the subsequent correlates. Four studies (33%;David et al., 2012; Dvorak & Poehlman, 1998; Erickson et al., 2013;James et al., 2012) captured PA with an objective measure, while theremaining eight studies (66%) used a range of self-report measures tocapture PA. None of the self-report PA measures were validated foruse in community dwelling adults with dementia. Table 1 presentsthe characteristics of the included participants, the quality of PAassessment and statistical analysis undertaken.

We conducted a two tailed Chi squared test and found therewas no statistical significant difference between the proportion ofrelated correlates reported in studies utilizing objective or self-report PA measures (p = 0.203).

3.3. Correlates of PA in community dwelling adults with dementia

In total 38 correlates were examined within the literatureacross five of the domains of the socio-ecological model. A

summary of the correlates with PA is presented in Table 2 andthese will be considered below.

3.4. Demographic correlates

In total six demographic correlates were examined. The mostcommonly assessed correlate was the influence of age upon PA.Increasing age was not consistently associated with PA with only 2/8 (25%, Erickson et al., 2013; James et al., 2012) finding a negativeassociation between increasing age and PA. Education wasexamined in four studies and the results were inconsistent (2/4,50%, Christofoletti et al., 2011; Vital et al., 2012 found a positiveassociation). Evidence from three studies suggests that theduration of dementia does not appear to be associated with PAparticipation (1/3, 33%, Vital et al., 2012, found a positiveassociation). Only one study investigated socioeconomic status,gender and ethnicity and none found any association with PA.

3.5. Biological correlates

In total 14 biological correlates were investigated. We did notidentify any consistent correlates (reported in four or morestudies), but there is evidence from two studies (100%, Auyeunget al., 2008; Watts et al., 2013) that faster gait speed/better walking

Table 2Summary of determinants of PA in patients with dementia.

Determinant variable Related to PA Unrelated to PA Summary code

Study Assoc. Study Assoc. % studies

reporting assoc.

Demographic

Age (higher) James et al. (2012),

Erickson et al. (2013)

� Christofoletti et al. (2011),

McCurry et al. (2010),

Auyeung et al. (2008),

Allan et al. (2006), Vital et al. (2012),

David et al. (2012)

00 25% (2/8)

Education (higher) Christofoletti et al. (2011),

Vital et al. (2012)

+ James et al. (2012), Erickson et al. (2012) ? 50% (2/4)

Duration of dementia (longer) Vital et al. (2012) + Christofoletti et al. (2011),

Allan et al. (2006)

0 33% (1/3)

Socioeconomic status Auyeung et al. (2008) 0 0% (0/1)

Gender Erickson et al. (2013) 0 0% (0/2)

Ethnicity Erickson et al. (2013) 0 0% (0/1)

Biological

BMI (higher) Auyeung et al. (2008),

Erickson et al. (2013)

� James et al. (2012), Allan et al. (2006),

Watts et al. (2013)

? 40% (2/5)

Comorbidity (higher nr. of conditions) McCurry et al. (2010) 0 0% (0/1)

Fat free mass (lower) Dvorak and Poehlman (1998) + + 100% (1/1)

Body fat % (lower) Dvorak and Poehlman (1998),

James et al. (2012),

Watts et al. (2013)

0 0% (3/3)

Resting metabolic rate (higher) Dvorak and Poehlman (1998) + + 100% (1/1)

Energy intake (higher) Dvorak and Poehlman (1998) + + 100% (1/1)

Grip strength (higher) Auyeung et al. (2008) 0 0% (0/1)

Gait speed and walking capacity (faster) Auyeung et al. (2008),

Watts et al. (2013)

+ + 100% (2/2)

Global motor function (higher) James et al. (2012) + + 100% (1/1)

Dizziness Allan et al. (2006) � � 100% (1/1)

Lower limb strength Watts et al. (2013) 0 100% (1/1)

VO2 peak Watts et al. (2013) 0 100% (1/1)

Balance Watts et al. (2013) 0 100% (1/1)

Medications (>4) Allan et al. (2006) � � 100% (1/1)

