Risk Factors for Falls in Community Stroke Survivors: A Systematic Review and Meta-Analysis

Risk Factors for Falls in Community Stroke Survivors: A Systematic Review and Meta-Analysis

Accepted Manuscript Risk factors for falls in community stroke survivors: A systematic review and metaanalysis Tianma Xu, MOT, Lindy Clemson, PhD, Kat...

3MB Sizes 0 Downloads 24 Views

Accepted Manuscript Risk factors for falls in community stroke survivors: A systematic review and metaanalysis Tianma Xu, MOT, Lindy Clemson, PhD, Kate O’Loughlin, PhD, Natasha A. Lannin, PhD, Catherine Dean, PhD, Gerald Koh, MD, PhD PII:

S0003-9993(17)30528-2

DOI:

10.1016/j.apmr.2017.06.032

Reference:

YAPMR 56975

To appear in:

ARCHIVES OF PHYSICAL MEDICINE AND REHABILITATION

Received Date: 26 March 2017 Revised Date:

19 June 2017

Accepted Date: 25 June 2017

Please cite this article as: Xu T, Clemson L, O’Loughlin K, Lannin NA, Dean C, Koh G, Risk factors for falls in community stroke survivors: A systematic review and meta-analysis, ARCHIVES OF PHYSICAL MEDICINE AND REHABILITATION (2017), doi: 10.1016/j.apmr.2017.06.032. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Running head: Fall risk factors after stroke

ACCEPTED MANUSCRIPT

Risk factors for falls in community stroke survivors: A systematic review and metaanalysis Tianma Xu, MOT,

a,b

a

a

Lindy Clemson, PhD, Kate O’Loughlin, PhD, Natasha A. Lannin,

c

d

e

RI PT

PhD, Catherine Dean, PhD, Gerald Koh, MD, PhD a

From the Ageing Work and Health Research Group, Faculty of Health Sciences, University of Sydney, Sydney, Australia;

b

Health & Social Sciences Cluster, Singapore Institute of Technology, Singapore;

c

SC

Department of Community and Clinical Allied Health, La Trobe Clinical School La Trobe University, d

e

M AN U

Melbourne, Australia; Department of Health Professions, Faculty of Medicine and Health Sciences, Macquarie University, Sydney, Australia; Saw Swee Hock School of Public Health, National University of Singapore, Singapore;

Presented with preliminary results to the National Occupational Therapy Conference, 10

TE D

October 2015, Singapore; and full findings to the 7th Biennial Australian and New Zealand Falls Prevention Conference, 29 November 2016, Melbourne, Australia.

EP

Acknowledgement

AC C

Tianma Xu is supported by Singapore Institute of Technology for his Ph.D. study at the University of Sydney. We also would like to thank Professor Deborah Black from the University of Sydney for her statistical support in the meta-analysis.

Conflicts of interest We have no conflict of interest.

Corresponding author:

Running head: Fall risk factors after stroke

ACCEPTED MANUSCRIPT

Tianma Xu (MOT) Ageing Work & Health Research Unit Faculty of Health Sciences

75 East St, Lidcombe, NSW, 2141, Australia

RI PT

The University of Sydney,

Telephone: +61290367483 or +6565928673

AC C

EP

TE D

M AN U

Systematic Review Registration No.: CRD42015023389

SC

Email address: [email protected] or [email protected]

ACCEPTED MANUSCRIPT Risk factors for falls in community stroke survivors: A systematic review and metaanalysis

Abstract

RI PT

Objective: To identify the risk factors for falls in community stroke survivors.

Data Sources: A comprehensive search for articles indexed on MEDLINE, EMBASE,

CINAHL, PsychINFO, Cochrane Library, and Web of Science databases was conducted.

SC

Study Selection: Prospective studies investigating fall risk factors in community stroke survivors were included. Reviewers in pair independently screened the articles and

M AN U

determined inclusion through consensus. Studies meeting acceptable quality rating using the Q-Coh were included in the meta-analysis.

Data Extraction: Data extraction was done in duplicate by four reviewers using a standardized data extraction sheet, and confirmed by another independent reviewer for

TE D

completeness and accuracy.

Data Synthesis: Twenty-one articles met the minimum criteria for inclusion; risk factors investigated by three or more studies (n=16) were included in a meta-analysis. The following

EP

risk factors had strong association with all fallers: impaired mobility (OR 4.36, CI 2.68-7.10);

AC C

reduced balance (OR 3.87, CI 2.39-6.26); use of sedative / psychotropic medications (OR 3.19, CI 1.36-7.48); disability in self-care (OR 2.30, CI 1.51-3.49); depression (OR 2.11, CI 1.18-3.75); cognitive impairment (OR 1.75, CI 1.02-2.99); and history of fall (OR 1.67, CI 1.03-2.72). A history of falling (OR 4.19, CI 2.05-7.01) had a stronger association with recurrent fallers. Conclusions: This study confirms that balance and mobility problems, assisted self-care, taking sedative or psychotropic medications, cognitive impairment, depression, and history of falling are associated with falls in community stroke survivors. We recommend that any

1

ACCEPTED MANUSCRIPT future research into falls prevention programs should consider addressing these modifiable risk factors. As the risk factors for falls in community stroke survivors are multifactorial,

Keywords: Community; Stroke; Accidental falls; Risk factors; Rehabilitation

SC

List of Abbreviations:

RI PT

interventions should be multi-dimensional.

activities of daily living

OR

odds ratios

CI

confidence interval

RR

Relative Risk, Risk Ratio or Rate Ratio

MMSE

Mini-Mental State Examination

BBS

Berg Balance Scale

10MWT

10 metre walk test

TUG

timed up and go

FIM

Functional Independence Measure

AC C

EP

TE D

M AN U

ADL

2

ACCEPTED MANUSCRIPT Falls are one of the most frequent complications after stroke.1 Stroke survivors are more likely to

2

fall than the general aging population and this could lead to greater deficits in activities of daily

3

living (ADL) and mobility functions.2-4 The prevalence of falls among stroke survivors during the

4

first six months after discharge from hospital is between 36% and 73%,5-7 and fall rates remain

5

high between 40% and 58% among these individuals one year after stroke.8-11 More stroke

6

survivors report falls in the first months post-stroke after discharge from the hospital and almost

7

one out of three have a near-fall each month.12 Two-thirds of the stroke survivors fall indoors13-15

8

and ambulatory stroke survivors are more likely to fall compared to those with higher dependency

9

levels and reduced mobility.16 Among those who fall, walking and transfers are the most

SC

frequently mentioned activities at the time of fall.2

M AN U

10

RI PT

1

11

The consequences of falls can cause further physical complications among stroke survivors in the

13

community, such as soft tissue injuries,7 fractures,17 and restriction in functional activities.17 Once

14

stroke survivors sustain a hip fracture, they are less likely to regain independent mobility than the

15

general population.18 At the same time, stroke survivors often develop psychosocial issues from

16

falling. The most common are depression19 and fear of falling,20 which can further reduce their

17

level of activity leading to further physical deconditioning and loss of independence.21

EP

AC C

18

TE D

12

19

Understanding the risk factors for falls among community stroke survivors is crucial to

20

developing an effective community falls prevention program. A number of cohort and case-

21

control studies have reported that the risk of falls among community stroke survivors may

22

increase with reduced balance,6 fear of falling,22, 23 depressive symptoms,24 a fall history during

23

hospitalization,6, 7 motor and sensory impairment,23 and environmental safety hazards.24 However,

24

there has been no systematic comprehensive review on fall predictors among stroke survivors

25

after their discharge from hospital. A previous systematic review25 in 2009 published in the Dutch

26

language included stroke participants from both hospital and community settings. However, it had 3

ACCEPTED MANUSCRIPT 27

broader inclusion criteria and did not include meta-analysis, presenting the results as a descriptive

28

review. Furthermore, they did not distinguish between hospital and community settings.25 This

29

systematic review and meta-analysis aims to identify the risk factors for falls in stroke survivors

30

discharged from hospital and residing in the community, the setting where most falls occur.