Behavioral attributes/skills

ADL function (lower) Auyeung et al. (2008),

James et al. (2012),

Allan et al. (2006)

� Winchester et al. (2013), Cedervall et al. (2012) � 60% (3/5)

History of falls Allan et al. (2006) � � 100% (1/1)

Psychological, cognitive, emotional

Global cognition Winchester et al. (2013) + Dvorak and Poehlman (1998),

McCurry et al. (2011), Auyeung et al. (2008),

James et al. (2012), Vital et al. (2012),

David et al. (2012), Christofoletti et al. (2011),

Allan et al. (2006), Cedervall et al. (2012)

00 10%% (1/10)

HRQOL overall (higher)

Allan et al. (2006)

+ + 100 (1/1)

Physical HRQOL (higher) Auyeung et al. (2008),

Allan et al. (2006)

+ + 100% (2/2)

Mental HRQOL higher) Allan et al. (2006) + Auyeung et al. (2008) ? 50% (1/2)

Dementia severity (higher) Christofoletti et al. (2011) 0 0% (0/1)

NPI patient (lower scores) Christofoletti et al. (2011) + Allan et al. (2006) ? 50% (1/2)

NPI caregiver (lower scores) Allan et al. (2006), Christofoletti et al. (2011) 0 0% (0/2)

Sleep (better) Christofoletti et al. (2011) + McCurry et al. (2010), David et al. (2012) 0 33% (1/3)

Depression (higher) Allan et al. (2006) � McCurry et al. (2010), David et al. (2012),

James et al. (2012)

0 33% (1/3)

Waking hours in day (less) McCurry et al. (2010) � � 100% (1/1)

Revised memory and behavior

problem checklist – disruption scale

McCurry et al. (2010) 0 100% (1/1)

Autonomic symptom scale

(higher = more problems)

Allan et al. (2006) � � 100% (1/1)

Delirium Allan et al. (2006) � � 100% (1/1)

Apathy (lower) David et al. (2012) + + 100% (1/1)

Social and cultural

Social functioning Allan et al. (2006) + + 100% (1/1)

Physical environment

–

Policy factors

–

Key: Assoc = association; � = negative, + = positive; 0 = no relation; ? = indeterminate, ++, ��, 00 and ?? = four or more studies investigating a correlate.

CAMCOG, The Cambridge Examination for Mental Disorders of the Elderly (Cognitive Section); NPI patient, neuropsychiatric inventory patient; NPI caregiver,

neuropsychiatric inventory caregiver.

B. Stubbs et al. / Archives of Gerontology and Geriatrics 59 (2014) 195–203 199

Records iden�fied through database searc hing

(n = 4,327)

Scre

enin

g In

clud

ed

Elig

ibili

ty

Iden

�fica

�on

Addi�ona l re cords id en�fi ed through other source s

2 ke y authors provid ed det ails of pote n�al papers

Records a� er duplicate s removed and scr eened

(n = 2,603 )

Full -text ar�cles ass ess ed for eli gibili ty

(n = 118)

Full -text ar�cles exclud ed (n=106), with reasons : 59 – Did not meas ure phys ical ac�vity 12 – Coho rt studi es in ves�ga�ng risk of demen�a over �me 15 – No diagno sis of demen�a 8 – Ins �tu�onalised popu la�o n 3 – Overlap ping stud y 1 – No data on corr elate s 8 – con tacted and excluded with reasons

Studies in clu ded in quan�ta�ve synt hesis

(n = 12)

Fig. 1. PRISMA (2009) flow diagram for search strategy.

B. Stubbs et al. / Archives of Gerontology and Geriatrics 59 (2014) 195–203200

capacity is positively associated with participation in PA. There isinconsistent evidence that body mass index (BMI) is related to PAwith 40% (2/5 studies, Auyeung et al., 2008; Erickson et al., 2013)finding an increased BMI was negatively associated with PAparticipation. There is an indication from the findings of singlestudies that global motor function, lower fat free mass, restingmetabolic rate and higher energy intake are all positivelyassociated with PA participation. One study (Allan et al., 2006)found that dizziness was negatively associated with PA. Surpris-ingly, other biological correlates including lower limb strength,VO2 peak and balance were not related to PA although investiga-tion of these was limited to one study.