RI PT

31

Methods

33

A systematic review protocol was developed before the review commenced. The detail of the

34

protocol was registered on the PROSPERO database under registration number

35

CRD42015023389. Review methods followed the Preferred Reporting Items for Systematic

36

Reviews and Meta-Analysis (PRISMA)26 and Meta-Analysis of Observational Studies in

37

Epidemiology (MOOSE)27 guidelines.

38

M AN U

SC

32

Search Strategy

40

A broad search strategy was used; combinations included Medical Subject Headings, text words

41

and word variants for three main themes: “stroke”, “falls”, and “study type”. As in similar

42

reviews,25, 28, 29 title and abstract were searched (up to June 2015) in the following electronic

43

databases: MEDLINE, EMBASE, PsycINFO, CINAHL, Cochrane Library (Cochrane Database of

44

Systematic Review) and Web of sciences. For example, under “study type”, we included the

45

following key words in our comprehensive search: prospective or cohort or case-control or

46

follow-up or observational or cohort study or case-control study or observational study. Full

47

search terms are available upon request. Additionally, bibliographies of identified publications

48

and published reviews were hand searched for potentially relevant articles. An email alert function

49

in electronic databases was created to keep track of any later published articles that may meet the

50

selection criteria based on saved search history up to 31 May 2016.

AC C

EP

TE D

39

51

4

ACCEPTED MANUSCRIPT Study Selection

53

We included prospective cohort studies and case-control studies where all participants were 18

54

years and above and clinically diagnosed with either first stroke or recurrent stroke. There was no

55

restriction on the year of publication but was limited to English literature. Studies were included if

56

at least 80% of the stroke participants in the study were being followed up in the community or

57

non-institutionalized setting for a minimum period of three months. Clinical trials and

58

retrospective studies were excluded. Studies conducted only during the hospitalization period or at

59

residential care facilities were excluded. The primary outcomes were the risk factors for all fallers

60

defined as community stroke survivors with at least one fall during follow-up. The secondary

61

outcomes were the risk factors for recurrent fallers defined as community stroke survivors with

62

two or more falls during follow-up.

63

M AN U

SC

RI PT

52

Data Extraction

65

One reviewer (TX) screened titles and based on inclusion criteria, rated them as relevant, not

66

relevant or maybe relevant. Abstracts for titles indicated with “yes” and “maybe” were

67

independently scrutinized by a pair of reviewers (TX & LC). The selected articles were then

68

reviewed independently in full text by two reviewers (TX, LC, KO, CD or NL). Any disagreement

69

was discussed among at least three reviewers until consensus was reached. The following data

70

was abstracted in duplicate by four reviewers (TX, LC, CD, and KO) using a standardized data

71

extraction sheet. The data extraction sheet included year of publication, country of study, sample

72

size, mean age of the participants, mean time since stroke at baseline assessment (duration of

73

stroke), frequency of fall assessment, duration of follow-up, proportion of fallers, and risk factors

74

for falls. The binary outcome variables reporting odds ratio (OR) with 95% Confidence Intervals

75

(CI) for each risk factor were extracted. Results reported in non-ORs, such as Relative Risk, Risk

76

Ratio or Rate Ratio (RR), were also extracted. When raw data is available to calculate the OR in

77

another study,30 we calculated the OR with 95% confidence intervals using an online statistical

AC C

EP

TE D

64

5

ACCEPTED MANUSCRIPT OR calculator (Appendix A, Table 1).31 The different categories of frequency of fall assessment

79

used in our data extraction were based on the Deandera et al’s systematic review.28 All extracted

80

data on proportions of fallers and risk factors were categorized into all fallers and recurrent fallers.

81

All findings from the selected studies were collated based on the common characteristics of

82

participants, and the primary and secondary outcomes of this review. Another independent

83

reviewer (NL) confirmed all data entries and checked for completeness and accuracy.

RI PT

78

84

Quality Assessment

86

The evaluation of the methodological quality of primary studies is a key process in systematic

87

review and meta-analysis. Strengthening the Reporting of Observational Studies in Epidemiology

88

statement (STROBE)32 was used as a general guide for the reviewers to determine whether the

89

selected observational studies were well reported. The quality assessment tool for cohort studies

90

called “Q-Coh”33 has established its inter-rater reliability (kappa: 0.68 to 0.87) and acceptable

91

validity, and it classifies methodological quality as either good, satisfactory or low.33 Hence, it

92

was used to select studies with good or satisfactory methodological quality to be included in the

93

meta-analysis. The Q-Coh tool was adapted by creating a standardized approach to decisions

94

when reviewing components and determining rating scales relevant to falls. A list of confounders

95

considered important to be controlled in fall studies was also defined before conducting the

96

quality assessment and they were age, gender, falls before stroke, falls in the hospital, length of

97

stay in the hospital, rehabilitation before and during the study.34 All selected studies were assessed

98

for risk of bias using the adapted Q-Coh by two independent reviewers (TX, LC, KO, CD). All

99

disagreements were solved in a consensus meeting. Any studies with good or satisfactory

100

AC C

EP

TE D

M AN U

SC

85

methodological quality were included in the meta-analysis.

101 102

Statistical analysis

6

ACCEPTED MANUSCRIPT In view of the wide range of potential risk factors being investigated across studies, only binary

104

outcome variables with 95% CI investigated by at least three studies in a comparable manner were

105

included in the meta-analysis. Decisions regarding inclusion of studies that can be meaningfully

106

compared were based on the validity of the outcome measures for measuring the domain of

107

interest. When different statistical analysis (e.g. both univariate analysis and multivariate analysis)

108

was used for the same risk factor in the study, the unadjusted results from the univariate statistical

109

analysis were pooled in the meta-analysis. However, when only adjusted results or results from

110

multivariate statistical analysis were presented in the study, these were included in the meta-

111

analysis.

SC

RI PT

103

M AN U

112

Since falls are more common in the studied population, the OR will considerably overestimate the

114

relative risk, or in other words, the relative risk will underestimate the OR if we combine both in

115

the meta-analysis.35, 36 Hence, we analysed all outcome variables reported in ORs and non-ORs

116

separately using the Comprehensive Meta-Analysis software (version 2.0)37 to generate pooled

117

estimates of effect sizes for each risk factor. We used the random-effects model with 95% CI for

118

all analyses in consideration of the diverse effect sizes of the included studies. The random-effects

119

model allowed us to estimate the mean of a distribution of effects from the selected studies in

120

meta-analysis.38 The i²-test was used to test the level of heterogeneity among the included studies.

121

The i²-test has been recommended as one of the most preferred and reliable tests for

122

heterogeneity,39 which represents the degree of variation in estimated effects across studies. It has

123

proposed that low heterogeneity is if the i² value falls lower than 50%, moderate for 50-75%, and

124

high for above 75%.40 All binary outcome variables or the risk factors reported in ORs were

125

pooled together for analyses and the respective risk factors reported in non-ORs were analysed

126

separately to support the findings from pooled ORs.

AC C

EP

TE D

113

127 128

Results 7

ACCEPTED MANUSCRIPT Study selection

130

Figure 1 summarises the flow of searches, inclusions, and exclusions. It included one eligible

131

study41 that was found from the database email alert system. Twenty-one studies including 19

132

cohort studies and two case-control studies met the selection criteria and were assessed for risk of

133

bias using the adapted Q-Coh tool.33

134



RI PT

129

135

Methodological quality

137

The summary of quality assessment is presented in appendix A, table 2. A study was considered

138

as “good” quality if at least six domain items in Q-Coh were given positive ratings and

139

“acceptable” quality if four or five domain items were rated as positive. Anything below was

140

considered as “low” quality. Out of the 21 papers, seven papers4, 5, 8, 24, 41-43 were rated as “good”

141

quality and nine papers6, 10, 11, 16, 30, 44-47 were rated as “acceptable” quality. The remaining five

142

papers7, 12, 48-50 were rated as “low” quality and excluded in the meta-analysis.

M AN U

TE D

143

SC

136

Characteristics of included studies

145

Study characteristics of the 16 included studies are displayed in table 1. Studies were conducted

146

in the community settings across different continents (Asia Pacific, Europe, and North America)

147

with a mean cohort age of sixty-nine years old, mean sample size less than 300 (range: 3011 –

148

117445), and published between 200224 and 2015.41 The range of follow-up period was from four

149

months24 to two years.44, 45 Out of the 16 studies, 11 studies5, 11, 16, 24, 30, 41-43, 45-47 identified risk

150

factors for all fallers, three studies4, 6, 8 identified risk factors for recurrent fallers, and two

151

studies10, 44 reported risk factors for both all fallers and recurrent fallers. A total number of 4,160

152

stroke participants from 16 studies and 1,116 (27%) stroke participants were classified as fallers.

153

The proportion of all fallers across studies ranged from 23%24 to 55%,8 and recurrent fallers from

154

5%44 to 42%.8 Summary characteristics of the included studies are shown in appendix A, table 3.

AC C

EP

144

8

ACCEPTED MANUSCRIPT 155



156

Meta-analysis

158

The meta-analysis revealed evidence on the risk factors for falling in people with stroke after

159

discharge from hospital. Based on previous literature on falls in older adults28, 51 and stroke

160

population,52 we proposed a classification dividing risk factors into the following domains:

161

sociodemographic, sensorimotor, cognitive, psychosocial, medical, balance and mobility, and self-

162

care risk factors. A total of 14 risk factors for both all fallers and recurrent fallers were

163

investigated by at least three studies, and at least one risk factor was represented under each fall

164

risk category (table 2).

165

< Table 2>

M AN U

SC

RI PT

157

166

Many of the risk factors examined across studies have used different scales and this is likely to

168

impact on heterogeneity. For four risk factors, cognitive impairment, balance and mobility, and

169

disability in self-care, we made post-priori decisions to exclude studies given lack of validity for

170

the measures used. For instance, under the cognitive domain, we pooled all Mini-Mental State

171

Examination (MMSE)5, 10, 30 together in the analysis and excluded the single item - Functional

172

Independence Measure (FIM) (memory score). We pooled all balance measures using Berg

173

Balance Scale (BBS) in the analysis,30, 46, 47 whereas studies using non-validated or single

174

performance test, such as Fugl Meyer Assessment postural stability score,5 “unable to semitandem

175

stand”,10 and “inability to recover by stepping”41 were excluded. Similarly, the mobility status

176

assessed using the 10 metre walk test (10MWT)53 and timed up and go (TUG)54 test were pooled

177

together, and “Unable to walk for ≤0.25ms”10 was deselected for analysis. Under the disability in

178

self-care domain, the FIM score5 is a broader measure of disability compared to the other scales,

179

such as the Modified Rakin Scale,44 ADL difficulties (bath/shower),10 and Barthel Index scores,42,

AC C

EP

TE D

167

9

ACCEPTED MANUSCRIPT 180 181

46

which measure a range of discrete self-care activities and can be meaningfully grouped together

for meta-analysis.

182

Table 2 shows the overall pooled results in ORs for each risk factor for 2 outcomes: all fallers and

184

recurrent fallers. Additional studies are listed where results were not available as an OR. Pooled

185

results for studies that contributed to significant risk factors for all fallers are shown in figure 2

186

and for recurrent fallers in figure 3.

RI PT

183

SC

187

Risk factors for all fallers

189

The overall pooled results based on comparable risk factors identified the following risk factors

190

for all fallers (figure 2) in declining magnitude of OR: impaired mobility (OR 4.36, 2.68 to 7.10);

191

reduced balance (OR 3.87, 2.39 to 6.26); use of sedative / psychotropic medications (OR 3.19,

192

1.36 to 7.48); disability in self-care (OR 2.30, 1.51 to 3.49); depression (OR 2.11, 1.18 to 3.75);

193

cognitive impairment (OR 1.75, 1.02 to 2.99); and history of fall (OR 1.67, 1.03 to 2.72). In

194

addition to the pooled ORs, individual studies reported in non-ORs for history of fall (rate ratio

195

1.30), depression (relative risk 1.60), reduced balance (risk ratio 2.10) and impaired mobility (risk

196

ratio 3.50) further supporting the above findings (table 2).

197



TE D

EP

AC C

198

M AN U

188

199

Table 2 also shows that the pooled results for the following variables were shown relatively low

200

relationship to all fallers and statistically non-significant: age (OR 1.02); gender (OR 1.01);

201

duration of stroke (OR 1.11); visual impairment (OR 1.39); and multiple strokes (OR 1.39). The

202

pooled results for the following variables were statistically non-significant with moderate to high

203

level of heterogeneity: motor impairment (OR 1.75); urinary incontinence (OR 1.54). The forest

204

plots of the above risk factors that were statistically non-significant can be found in appendix B,

205

figure 1. 10

ACCEPTED MANUSCRIPT 206

Risk factors for recurrent fallers

208

Table 2 also shows the 4 risk factors for recurrent fallers which were investigated by three or

209

more studies. The pooled OR for history of fall (OR 4.19, 2.50 to 7.01) indicates that people with

210

stroke who had a history of fall post-stroke are strongly associated with risk of recurrent falling

211

(figure 3). It is difficult to draw conclusions for use of sedative and psychotropic medications,

212

reduced balance and motor impairment in relationship to recurrent falling because of the high

213

heterogeneity. Forest plots of risk factors for recurrent fallers with high levels of heterogeneity or

214

non-significance can be found in appendix B, figure 2-3.

215



M AN U

SC

RI PT

207

216

Discussion

218

This meta-analysis showed that impaired balance and mobility, the use of sedative/psychotropic

219

medications, and disability in self-care were strongly associated with falling among stroke

220

survivors living in the community, while depression, cognitive impairment, and history of fall

221

were moderately associated with falling among this population.

EP

222

TE D

217

Pooling results from a total of 16 studies reveal that there are common risk factors for falls

224

between community stroke survivors and community-dwelling older adults. Both our study

225

findings and another systematic review study on fall risk factors in community older adults28

226

support that history of fall and mobility problem are associated with falls. We found impaired

227

mobility and depression lead to a higher risk of falls in community stroke survivors (OR 4.36; OR

228

2.11) compared to community older adults (OR 2.1; OR 1.63).28 Our findings also highlight that

229

having a history of a fall including a near-fall in the home, community or hospital setting predicts

230

a higher risk of recurrent falling in the stroke group (OR 4.19) than found in community older

AC C

223

11

ACCEPTED MANUSCRIPT adults (OR: 3.46) as reported in the Deandera et al’s systematic review of fall risk factors in

232

community older adults.28 A history of falling was also highly associated with recurrent falls (OR

233

>4), again similar to the findings of the Deandera et al’s study.28 Given that falls are common in

234

this group, our study confirms that clinical attention to fall prevention is needed for any stroke

235

survivor with a history of near-fall8 or fall in the hospital or rehabilitation setting6 or a fall in the

236

past twelve months.10, 44 Clinicians should thus screen for falls as one component of their routine

237

clinical care for stroke patients.

RI PT

231

SC

238

While the risk of falls is associated with increasing age in community older adults28 our meta-

240

analysis shows no relationship between age, gender and all fallers among community stroke

241

survivors, despite participants in included studies generally being older with an average age of

242

sixty-nine. This lack of relationship may, however, be a factor of the design of included studies.

243

There were only a small number of studies that presented OR for both age and gender, as some

244

studies included age and/or gender as potential confounding factors and therefore did not

245

individually analyse them.10, 11, 24, 41, 46, 47 Furthermore, we were unable to include the variables of

246

age and gender in logistic regression analysis due to the different statistical methodologies used in

247

some selected studies. For instance, one study set the statistical power to 15% in order for the

248

variables to be selected for further analysis.8

TE D

EP

AC C

249

M AN U

239

250

Motor impairment, while clinically thought to increase risk of falls, was not found to be associated

251

with falling in our study. Motor impairment, a sensorimotor risk factor, can be understood as a

252

limitation of function in terms of muscle control, movement or mobility55 and can affect the

253

movement control of the affected side of the body. However, the lack of consistency in the use of

254

terminology that describes motor impairment after stroke55 and the variability in the use of motor

255

impairment measures across studies led to high heterogeneity and likely the statistically non-

256

significance. Thus, it is difficult to conclude whether motor impairment is a risk factor for falling 12

ACCEPTED MANUSCRIPT 257

and/or recurrent falls. The methodological limitations of the included studies thus make further

258

research necessary.