3.6. Behavioral attributes and skills

Two correlates were studied within this category and the mostcommonly investigated variable was how a participant’s difficultycompleting their ADL affected PA participation. This analysisestablished that three out of five studies (60%, Allan et al., 2006;Auyeung et al., 2008; James et al., 2012) found that difficultiesengaging in ADL were negatively associated with PA. There is someevidence from another study (Allan et al., 2006) that a history offalls is negatively associated with PA.

3.7. Psychological, cognitive and emotional correlates

In total 15 correlates were analyzed within this domain. Wefound consistent evidence from 10 studies that lower globalcognition is not related to PA participation with 90% (9/10, Allanet al., 2006; Auyeung et al., 2008; Cedervall et al., 2012;Christofoletti et al., 2011; David et al., 2012; Dvorak & Poehlman,1998; James et al., 2012; McCurry et al., 2010; Vital et al., 2012)

finding no significant association. There is some evidence from twostudies (Auyeung et al., 2008; Allan et al., 2006) that higherphysical health related quality of life (QOL) is positively associatedwith PA. There were inconsistent findings regarding the influenceof mental health related QOL while sleep and depression were notrelated to PA participation. There was some evidence frominvestigations limited to a single study that overall health relatedQOL and reduced apathy were positively associated with PA, whilewaking hours in the day and delirium were negatively associatedwith PA.

3.8. Social and cultural factors

Only one study (Allan et al., 2006) investigated a correlatewithin this domain and this study found that higher socialfunctioning was positively associated with PA participation.

3.9. Physical environment and policy factor correlates

Within this systematic review we did not identify any studythat investigated the influence of the physical environment orpolicy factors upon PA participation.

3.10. Discussion

To the best of our knowledge this review is the first to detail thequantitative correlates of PA participation in community dwellingadults with dementia. We were able to identify significantcorrelates in four of the five domains across the socio-ecologicalmodel. Increased dietary (energy) intake, resting metabolic rate,fat free mass, gait speed, global motor function, overall HRQOL,physical HRQOL, higher levels of social functioning and reduced

B. Stubbs et al. / Archives of Gerontology and Geriatrics 59 (2014) 195–203 201

apathy were all positively associated with PA. In addition, takingfour or more medications, dizziness, lower ADL function, a historyof falls, less waking hours in the day, more autonomic problemsand delirium were negatively associated with PA participation.However, most of these associations were only supported byfindings in one or two studies and we did not find any consistentlyreported (in four or more studies) statistically significantcorrelates.

It is of interest that we found that increasing age, education andglobal cognition were not consistently associated with lowerparticipation in PA. The finding from eight studies that increasingage is not negatively associated with PA participation suggests thatregardless of age community-dwelling adults with dementia canstill engage in PA and therefore potentially obtain the multiplebenefits associated with participation in it (Pitkala, Savikko, Poysti,Strandberg, & Laakkonen, 2013; Potter et al., 2011). Similarly, nineout of ten studies showed that lower global cognition is notassociated with lower levels of PA in community dwelling adultswith dementia. This is a somewhat surprising finding and apossible explanation could be that caregivers compensate withadditional assistance to engage the person with dementia to bemore active. A recent qualitative study of people with AD and theircaregivers gave an insight into this phenomenon and stressed theimportance of dual involvement of the caregiver in order for PAparticipation to be successfully increased (Malthouse & Fox, 2014).Clearly as the dementia advances the promotion of PA becomesmore complex and it is imperative that caregivers are involved inorder for PA uptake to be successful but this has also beendemonstrated to result in reductions in burden (Lowery et al.,2013). From this review, it appears that within dementia, otherfactors such as polypharmacy, difficulties in ADL and functionaldecline are more pertinent factors negatively associated with PAthan cognition.