259

The results of our meta-analysis support that cognitive impairment is associated with falling in

261

community stroke survivors. This is worthy of further investigation, given the limitations inherent

262

with the use of different cognitive screening tools used in the included studies. Cognitive

263

impairment after stroke is common; almost one out of four stroke survivors have some form of

264

cognitive impairment in the first three months after a stroke.56 Additionally, stroke survivors with

265

cognitive impairment may experience reduced safety awareness and require more attention from

266

their family members or other caregivers. Future studies should use cognitive tests that are more

267

sensitive in identifying stroke survivors with cognitive impairment. For example, the Montreal

268

Cognitive Assessment57 is a validated screening tool in detecting people with mild cognitive

269

impairment and more sensitive in identifying cognitive abnormalities after stroke than MMSE.57,

271

SC

M AN U

58

TE D

270

RI PT

260

Depression too is common after stroke; almost one third of all stroke survivors experience

273

depressive symptoms at some time after the onset of stroke.19 The included studies used different

274

cut-off points for the same scale to define severe depression10 making interpretation of data

275

difficult. Nevertheless study findings do suggest screening stroke survivors for depressive

276

symptoms and being aware of the need for appropriate psychosocial interventions for stroke

277

survivors with depressive symptoms.

AC C

EP

272

278 279

Our analysis showed a three-fold increased risk of falls associated with the use of

280

sedative/psychotropic medications as a risk factor of falling among community stroke survivors.

281

The high level of heterogeneity (i²=94%) present when we calculated the effect size of fall risk for

282

recurrent fallers may be due to the differences in measures, such as types10, 44 and number of 13

ACCEPTED MANUSCRIPT medications.6 In view of the small number of studies included in the analysis, there is also a need

284

to explore the influence of different types of medications commonly prescribed post-stroke that

285

may cause giddiness, such as antiepileptic drugs which are often prescribed for seizure control59

286

and pain60 in the stroke population. Since post-stroke central pain is common,61 clinicians should

287

consider the possible adverse events (such as falls) when prescribing medications for pain

288

management post stroke. In addition, other non-pharmacological interventions for pain

289

management, such as neurostimulation therapy and cognitive behavioral therapy could be

290

explored as alternative options.61

SC

RI PT

283

291

Our meta-analysis shows that community stroke survivors with impaired balance and mobility

293

have a four-fold increased risk of falls. Using the standardized and validated functional screening

294

tools, such as BBS, 10MWT and TUG test can help identify stroke survivors at risk of falls.

295

Previous structured exercise-based interventions with stroke survivors targeting strength, balance

296

and mobility were found to have no significant effect on fall reduction,23,62 whereas a recent 12-

297

week simplified Tai Chi is the first study to show a better effect in reducing falls in community

298

stroke survivors than exercise-based intervention using post hoc test (x²=5.6; p=.06).63 However,

299

this result should be interpreted with cautions as no fall rate or fall risk analysis was conducted,

300

and no further follow-up was done after the intervention. Dean et al.62 demonstrated a differential

301

effect in a group based exercise program between the stroke participants who had slower mobility

302

compared to those who had a faster gait, however, these effects did not reach statistical

303

significance within the subgroups. In view of the neurological deficits presented in people after

304

stroke, such as muscle weakness, sensory impairment and spasticity, we suggest future research to

305

find out if stroke-specific exercises for improving strength, balance and mobility in this

306

population reduce the risk of falls.

AC C

EP

TE D

M AN U

292

307

14

ACCEPTED MANUSCRIPT Our analysis revealed that disability in self-care (OR>2) was a risk factor for falls in community

309

stroke survivors. Unlike most other risk factors for falls in community stroke survivors, disability

310

in self-care as a risk factor is unique to the stroke population. Many people after stroke experience

311

some form of difficulty in performing self-care activities. This may help identify any stroke

312

survivors at a high risk of falls and likely to benefit from self-care retraining as part of a fall

313

prevention program.

RI PT

308

314

Identification of major risk factors is key to better understanding how falls may be prevented

316

among stroke survivors. Several falls risk assessment tools have been used in predicting falls in

317

the general population. For example, the Falls Risk for Older People – Community Setting

318

(FROP-Com)64 is one the comprehensive fall risk screening tool for general population. Our key

319

risk factors that we identified for the community stroke survivors are included in the FROP-Com

320

with the exception of depression. If using the FROP-Com for community stroke survivors, we

321

would recommend the addition of a depression screening tool. The Downtown index65 was

322

another risk screening tool purported to predict falls. It reported excellent sensitivity (91%) but

323

with very poor specificity (27%) when it was used to predict elderly stroke survivors at risk of

324

falls66 and would not be recommended.

M AN U

TE D

EP

325

SC

315

In the general population, the Cochrane review67 has recommended both single interventions (e.g.

327

exercises, home safety) and multifactorial interventions as being effective. It seems people with

328

stroke are at high risk of falling, and that the risk of falls is multifactorial. It is clear that we need

329

to address each of these risk factors, such as reduced balance and mobility, taking

330

sedative/psychotropic medications, disability in self-care, as well as cognitive and psychosocial

331

factors of depression and cognitive impairment. As the risk factors for falls in the general

332

population are very similar, it does suggest that we could use the falls prevention programs that

333

work for the general population. However, further research is needed to understand the underlying

AC C

326

15

ACCEPTED MANUSCRIPT reasons for risk specific to stroke. For example, which components of the balance and mobility or

335

which aspects of the self-care disabilities could lead to falls in people with stroke? Observational

336

studies and interviews with stroke survivors, caregivers and healthcare professionals could

337

elucidate how current general multifactorial programs could potentially be refined and translated

338

for stroke.

RI PT

334

339

In summary, the findings of this review have several implications. Firstly, since community stroke

341

survivors with reduced balance and mobility have four-fold increased risk of falls, falls prevention

342

programs should focus on improving balance and mobility using evidence-based interventions

343

(e.g. programs targeting muscle strength and balance deficits). Secondly, stroke survivors who are

344

taking sedative or psychotropic medications should take extra precautions especially when there

345

are other risk factors present at the same time. Thirdly, falls prevention programs tend to include

346

home safety,69 but do not specifically look at self-care activities or participation at home or in the

347

community. In order to address the disability in self-care after stroke, we suggest occupational

348

therapy interventions should include both home safety and setting of occupational goals to

349

maximize their level of independence in self-care activities and activity participation post-

350

stroke.68 More importantly, proper patient and caregiver education on the cognitive and

351

psychosocial risk factors for falls is highly recommended to prevent people with stroke from

352

falling, especially before discharge from the hospital settings. Lastly, early screening of these risk

353

factors upon discharge from the hospitals is crucial in identifying potential fallers.

M AN U

TE D

EP

AC C

354

SC

340

355

Study limitations

356

There were some limitations in this meta-analysis. First, as discussed above, the number of

357

observational studies reporting fall risk factors in community stroke survivors was small. Testing

358

for heterogeneity among the selected studies and excluding those with high heterogeneity further

359

reduced the number of potential studies that could be included in our meta-analysis. Second, 16

ACCEPTED MANUSCRIPT results were interpreted based on the 16 studies that we could pool and potential selection bias

361

may exist as we excluded non-English and abstract-only publications. Third, the lack of

362

standardization in fall data collection methods and statistical analysis methods are likely to affect

363

the homogeneity across all studies. Hence, only data from the studies with ‘good’ and ‘acceptable’

364

quality ratings were included in this review. Fourth, we could not pool all data from each paper in

365

view of such divergent outcome measures used. The calculation of pooled ORs for some of the

366

included risk factors did not include non-validated measures reporting OR and the data reporting

367

non-OR. In order to have at least three studies for each risk factor in the meta-analysis, we had to

368

pool the estimates derived from the selected studies with different levels of possibility of residual

369

confounding or bias, which may have affected the overall effect size computation and result

370

interpretation. This factor was carefully considered when we looked into the comparable risk

371

factors that could be meaningfully grouped together to minimize the level of heterogeneity. In

372

view of the above shortfalls, the study results should be interpreted with caution.