It is also of interest that HRQOL and in particular physicalHRQOL are positively associated with PA but the results are limitedto one and two studies respectively. A RCT (Teri et al., 2003)including community dwelling adults with AD established that astructured PA program improved physical HRQOL in the interven-tion group compared to the control group. The group undertakingthe PA program also had lower levels of depression and improvedphysical function. The findings from our review are backed up bythis RCT that physical HRQOL is a key factor associated with PAparticipation in dementia. Somewhat surprisingly, we foundinconsistent results regarding the association between depressionand PA.

Perhaps the most interesting biological correlate we identifiedis that higher gait speed and walking capacity were positivelyassociated with PA (two studies). While this may not be surprising,it is important, since variations in gait speed and in particular slowgait are associated with an elevated risk of falls (Barak, Wagenaar,& Holt, 2006; Callisaya, Blizzard, Schmidt, McGinley, & Srikanth,2010) and it should be noted that falls are a travesty in communitydwelling adults with dementia (Wesson et al., 2013). We alsofound evidence from one study that a history of falls wasnegatively associated with PA participation (Allan et al., 2006).This highlights the challenge that clinicians face when older adultswith dementia start to develop slow walking speed, their falls riskincreases and they actually start falling. Within the general olderadult setting, targeted PA programs are the most effectiveinterventions to prevent falls (Sherrington et al., 2008) yet theirrole of PA programs in preventing falls in community dwellingadults with dementia is unclear and warrants exploration (Wessonet al., 2013).

Within the behavioral attributes domain of the socio-ecologicalmodel we found that older adults with dementia who experienceddifficulty undertaking their ADL were significantly less active. The

relationship between PA and ADL is well established in the generalolder population with a recent review establishing that increasingPA is effective in slowing the progression of ADL disability, frailtyand ultimately dependency (Tak, Kuiper, Chorus, & Hopman-Rock,2013). A recent Cochrane review established that PA can have apositive influence upon people with dementias ability to under-take their ADL (Forbes et al., 2013). While increasing difficultyparticipating in ADL is a key feature of the progression of dementia,it remains unclear if PA can prevent this progression. The currentreview demonstrates that difficulties with ADL and a history offalls were negatively associated with PA while a higher gait speedis positively associated with PA participation. While it is notpossible to make any claims regarding causation, current results dosuggest the importance of maintaining PA in older adults withdementia particularly when the wider well established benefits ofPA are considered.

Within the literature (Bilotta et al., 2012; Gray et al., 2013)there has been an increased discussion regarding the importanceof preventing the development of frailty. Frailty is related with anumber of factors that we found were associated with PAincluding weight loss (energy intake), exhaustion, muscleweakness, slow walking and physical inactivity (Kulmala,Nykanen, Manty, & Hartikainen, 2013). Preventing the develop-ment of frailty in community dwelling adults with dementia isvitally important since this is considered an independentpredictor of death in outpatients with AD (Bilotta et al., 2012).In addition frailty and dependence in ADL are associated withnursing home admission (Gaugler, Yu, Krichbaum, & Wyman,2009). PA is an important component to prevent sarcopenia andfrailty (Cooper et al., 2012) and so maintaining PA is clearlyimportant. Whether increasing PA levels may actually prevent thedevelopment of frailty in community dwelling adults withdementia and postpone nursing home admission is purelyspeculative at this stage but warrants exploration in prospectivelongitudinal studies.

3.11. Strengths and limitations

We conducted a comprehensive systematic review to identifythe multiple layers of factors over which PA participation isinfluenced. In order to reduce heterogeneity we only includedparticipants only with a confirmed diagnosis with dementiaaccording to recognized criteria and excluded those with mildcognitive impairment. In addition, nine research groups (out of 12)sent us their raw data and/or conducted additional analysis on ourbehalf, thus our results go considerably beyond what is available inthe published literature. However, it is important that the results ofthis review are interpreted with caution in view of a number oflimitations. First, all of the studies collected PA and theaccompanying correlates with a cross sectional measurement,thus it is not possible to identify causal relationships but only referto association. On the other hand, others (Bauman et al., 2012) haveargued that this is also a strength enabling potential mediators tobe established for planning interventions and to enable theprioritization of target population groups. Secondly, two-thirds ofthe studies within this review measured PA with self-reportoutcomes that are of unknown reliability and validity in dementia.Clearly there are concerns about the accuracy of the PAmeasurement obtained from self-report in this population, butwe did not find any statistical difference in the proportion ofcorrelates found in studies with studies using self-report orobjective measures. Third, dementia is a very heterogeneous groupand most within our review had a diagnosis of AD (89%), but onestudy (Allan et al., 2006) included participants with other knowntypes of dementia and it is unclear how this may have impacted theresults of our review. Fourth, as previous authors have noted

B. Stubbs et al. / Archives of Gerontology and Geriatrics 59 (2014) 195–203202

(Bauman et al., 2012; Vancampfort et al., 2014) this approach tostudying correlates only enables the consistency of associations tobe established and not the magnitude, thus meta-analysis is rarelypossible. Meta-analysis is also very difficult due to the heteroge-neity in the use of outcome measures, assessments tools andstatistical procedures. Finally, most of the included studies hadrelatively small sample sizes and included a limited number ofcorrelates.

3.12. Clinical overview and implications

This review offers relevant information for the stimulation of PAparticipation in clinical practice and in clinical trials withincommunity settings. It has highlighted that participation in PA iscomplex and affected by multiple factors. Future researchendeavors should explore in RCTs whether the identified variablesassociated with PA, in particular ADL, gait speed and dietary intakecould reduce the onset of frailty, progression of dementia,functional dependence, and ultimately nursing home admission(Kulmala et al., 2013). At this stage, this link is purely speculativeand theoretical but warrants investigation. We did not find anycorrelates within the domains of physical environment and policyfactors of the socio-ecological model. This is not uncommon sincethese factors have been understudied in other populations;however their importance on PA levels cannot be underestimated(Bauman et al., 2012). It may be interesting to determinedifferences in PA participation between urban and rural settingsor whether level of PA participation depends on coverage ofphysiotherapy programs by health care insurance. When con-ducting future correlate investigations of PA authors should seek touse appropriate measures to calculate PA and prioritize the use ofobjective measures.

It is known that carers and in particular family members arefaced with many challenges including difficulty in encouragingpeople with dementia to engage in PA. Future research shouldinvestigate the influence of carers upon participation as this islikely to be essential but this was not established in our review.However, there is promising research that engaging in PA canreduce carer burden (Orgeta and Miranda-Castillo, 2014). Futureresearch should therefore seek to determine the amount and typeof social support required to initiate or maintain PA in communitydwelling adults with dementia. Previous research (Teri et al., 2003)has established the importance of social support from caregivers inengaging people with dementia in PA and this is likely to beinfluential in its uptake. Other professionals such as physiothera-pists are likely to be able to provide advice on the appropriateforms of PA that can be matched to the individual. This may beparticularly useful when a person’s functional abilities decline,balance deteriorates and falls risk increases and it becomesimportant to encourage the participation in PA that is safe for theindividual.

4. Conclusion

There are indications that difficulty in undertaking ADL, gaitspeed, dietary intake and physical HRQOL are important factorsassociated with PA participation. From the review we were able todetermine that increasing age and lower global cognition do notappear to be negatively associated with participation in PA.However, we found no consistent correlates (reported in >4studies) of factors associated with PA in community dwellingadults with dementia. With the mounting political and healthinterest in PA in dementia, it is integral that future research isrequired to investigate these and other correlates further withobjective measurement of PA and prospective measurement beinga priority.

Conflict of interest

All authors report that they do not have any conflicts of interestto declare.

Authors contribution

All authors formulated the research question and designed thestudy. BS, LH and DV undertook the searches and decided on thefinal list of included articles. BS and LH undertook data extraction.BS, LH and DV analyzed the data and all authors contributed to theinterpretation of the data analysis and writing of the manuscript.All authors have approved the final version of the manuscript.

Role of funding source

Not applicable, the project was self-funded.