M AN U

SC

RI PT

360

TE D

373

Conclusions

375

Stroke survivors with balance and mobility problems, assisted self-care, taking sedative or

376

psychotropic medications, cognitive impairment, depression, and history of falling are at risk of

377

falls. All the above risk factors except history of falling are amenable through interventions. We

378

recommend that any future research into falls prevention programs for community stroke

379

survivors should consider addressing these modifiable risk factors. As the risk factors for falls in

380

community stroke survivors are multifactorial and few single interventions has been shown

381

effective in this population, interventions should be multi-dimensional.

AC C

EP

374

17

ACCEPTED MANUSCRIPT

References 1.

Tilson JK, Wu SS, Cen SY, Feng Q, Rose DR, Behrman AL, et al. Characterizing and identifying risk for falls in the LEAPS study: a randomized clinical trial of interventions to improve walking poststroke. Stroke 2012;43:446-52. Weerdesteyn V, de Niet M, van Duijnhoven HJ, Geurts AC. Falls in individuals with stroke. J

RI PT

2.

Rehabil Res Dev 2008;45:1195-213. 3.

Pouwels S, Lalmohamed A, Leufkens B, de Boer A, Cooper C, van Staa T, et al. Risk of

SC

hip/femur fracture after stroke: a population-based case-control study. Stroke 2009;40:32815.

Simpson LA, Miller WC, Eng JJ. Effect of stroke on fall rate, location and predictors: a

M AN U

4.

prospective comparison of older adults with and without stroke. PloS One 2011;6:e19431. 5.

Alemdaroglu E, Ucan H, Topcuoglu AM, Sivas F. In-hospital predictors of falls in community-dwelling individuals after stroke in the first 6 months after a baseline evaluation:

6.

TE D

a prospective cohort study. Arch Phys Med Rehabil 2012;93:2244-50. Mackintosh SF, Hill KD, Dodd KJ, Goldie PA, Culham EG. Balance score and a history of falls in hospital predict recurrent falls in the 6 months following stroke rehabilitation. Arch

Forster A, Young J. Incidence and consequences offalls due to stroke: a systematic inquiry.

AC C

7.

EP

Phys Med Rehabil 2006;87:1583-9.

BMJ 1995;311:83-6. 8.

Ashburn A, Hyndman D, Pickering R, Yardley L, Harris S. Predicting people with stroke at risk of falls. Age Ageing 2008;37:270-6.

9.

Sackley C, Brittle N, Patel S, Ellins J, Scott M, Wright C, et al. The prevalence of joint contractures, pressure sores, painful shoulder, other pain, falls, and depression in the year after a severely disabling stroke. Stroke 2008;39:3329-34.

18

ACCEPTED MANUSCRIPT 10. Lamb SE, Ferrucci L, Volapto S, Fried LP, Guralnik JM. Risk factors for falling in homedwelling older women with stroke the women’s health and aging study. Stroke 2003;34:494501. 11. Blennerhassett JM, Dite W, Ramage ER, Richmond ME. Changes in balance and walking

RI PT

from stroke rehabilitation to the community: a follow-up observational study. Arch Phys Med Rehabil 2012;93:1782-7.

12. Wagner LM, Phillips VL, Hunsaker AE, Forducey PG. Falls among community-residing

SC

stroke survivors following inpatient rehabilitation: a descriptive analysis of longitudinal data. BMC Geriatr 2009;9:46.

M AN U

13. Harris JE, Eng JJ, Marigold DS, Tokuno CD, Louis CL. Relationship of balance and mobility to fall incidence in people with chronic stroke. Phys Ther 2005;85:150-8. 14. Belgen B, Beninato M, Sullivan PE, Narielwalla K. The association of balance capacity and falls self-efficacy with history of falling in community-dwelling people with chronic stroke.

TE D

Arch Phys Med Rehabil 2006;87:554-61.

15. Soyuer F, Ozturk A. The effect of spasticity, sense and walking aids in falls of people after chronic stroke. Disabil Rehabil 2007;29:679-87.

EP

16. Yates JS, Lai SM, Duncan PW, Studenski S. Falls in community-dwelling stroke survivors:

AC C

an accumulated impairments model. J Rehabil Res Dev 2002;39:385-94. 17. Mackintosh SF, Hill KD, Dodd KJ, Goldie PA, Culham EG. Falls and injury prevention should be part of every stroke rehabilitation plan. Clin Rehabil 2005;19:441-51. 18. Ramnemark A, Nilsson M, Borssén B, Gustafson Y. Stroke, a major and increasing risk factor for femoral neck fracture. Stroke 2000;31:1572-7. 19. Hackett ML, Yapa C, Parag V, Anderson CS. Frequency of depression after stroke a systematic review of observational studies. Stroke 2005;36:1330-40. 20. Watanabe Y. Fear of falling among stroke survivors after discharge from inpatient rehabilitation. Int J Rehabil Res 2005;28:149-52. 19

ACCEPTED MANUSCRIPT 21. Herrmann N, Black S, Lawrence J, Szekely C, Szalai J. The Sunnybrook stroke study a prospective study of depressive symptoms and functional outcome. Stroke 1998;29:618-24. 22. Friedman SM, Munoz B, West SK, Rubin GS, Fried LP. Falls and fear of falling: which

prevention. J Am Geriatr Soc 2002;50:1329-35.

RI PT

comes first? A longitudinal prediction model suggests strategies for primary and secondary

23. Batchelor FA, Hill KD, Mackintosh SF, Said CM, Whitehead CH. Effects of a multifactorial falls prevention program for people with stroke returning home after rehabilitation: a

SC

randomized controlled trial. Arch Phys Med Rehabil 2012;93:1648-55.

24. Jorgensen L, Engstad T, Jacobsen BK. Higher incidence of falls in long-term stroke survivors

M AN U

than in population controls: depressive symptoms predict falls after stroke. Stroke 2002;33:542-7.

25. Rensink M, Schuurmans M, Lindeman E, Hafsteinsdottir TB. Falls: incidence and risk factors after stroke. A systematic literature review. Tijdschr Gerontol Geriatr 2009;40:156-67.

TE D

26. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Ann Inter Med 2009;151:264-9. 27. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, et al. Meta-analysis of

EP

observational studies in epidemiology: a proposal for reporting. JAMA 2000;283:2008-12.

AC C

28. Deandrea S, Lucenteforte E, Bravi F, Foschi R, La Vecchia C, Negri E. Risk factors for falls in community-dwelling older people: a systematic review and meta-analysis. Epidemiol 2010;21:658-68.

29. Verheyden GS, Weerdesteyn V, Pickering RM, Kunkel D, Lennon S, Geurts AC, et al. Interventions for preventing falls in people after stroke. Cochrane Database Syst Rev 2013;5:CD008728. 30. Andersson AG, Kamwendo K, Appelros P. Fear of falling in stroke patients: relationship with previous falls and functional characteristics. Int J Rehabil Res 2008;31:261-4.

20

ACCEPTED MANUSCRIPT 31. MedCalc Software. Odds Ratio Calculator. Available from: https://www.medcalc.org/calc/odds_ratio.php [Accessed 1st May 2016]. 32. Von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement:

RI PT

guidelines for reporting observational studies. Prev Med 2007;45:247-51.

33. Jarde A, Losilla J-M, Vives J, Rodrigo MF. Q-Coh: A tool to screen the methodological quality of cohort studies in systematic reviews and meta-analyses. Int J Clin Health Psychol

SC

2013;13:138-46.

34. Jarde A, Losilla J-M, Vives J. Methodological quality assessment tools of non-experimental

M AN U

studies: a systematic review. Ann Psychol 2012;28:617-28.