Acknowledgments

We would like to express our gratitude to the authors whokindly provided us with additional data including Dr Louise Allen,Newcastle University, Dr Jason Leung, Chinese University HongKong, Dr Ylva Cedervall, Uppsala University, Dr Renaud David, NiceUniversity, Dr Kirk Erickson, University of Pittsburgh, Dr Bryan D.James, Rush Alzheimer’s Disease Center, Dr Sue McCurry,University of Washington, Dr Thays Martins Vital, Cognitiva paraPacientes com Doenca de Alzheimer (PRO-CDA), Dr Amber Watts,University of Kansas.

References

Ahlskog, J. E., Geda, Y. E., Graff-Radford, N. R., & Petersen, R. C. (2011). Physical exerciseas a preventive or disease-modifying treatment of dementia and brain aging. MayoClinic Proceedings, 86(September (9)), 876–884. http://dx.doi.org/10.4065/mcp.2011.0252

Allan, L., McKeith, I., Ballard, C., & Kenny, R. A. (2006). The prevalence of autonomicsymptoms in dementia and their association with physical activity, activities ofdaily living and quality of life. Dementia Geriatric and Cognitive Disorders, 22(3),230–237.

American Psychiatric Association (1994). Diagnostic and statistical manual of mentaldisorders (DSM-IV) (4th ed.).

Auyeung, T. W., Kwok, T., Lee, J., Leung, P. C., Leung, J., & Woo, J. (2008). Functional declinein cognitive impairment – the relationship between physical and cognitive function.Neuroepidemiology, 31(3), 167–173.

Barak, Y., Wagenaar, R. C., & Holt, K. G. (2006). Gait characteristics of elderly peoplewith a history of falls: A dynamic approach. Physical Therapy, 86(November (11)),1501–1510.

Bauman, A. E., Reis, R. S., Sallis, J. F., Wells, J. C., Loos, R. J., Martin, B. W., et al. (2012).Correlates of physical activity: Why are some people physically active and othersnot? Lancet, 380(July (9838)), 258–271.

Bilotta, C., Bergamaschini, L., Nicolini, P., Case, A., Pina, G., Rossi, S. V., et al. (2012).Frailty syndrome diagnosed according to the Study of Osteoporotic Fracturescriteria and mortality in older outpatients suffering from Alzheimer’s disease: Aone-year prospective cohort study. Aging and Mental Health, 16(3), 273–280.

Blankevoort, C. G., van Heuvelen, M. J., Boersma, F., Luning, H., de Jong, J., & Scherder, E.J. (2010). Review of effects of physical activity on strength, balance, mobility andADL performance in elderly subjects with dementia. Dementia and Geriatric Cogni-tive Disorders, 30(5), 392–402. http://dx.doi.org/10.1159/000321357

Butler, R., & Radhakrishnan, R. (2012). Dementia. Clinical Evidence, (September).pii:1001.

Callisaya, M. L., Blizzard, L., Schmidt, M. D., McGinley, J. L., & Srikanth, V. K. (2010).Ageing and gait variability – a population-based study of older people. Age andAgeing, 39(March (2)), 191–197.

Cedervall, Y., Kilander, L., & Aberg, A. C. (2012). Declining physical capacity butmaintained aerobic activity in early Alzheimer’s disease. American Journal ofAlzheimer’s Disease and Other Dementia, 27(May (3)), 180–187.

Christofoletti, G., Oliani, M. M., Bucken-Gobbi, L. T., Gobbi, S., Beinotti, F., & Stella, F.(2011). Physical activity attenuates neuropsychiatric disturbances and caregiverburden in patients with dementia. Clinics, 66(4), 613–618.

Cooper, C., Dere, W., Evans, W., Kanis, J. A., Rizzoli, R., Sayer, A. A., et al. (2012). Frailtyand sarcopenia: Definitions and outcome parameters. Osteoporosis International,23(July (7)), 1839–1848.