35. Akobeng A. Understanding systematic reviews and meta-analysis. Arch Dis Child 2005;90:845-8.

36. Grant RL. Converting an odds ratio to a range of plausible relative risks for better

TE D

communication of research findings. BMJ 2014;348:f7450.

37. Borenstein M, Rothstein D, Cohen J. Comprehensive meta-analysis: A computer program for research synthesis [Computer software]. Englewood, NJ: Biostat. 2005.

EP

38. Borenstein M, Hedges LV, Higgins J, Rothstein HR. A basic introduction to fixed‐effect and

AC C

random‐effects models for meta‐analysis. Res Synth Methods 2010;1:97-111. 39. Ried K. Interpreting and understanding meta-analysis graphs: a practical guide. 2006. 40. Harris R, Bradburn M, Deeks J, Harbord R, Altman D, Sterne J. Metan: fixed-and randomeffects meta-analysis. Stata J 2008;8:3. 41. Mansfield A, Wong J, McIlroy W, Biasin L, Brunton K, Bayley M, et al. Do measures of reactive balance control predict falls in people with stroke returning to the community? Physiotherapy 2015;101:373-80.

21

ACCEPTED MANUSCRIPT 42. Kerse N, Parag V, Feigin VL, McNaughton H, Hackett ML, Bennett DA, et al. Falls after stroke: results from the Auckland Regional Community Stroke (ARCOS) Study, 2002 to 2003. Stroke 2008;39S:1890-3. 43. Wada N, Sohmiya M, Shimizu T, Okamoto K, Shirakura K. Clinical Analysis of Risk Factors

RI PT

for Falls in Home-Living Stroke Patients Using Functional Evaluation Tools. Arch Phys Med Rehabil 2007;88:1601-5.

44. Callaly E, Chroinin DN, Hannon N, Sheehan O, Marnane M, Merwick A, et al. Falls and

SC

fractures 2 years after acute stroke: the North Dublin Population Stroke Study. Age Ageing 2015;44:882-6.

M AN U

45. Divani AA, Vazquez G, Barrett AM, Asadollahi M, Luft AR. Risk factors associated with injury attributable to falling among elderly population with history of stroke. Stroke 2009;40:3286-92.

46. Jalayondeja C, Sullivan PE, Pichaiyongwongdee S. Six‐month prospective study of fall risk

TE D

factors identification in patients post‐stroke. Geriatr Gerontol Int 2014;14:778-85. 47. Persson CU, Hansson P-O, Sunnerhagen KS. Clinical tests performed in acute stroke identify the risk of falling during the first year: postural stroke study in Gothenburg (POSTGOT). J

EP

Rehabil Med 2011;43:348-53.

AC C

48. Andersson AG, Kamwendo K, Seiger A, Appelros P. How to identify potential fallers in a stroke unit: validity indexes of 4 test methods. J Rehabil Med 2006;38:186-91. 49. Hyndman D, Ashburn A. Stops walking when talking as a predictor of falls in people with stroke living in the community. J Neurol Neurosurg Psychiatry 2004;75:994-7. 50. Said CM, Galea MP, Lythgo N. People with stroke who fail an obstacle crossing task have a higher incidence of falls and utilize different gait patterns compared with people who pass the task. Physical Therapy 2013;93:334-44. 51. Lord SR, Sherrington C, Menz HB, Close JC. Falls in older people: risk factors and strategies for prevention. Cambridge University Press; 2007. 22

ACCEPTED MANUSCRIPT 52. Hanger HC, Wills KL, Wilkinson T. Classification of falls in stroke rehabilitation–not all falls are the same. Clin Rehabil 2014;28:183-95. 53. Collen FM, Wade DT, Bradshaw CM. Mobility after stroke: reliability of measures of impairment and disability. Int Disabil Stud 1990;12:6-9.

elderly persons. J Am Geriatr Soc 1991;39:142-8.

RI PT

54. Podsiadlo D, Richardson S. The timed “Up & Go”: a test of basic functional mobility for frail

55. Langhorne P, Coupar F, Pollock A. Motor recovery after stroke: a systematic review. Lancet

SC

Neurology 2009;8:741-54.

56. Haring H-P. Cognitive impairment after stroke. Curr Opin Neurol 2002;15:79-84.

M AN U

57. Nasreddine ZS, Phillips NA, Bédirian V, Charbonneau S, Whitehead V, Collin I, et al. The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc 2005;53:695-9.

58. Pendlebury ST, Cuthbertson FC, Welch SJ, Mehta Z, Rothwell PM. Underestimation of

TE D

cognitive impairment by Mini-Mental State Examination versus the Montreal Cognitive Assessment in patients with transient ischemic attack and stroke. Stroke 2010;41:1290-3. 59. Huang YH, Chi NF, Kuan YC, Chan L, Hu CJ, Chiou HY, et al. Efficacy of phenytoin,

EP

valproic acid, carbamazepine and new antiepileptic drugs on control of late‐onset post‐stroke

AC C

epilepsy in Taiwan. Eur J Neurol 2015;22:1459-68. 60. Frese A, Husstedt I, Ringelstein E, Evers S. Pharmacologic treatment of central post-stroke pain. Clin J Pain 2006;22:252-60. 61. Klit H, Finnerup NB, Jensen TS. Central post-stroke pain: clinical characteristics, pathophysiology, and management. Lancet Neurology 2009;8:857-68. 62. Dean CM, Rissel C, Sherrington C, Sharkey M, Cumming RG, Lord SR, et al. Exercise to enhance mobility and prevent falls after stroke: the community stroke club randomized trial. Neurorehabil Neural Repair 2012;26:1046-57.

23

ACCEPTED MANUSCRIPT 63. Taylor-Piliae RE, Hoke TM, Hepworth JT, Latt LD, Najafi B, Coull BM. Effect of Tai Chi on physical function, fall rates and quality of life among older stroke survivors. Arch Phys Med Rehabil 2014;95:816-24. 64. Garvey J, Connolly D, Boland F, Smith SM. OPTIMAL, an occupational therapy led self-

randomized controlled trial. BMC Fam Pract 2015;16:1.

RI PT

management support programme for people with multimorbidity in primary care: a

65. Downton J. Falls in the elderly, London. Edward Arnold 1993;64:128-30.

rehabilitation. Stroke 1996;27:1821-4.

SC

66. Nyberg L, Gustafson Y. Using the Downton index to predict those prone to falls in stroke

M AN U

67. Gillespie LD, Robertson MC, Gillespie WJ, Sherrington C, Gates S, Clemson LM, Lamb SE. Interventions for preventing falls in older people living in the community. Cochrane Database Syst Rev 2012;9.

68. Russell MA, Hill KD, Blackberry I, Day LM, Dharmage SC. The reliability and predictive

TE D

accuracy of the falls risk for older people in the community assessment (FROP-Com) tool. Age Ageing 2008;37:634-9.

69. Clemson L, Mackenzie L, Ballinger C, Close JC, Cumming RG. Environmental interventions

EP

to prevent falls in community-dwelling older people: A meta-analysis of randomized trials. J

AC C

Aging Health 2008;20:954-71.

Supplier a.

Comprehensive Meta-Analysis software (version 2.0); Biostat, USA.

Figure Legends Figure 1: Flowchart of study selection

24

ACCEPTED MANUSCRIPT Figure 2: The overall pooled results for risk factors for all fallers which were statistically significant with low level of heterogeneity. Note: Under random effects, the smaller studies get more weight and the larger studies get less weight.

AC C

EP

TE D

M AN U

SC

significant with low level of heterogeneity.

RI PT

Figure 3: The overall pooled results for risk factor for recurrent fallers which were statistically

25

ACCEPTED MANUSCRIPT

Table 1: Summary characteristic of 16 prospective observational studies investigating fall risk factors in community stroke survivors Mean time since stroke at baseline assessment 4 months

Frequency of fall assessment* Intermediate

Duration of follow up (months) 6

Outcome (fallers) ˆ All

First/ Recurrent stroke Both

Proportion of all fallers % (n) 36.0% (24)

Proportion of recurrent fallers % (n) 15% (10)

Year 2012

Country of Study Turkey

Sample size 66

Male 32

Andersson et al

2008

Sweden

140

78

75

NA

Low

12

All

NA

41.4% (58)

NA

Ashburn et al

2008

UK

115

77

70.2

80 days

Low

12

Rec

Both

55.0% (63)

42% (48)

Blennerhassett et al

2012

Australia

30

20

66

>1 year

Low

14.5

All

NA

40.0% (12)

Callaly et al

2015

Ireland

522

262

70.6

3 days

Low

24

All/ Rec

Both

23.5% (124)

33% (10) 5% (27)

Divani et al

2009

USA

1174

550

74

3.5 years

Low

24

All

Both

46.0% (540)

NA

Jalayondeja et al

2014

Thailand

97

59

61.9

1 month

High

6

All

1st stroke

25.8% (25)

13% (13)

Jorgensen et al

2002

Norway

111

63

68

>10 years

High

4

All

Both

23.0% (25)

12% (13)

Kerse et al

2008

1104

546

70.7

1 day

Intermediate

6

All

Both

37.0% (407)

19% (211)

Lamb et al

2003

New Zealand UK

94

0

76

48 months

Intermediate

12

All/ Rec

Both

48.0% (45)

29% (27)

Mackintosh et al

2006

Australia

55

25

68.1

2.3 months

High

6

Rec

Both

45.0% (25)

22% (12)

Mansfield et al

2015

Canada

95

60

62.5

51.2 days

High

6

All

NA

36.8% (35)

15% (14)

Persson et al

2011

Norway

96

56

73

2 days

Intermediate

12

All

1st stroke

48.0% (46)

NA

Simpson et al

2011

Canada

80

58

67.6

NA

High

12

Rec

1st stroke

50.0% (40)

31% (26)

Wada et al

2007

Japan

101

62

67.2

6.1 year

High

12

All

NA

44.6% (45)

20% (20)

Yates et al

2002

USA

280

140

69.3

3-14 days

High

6

All

NA

50.7% (142)

NA

SC

M AN U

TE D

EP

AC C

ˆ Rec: Recurrent.

RI PT

Authors Alemdaroglu et al

Mean or median age (yrs) 64

*High: fall information was obtained at least every 3 months; intermediate: participants were interviewed less frequently than for “high” but at least every 6 months; low: when the participants were interviewed only once a year.

ACCEPTED MANUSCRIPT

Table 2: Meta-analysis results from 16 studies.

Risk factors for recurrent fallers Sociodemographic risk factors History of fall Sensorimotor risk factor Motor impairment Medical risk factors Use of sedative / psychotropic medications Mobility & self-care risk factors Reduced balance Note:

0.07 0.94 0.04

Mixed Mixed Mixed

5

1.75 (0.98-3.12)

81%

0.06

Mixed

3

1.75 (1.02-2.99)

0%

0.04

Mixed

3

2.11 (1.18-3.75)

59%

0.01

3 3 2 2

1.39 (0.86-2.25) 1.11 (0.90-1.35) 1.39 (0.79-2.46) 1.54 (0.59-4.04)

0% 74% 43% 44%

3

3.19 (1.36-7.48)

3 3 4

3.87 (2.39-6.26) 4.36 (2.68-7.10) 2.30 (1.51-3.49)

No. of studies reporting OR

OR (95% CI)

Statistical methods

Rate ratio Rate ratio Rate ratio

1 1 1

1.10 (1.01-1.20) 1.10 (0.93-1.30) 1.30 (1.13-1.50)

0% 0% 0%

0.03 0.26 <.01

Multivariate Multivariate Multivariate

Rate ratio Relative risk

1 1

1.10 (1.01-1.20) 0.80 (0.64-1.00)

0% 0%

0.03 0.05

Multivariate Multivariate

Rate ratio

1

1.10 (0.86-1.40)

0%

0.44

Multivariate

Mixed

Relative risk

1

1.60 (1.11-2.30)

0%

0.01

Multivariate

0.18 0.33 0.15 0.38

Mixed Mixed Multivariate Mixed

Rate ratio Rate ratio Rate ratio Rate ratio

1 1 1 1

0.97 (0.86-1.10) 1.10 (1.01-1.20) 0.91 (0.75-1.10) 1.30 (1.21-1.40)

0% 0% 0% 0%

0.64 0.03 0.33 <.01

Multivariate Multivariate Multivariate Multivariate

22%

0.01

Mixed

0% 0% 46%

0.01 <.01 <.01

Mixed Mixed Mixed

Risk Ratio Risk Ratio

1 1

2.10 (1.16-3.80) 3.50 (1.00-12.23)

0% 0%

0.01 0.05

Univariate Univariate

IRR (95% CI)

Heterogeneity i²

p

Statistical methods*

0.91 (0.85-0.98)

0%

0.01

Heterogeneity i²

p

Statistical methods*

3

0.88 (0.30-2.53)

72%

0.81

Mixed

3

2.23 (1.18-4.23)

94%

0.01

Mixed

3

2.40 (0.68-8.49)

81%

0.17

Mixed

4.19 (2.50-7.01)

13%

<.01

Mixed

i² value= 50-75%: moderate heterogeneity; i² value >75%: high heterogeneity *Mixed statistical methods include univariate and multivariate analyses.

RI PT

0% 6% 37%

SC

1.02 (1.00-1.03) 1.01 (0.76-1.34) 1.67 (1.03-2.72)

M AN U

4 3 3

4

Fall risk factors in non-Odds Ratio (non-OR) Heterogeneity No. of studies RR** (95% CI) p reporting non-OR i²

Statistical methods

TE D

Cognitive risk factor Cognitive impairment Psychosocial risk factor Depression Medical risk factors Visual impairment Duration of stroke Multiple strokes Urinary incontinence Use of sedative / psychotropic medications Mobility & self-care risk factors Reduced balance Impaired mobility Disability in self-care

p

EP

Sociodemographic risk factors Age Gender (female) History of fall Sensorimotor risk factor Motor impairment (lower Extremities)

Heterogeneity i²

AC C

Risk factors for all fallers

Fall risk factors in Odds Ratio (OR) No. of studies OR (95% CI) reporting OR

No. of studies reporting non-OR

Incidence rate ratio 1

RR** includes: Rate Ratio, Relative Risk or Risk Ratio; IRR: Incidence Rate Ratio

Multivariate

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

Appendix A

Table 1: Calculating Odds Ratio for Andersson et al.’s study

Andersson et al 2008 (n=140)

42

16

Age Female Visual impairment Cognitive impairment (MMSE<=23) Low mood (GDS>=6) Motor impairment (UE, BL<=56) Motor impairment (LE, BL<=35) Impaired functional mobility (TUG>14s) Impaired balance (BBS<45)

23 18 23 19 26 29 30 28 27

7 6 3 2 5 3 1 1 2

58 Yes 30 24 26 21 31 32 31 29 29

No 28 34 32 37 27 26 27 29 29

Odds

28

54

1.071 0.706 0.813 0.568 1.148 1.231 1.148 1.000 1.000

20 20 16 7 14 21 16 13 12

21 18 14 9 14 15 12 3 3

82 Yes 41 38 30 16 28 36 28 16 15

95% CI Lower limit

Upper limit p value

RI PT

Total

Non-faller Low SE high SE Total Odds

No 41 44 52 66 54 46 54 66 67

1.0000 0.864 0.577 0.242 0.519 0.783 0.519 0.242 0.224

OR

1.071 0.547 2.100 0.82 0.414 1.612 1.41 0.710 2.795 1.090 5.031 2.34 2.21 1.112 4.410 1.57 0.799 3.094 1.112 4.410 2.21 1.948 8.737 4.13 4.47 2.088 9.556

SC

Faller Low SE high SE Total

0.841 0.561 0.328 0.029 0.024 0.190 0.024 0.000 0.000

M AN U

OR and 95% CI were calculated in https://www.medcalc.org/calc/odds_ratio.php

Table 2: Q-Coh Quality Assessment for prospective observational studies SN

Representativeness

Ax Domain

Comparability of Maintenance of Exposure measure Outcome measure the groups the comparability