David, R., Mulin, E., Friedman, L., Le Duff, F., Cygankiewicz, E., Deschaux, O., et al. (2012).Decreased daytime motor activity associated with apathy in Alzheimer disease:

B. Stubbs et al. / Archives of Gerontology and Geriatrics 59 (2014) 195–203 203

An actigraphic study. American Journal of Geriatric Psychiatry, 20(September (9)),806–814.

Dvorak, R. V., & Poehlman, E. T. (1998). Appendicular skeletal muscle mass, physicalactivity, and cognitive status in patients with Alzheimer’s disease. Neurology,51(November (5)), 1386–1390.

Erickson, K. I., Barr, L. L., Weinstein, A. M., Banducci, S. E., Akl, S. L., Santo, N. M., et al.(2013). Measuring physical activity using accelerometry in a community samplewith dementia. Journal of American Geriatrics Society, 61(January (1)), 158–159.

Erickson, K. I., Weinstein, A. M., & Lopez, O. L. (2012). Physical activity, brain plasticity,and Alzheimer’s disease. Archives of Medical Research, 43(November (8)), 615–621.http://dx.doi.org/10.1016/j.arcmed.2012.09.008

Forbes, D., Thiessen, E. J., Blake, C. M., Forbes, S. C., & Forbes, S. (2013). Exerciseprograms for people with dementia. Cochrane Database Systematic Reviews, (De-cember (12)), CD006489.

Gaugler, J. E., Yu, F., Krichbaum, K., & Wyman, J. F. (2009). Predictors of nursing homeadmission for persons with dementia. Medical Care, 47(February (2)), 191–198.

Gray, S. L., Anderson, M. L., Hubbard, R. A., LaCroix, A., Crane, P. K., McCormick, W., et al.(2013). Frailty and incident dementia. Journal of Gerontology, Series A: BiologicalSciences, 68(September (9)), 1083–1090.

Hauer, K., Schwenk, M., Zieschang, T., Essig, M., Becker, C., & Oster, P. (2012). Physicaltraining improves motor performance in people with dementia: A randomizedcontrolled trial. Journal of American Geriatrics Society, 60(January (1)), 8–15. http://dx.doi.org/10.1111/j.1532-5415.2011.03778.x

James, B. D., Boyle, P. A., Bennett, D. A., & Buchman, A. S. (2012). Total daily activitymeasured with actigraphy and motor function in community-dwelling olderpersons with and without dementia. Alzheimer Disease and Associate Disorders,26(July–September (3)), 238–245.

Kulmala, J., Nykanen, I., Manty, M., & Hartikainen, S. (2013). Association between frailtyand dementia: A population-based study. Gerontology, (August). (Epub aheadof print).

Lowery, D., Cerga-Pashoja, A., Iliffe, S., Thune-Boyle, I., Griffin, M., Lee, J., et al. (2013).The effect of exercise on behavioural and psychological symptoms of dementia:The EVIDEM-E randomised controlled clinical trial. International Journal of GeriatricPsychiatry, (December) http://dx.doi.org/10.1002/gps.4062

Malthouse, R., & Fox, F. (2014). Exploring experiences of physical activity among peoplewith Alzheimer’s disease and their spouse carers: A qualitative study. Physiothera-py, 100(2), 169–175.

McCurry, S. M., Pike, K. C., Logsdon, R. G., Vitiello, M. V., Larson, E. B., & Teri, L. (2010).Predictors of short- and long-term adherence to a daily walking program inpersons with Alzheimer’s disease. American Journal of Alzheimer’s Disease and OtherDementia, 25(September (6)), 505–512.

McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., & Stadlan, E. M. (1984).Clinical diagnosis of Alzheimer’s disease: Report of the NINCDS-ADRDA WorkGroup under the auspices of Department of Health and Human Services Task Forceon Alzheimer’s Disease. Neurology, 34(July (7)), 939–944.