Statistical analyses

Attrition

Overall Rating

Quality

Studies 1 Alemdaroglu et al, 2012

3+

1

4+

1

4+

1

2+

1

8+2-

1

2+

1

1+

1

7

Good

2 Ashburn et al., 2008

2+1-

1

4+

1

4+

1

2+

1

10+

1

2+

1

1+

1

7

Good

3 Mansfield et al., 2015

2+1-

4 Wada et al., 2007

1+2-

5 Kerse et al., 2008

3+

6 Jogensen et al., 2002

2+1-

7 Simpson etal., 2011

2+1-

8 Andersson et al., 2008

2+1-

9 Blennerhassett et al., 2012 11 Divani et al., 2009 12 Jalayondeja et al., 2014

Items

Rating

Items

Rating

Items

TE D

Rating

Rating

Items

Rating

Items

Rating

Items

Rating

1

4+

1

4+

1

2+

1

10+

1

2+

1

1+

1

7

Good

0

4+

1

4+

1

2+

1

8+2-

1

2+

1

1+

1

6

Good

1

4+

1

4+

1

2+

1

7+3-

0

2+

1

1+

1

6

Good

1

4+

1

4+

1

2+

1

10+

1

2-

0

1+

1

6

Good

1

4+

1

4+

1

2+

1

10+

1

2-

0

1+

1

6

Good

1

4+

1

4+

1

2+

1

5+5-

0

1+1-

0

1+

1

5

Acceptable

2+1-

1

4+

1

4+

1

1+1-

0

6+4-

0

1+1-

0

1+

1

4

Acceptable

2+1-

1

4+

1

4+

1

2+

1

7+3-

0

1+1-

0

1+

1

5

Acceptable

2+1-

1

4+

1

4+

1

2+

1

7+3-

0

2-

0

1+

1

5

Acceptable

1+2-

EP

10 Callaly et al., 2015

Items

4+

1

4+

1

2+

1

10+

1

2-

0

1+

1

5

13 Lamb et al., 2003

3+

1

4+

1

4+

1

2+

1

7+3-

0

1+1-

0

1+

1

5

Acceptable

14 Mackintosh et al., 2006

1+2-

0

4+

1

4+

1

2+

1

10+

1

2-

0

1+

1

5

Acceptable

15 Persson et al., 2011

2+1-

1

4+

1

4+

1

2+

1

7+3-

0

2-

0

1+

1

5

Acceptable

16 Yates et al., 2002

1+2-

0

4+

1

4+

1

1+1-

0

7+3-

0

2+

1

1+

1

4

Acceptable

1 Andersson et al., 2006

3+

1

4+

1

4+

1

2-

0

5+5-

0

2-

0

1-

0

3

Low

2 Forster & Young, 1995

2+1-

1

3+1-

0

4+

1

1+1-

0

7+3-

0

2-

0

1+

1

3

Low

3 Hyndman & Ashburn, 2004

1+2-

0

3+1-

0

4+

1

1+1-

0

10+

1

1+1-

0

1-

0

2

Low

4 Said et al., 2013

1+2-

0

4+

1

4+

1

1+1-

0

6+4-

0

2-

0

1+

1

3

Low

5 Wagner et al., 2009

3-

0

3+1-

0

3+1-

0

2-

0

8+2-

1

2-

0

1-

0

1

Low

AC C

0

Acceptable

"+" indicates positive rating and "-" indicates negative rating for the assessed item in each category For rating under each Category, "1" indicates positive rating and "0" indicates "negative" rating.

1

ACCEPTED MANUSCRIPT

Table 3: Summary characteristics of the included studies

No. of Studies*

3 7 3

2 3 1

M AN U

3 9 1

2 3 1

SC

6 3 4

RI PT

All Fallers

EP

TE D

Characteristics Location Asia Pacific Europe North America No. of subjects included in the study <100 100-200 >200 Mean or median age of study poulation (years) <65 65-75 >75 Duration of follow-up (months) 3-6 7-12 >12 Year of publication 2000-2005 2006-2010 2011-2015 Frequency of fall assessment High Intermediate low

Recurrent Fallers

0 5 1

6 4 3

1 4 1

0 6 7

2 1 3

5 6 2

3 2 1

AC C

*The sum does not add up to the total because two papers presented both outcomes

2

ACCEPTED MANUSCRIPT

Appendix B

Risk factors (for all fallers) Author, year

Weight %

Odds ratio and 95% CI

OR (95% CI)

Age 12.53 0.07 77.89 9.51

1.02 (0.97 to 1.07) 1.07 (0.55 to 2.10) 1.01 (0.99 to 1.03) 1.06 (1.00 to 1.12) 1.02 (1.00 to 1.03); p=0.07 Heterogeneity i² = 0%;

RI PT

Alemdaroglu et al 2012 Andersson et al 2008 Callaly et al 2015 Kerse et al 2008

Gender (Female) 2.02 (0.72 to 5.63) 0.82 (0.42 to 1.62) 0.99 (0.75 to 1.31) 1.01 (0.76 to 1.34); p=0.94 Heterogeneity i² = 6%;

SC

Alemdaroglu et al 2012 Andersson et al 2008 Kerse et al 2008

Motor impairment 0.97 (0.91 to 1.04) 2.21 (1.11 to 4.40) 1.20 (0.51 to 2.83) 3.71 (1.67 to 8.25) 2.20 (1.04 to 4.66) 1.75 (0.98 to 3.12); p=0.06 Heterogeneity i² = 81%;

M AN U

Alemdaroglu et al 2012 Andersson et al 2008 Lamb et al 2003 Persson et al 2011 Yates et al 2002

Visual impairment

1.47 (0.54 to 4.04) 1.41 (0.71 to 2.80) 1.30 (0.53 to 3.19) 1.39 (0.86 to 2.25); p=0.18

26.98 19.56 16.99 17.86 18.62

22.54 48.94 28.51

TE D

Alemdaroglu et al 2012 Andersson et al 2008 Lamb et al 2003

7.39 16.36 76.26

Heterogeneity i² = 0%; Duration of stroke

Multiple strokes

Heterogeneity i² = 74%; 71.32 28.68

1.16 (0.83 to 1.62) 2.20 (0.91 to 5.32) 1.39 (0.79 to 2.46); p=0.25

AC C

Kerse et al 2008 Lamb et al 2003

55.16 3.84 41.01

1.00 (0.99 to 1.02) 1.10 (0.41 to 2,99) 1.26 (1.07 to 1.48) 1.11 (0.90 to 1.35); p=0.33

EP

Alemdaroglu et al 2012 Lamb et al 2003 Wada et al 2007

Heterogeneity i² = 43%;

Urinary incontinence

Alemdaroglu et al 2012 Lamb et al 2003

0.84 (0.25 to 2.84) 2.30 (1.00 to 5.30) 1.54 (0.59 to 4.04); p=0.38 Heterogeneity i² = 44%; 0.1 0.2

39.99 60.01 0.5

Decrease odds in falling

1

2

5

10

Increase odds in falling

Figure 1 The overall pooled results for risk factors for all fallers which were statistically non-significant.

3

ACCEPTED MANUSCRIPT Risk factors (for recurrent fallers)

Author, Year

OR (95% CI)

OR (Random) and 95% CI

Use of Sedative and Psychotropic Medications 3.84 (1.45-10.18) 2.80 (2.66-2.95) 1.30 (1.00-1.70) 2.23 (1.18-4.23); p=0.014 Heterogeneity i²=94%

20.99 40.79 38.22

0.1 0.2

0.5

Decrease odds of falling

1

2

RI PT

Callaly et al 2015 Lamb et al 2003 Mackintosh et al 2006

Weight %

5

10

Increase odds of falling

Figure 2 The overall pooled results for risk factors for recurrent fallers which were statistically

SC

significant with high level of heterogeneity.

EP

TE D

M AN U

Risk factors (for recurrent fallers)

AC C

Figure 3 The overall pooled results for risk factors for recurrent fallers which were statistically non-significant with high level of heterogeneity.

4