Orgeta, V., & Miranda-Castillo, C. (2014). Does physical activity reduce burden in carersof people with dementia? A literature review. International Journal of GeriatricPsychiatry. http://dx.doi.org/10.1002/gps.4060

Pitkala, K., Savikko, N., Poysti, M., Strandberg, T., & Laakkonen, M. L. (2013). Efficacy ofphysical exercise intervention on mobility and physical functioning in older peoplewith dementia: A systematic review. Experimental Gerontolology, 48(January (1)),85–93. http://dx.doi.org/10.1016/j.exger.2012.08.008

Prince, M., Bryce, R., Albanese, E., Wimo, A., Ribeiro, W., & Ferri, C. P. (2013). The globalprevalence of dementia: A systematic review and meta-analysis. Alzheimers andDementia, 9(January (1)), 63–75.e2. http://dx.doi.org/10.1016/j.jalz.2012.11.007

Potter, R., Ellard, D., Rees, K., & Thorogood, M. (2011). A systematic review of the effectsof physical activity on physical functioning, quality of life and depression in olderpeople with dementia. International Journal of Geriatric Psychiatry, 26(October (10)),1000–1011. http://dx.doi.org/10.1002/gps.2641

Sallis, J. F., Cervero, R. B., Adcher, W., Henderson, K. A., Kraft, M. K., & Kerr, J. (2006). Anecological approach to creating active living communities. Annual Review of PublicHealth, 27, 297–322.

Sherrington, C., Whitney, J. C., Lord, S. R., Herbert, R. D., Cumming, R. G., & Close, J. C.(2008). Effective exercise for the prevention of falls: A systematic review and meta-analysis. Journal of American Geriatrics Society, 56(December (12)), 2234–2243.

Smith, J. C., Nielson, K. A., Woodard, J. L., Seidenberg, M., & Rao, S. M. (2013). Physicalactivity and brain function in older adults at increased risk for Alzheimer’s disease.Brain Sciences, 3, 54–83. http://dx.doi.org/10.3390/brainsci301005

Tak, E., Kuiper, R., Chorus, A., & Hopman-Rock, M. (2013). Prevention of onset andprogression of basic ADL disability by physical activity in community dwellingolder adults: A meta-analysis. Ageing Research Reviews, 12(January (1)), 329–338.

Teri, L., Gibbons, L. E., McCurry, S. M., Logsdon, R. G., Buchner, D. M., Barlow, W. E., et al.(2003). Exercise plus behavioral management in patients with Alzheimer disease:A randomized controlled trial. Journal of American Medical Association, 290(October(15)), 2015–2022.

Vancampfort, D., Vanderlinden, J., Stubbs, B., Soundy, A., Pieters, G., Hert, M. D., et al.(2014). Physical activity correlates in persons with binge eating disorder: Asystematic review. European Eating Disorders Review, (September) http://dx.doi.org/10.1002/erv.2255

Vital, T. M., Hernandez, S. S., Stein, A. M., Garuffi, M., Corazza, D. I., de Andrade, L. P., et al.(2012). Depressive symptoms and level of physical activity in patients with Alzhei-mer’s disease. Geriatric and Gerontology International, 12(October (4)), 637–642.

Warren, J. M., Ekelund, U., Besson, H., Mezzani, A., Geladas, N., & Vanhees, L. (2010).Assessment of physical activity – a review of methodologies with reference toepidemiological research: A report of the exercise physiology section of theEuropean Association of Cardiovascular Prevention and Rehabilitation. EuropeanJournal of Cardiovascular Prevention and Rehabilitation, 17, 127–139.

Watts, A., Vidoni, E., Loskutova, N., Johnson, D. K., & Burns, J. (2013). Measuring physicalactivity in older adults with and without Alzheimer’s disease. Clinical Gerontologist,36(4), 1–20.

Wesson, J., Clemson, L., Brodaty, H., Lord, S., Taylor, M., Gitlin, L., et al. (2013). Afeasibility study and pilot randomised trial of a tailored prevention program to reducefalls in older people with mild dementia. BMC Geriatrics, 13(September (1)), 89.

Winchester, J., Dick, M. B., Gillen, D., Reed, B., Miller, B., Tinklenberg, J., et al. (2013).Walking stabilizes cognitive functioning in Alzheimer’s disease (AD) across oneyear. Archives of Gerontology and Geriatrics, 56(January–February (1)), 96–103.