JAMDA xxx (2015) 1e7 JAMDA journal homepage: www.jamda.com Review Article Frailty as a Predictor of Future Falls Among Community-Dwelling Older People: A Systematic Review and Meta-Analysis Gotaro Kojima MD * Japan Green Medical Centre, London, United Kingdom a b s t r a c t Keywords: Frailty falls community-dwelling older people Background: Although multiple longitudinal studies have investigated frailty as a predictor of future falls, the results were mixed Thus far, no systematic review or meta-analysis on this topic has been conducted Objective: To review the evidence of frailty as a predictor of future falls among community-dwelling older people Methods: Systematic review of literature and meta-analysis were performed using electronic databases (Embase, Scopus, MEDLINE, CINAHL Plus, PsycINFO, and the Cochrane Library) searching for studies that prospectively examined risk of future fall risk according to frailty among community-dwelling older people published from 2010 to April 2015 with no language restrictions Results: Of 2245 studies identified through the systematic review, 11 studies incorporating 68,723 individuals were included in the meta-analysis Among studies reporting odds ratios (ORs), frailty and prefrailty were significantly associated with higher risk of future falls (pooled OR ¼ 1.84, 95% confidence interval [95% CI] ¼ 1.43e2.38, P < 001; pooled OR ¼ 1.25, 95% CI ¼ 1.01e1.53, P ¼ 005, respectively) Among studies reporting hazard ratios (HRs), whereas frailty was significantly associated with higher risk of future falls (pooled HR ¼ 1.24, 95% CI ¼ 1.10e1.41, P < 001), future fall risk according to prefrailty did not reach statistical significance (pooled HR ¼ 1.14, 95% CI ¼ 0.95e1$36, P ¼ 15) High heterogeneity was noted among studies reporting ORs and seemed attributed to difference in gender proportion of cohorts according to subgroup and meta-regression analyses Conclusions: Frailty is demonstrated to be a significant predictor of future falls among communitydwelling older people despite various criteria used to define frailty The future fall risk according to frailty seemed to be higher in men than in women Ó 2015 AMDA e The Society for Post-Acute and Long-Term Care Medicine Older people are a highly heterogeneous population Although people generally develop diseases and disabilities as they age, the trajectory and rate of change in health and functional status vary widely in each individual and persons with the same chronological age can have very different biological ages.1 Therefore, it is challenging to measure the heterogeneity of the aging process in the elderly One of the potential concepts to quantify the overall health diversity of older people is frailty Frailty is a biological syndrome characterized by reduced reserve capacity in multiple physiologic systems and increased vulnerability to stressors due to age-related cumulative deficits.2 In general, people are more likely to develop The author declares no conflicts of interest * Address correspondence to Gotaro Kojima, MD, Japan Green Medical Centre, 10 Throgmorton Avenue, London EC2N 2DL, UK E-mail address: gotarokojima@yahoo.co.jp http://dx.doi.org/10.1016/j.jamda.2015.06.018 1525-8610/Ó 2015 AMDA e The Society for Post-Acute and Long-Term Care Medicine frailty as they get older.2,3 Prevalence of frailty among communitydwelling people aged 65 years and older is widely variable depending on settings, ranging from 4.0% to 59.1%.3 Frailty has been shown to be associated with multiple adverse health outcomes, including disability, falls, hospitalization, institutionalization, and death Among these, fall is a leading cause of mortality in older people.4 Fall is not only associated with a wide range of negative consequences, such as disabilities, fear of falling, and impaired quality of life,4,5 but also associated with increased health care burden and costs.6 Incidence of fall is high among older people; one-third of elderly aged 65 and older fall every year, and the incidence of falling increases up to 50% among those 80 years and older.7 Given the expanding elderly population worldwide, preventing falls has been a major public concern of authorities in many countries.4,8,9 One of the important key issues for preventing falls is identification of risk factors for falling Weakness, impaired balance, and abnormal gait are major components of physical frailty2,10 and are likely to increase the risk of G Kojima / JAMDA xxx (2015) 1e7 falling in older people Furthermore, frail older people may be at high risk of falling because of decreased functional reserve capacity in maintaining position, balance, and coordination, and increased vulnerability to such stressors as accidents, disease symptoms, or adverse drug reactions The evidence of frailty as a predictor of falls in community-dwelling older people comes from prospective cohort studies with mixed results Most of the studies demonstrated that the frail elderly were more likely to fall than the nonfrail,10e17 but a few showed nonsignificant results.18e20 Thus far, no systematic review or meta-analysis studies on this topic have been conducted in the literature Therefore, the objectives of this systematic review were (1) to identify and compare prospective cohort studies examining frailty as a predictor of future falls among community-dwelling older people, and (2) to combine those data to synthesize pooled risk estimates of frailty for future falls Methods This study was conducted according to a protocol developed with adherence to Meta-analysis of Observational Studies in Epidemiology (MOOSE)21 statements by a clinician researcher who was trained for internal medicine and geriatric medicine and is currently working as a general practitioner Data Sources and Search Strategy A systematic search of the literature was performed in April 2015 using Embase, Scopus, MEDLINE, CINAHL Plus, PsycINFO, and the Cochrane Library for studies written in any languages and published from 2000 through present The search terms used included (Accidental falls (Medical Subject Heading (MeSH))) OR (Falling (MeSH)) OR (Falls (MeSH)) OR (Fall*) AND (Frailty) using an explosion function if available PubMed and reference lists of relevant studies were also hand searched Study Selection Prospective cohort studies examining frailty as a risk factor for future falls were selected using the following inclusion criteria: Prospective study design Community-dwelling individuals Sample size at least 100 individuals Individuals aged 60 years or older or mean age of 70 years or older Frailty was defined by criteria originally designed to measure frailty and validated in population-based studies or its modified versions Adjusted or unadjusted odds ratio (OR), risk ratio (RR), or hazard ratio (HR) as a risk measure reported or able to be calculated from available data Studies were excluded if they substituted other measures, such as disability or walking speed, to define frailty or used selected samples with certain conditions or diseases If multiple studies used the same data or cohort, a study with the largest number of individuals was selected Data Extraction A standardized data collection tool was used to collect data from the eligible studies The data extracted included the following: first author, year of publication, location, sample size, proportion of male individuals, age, frailty criteria, outcome, follow-up period, frequency of fall monitoring, and effect measure When single fallers and recurrent fallers were used as separate outcomes and data of any fallers (single fallers þ recurrent fallers) were not available, calculation of an OR of any fallers compared with nonfallers was attempted, or the data of only recurrent fallers were used Some frailty criteria define “prefrail” or similar terminology, which is an intermediate frailty status between frail and nonfrail/robust, and these data were also collected and used for meta-analyses if available When or more frailty criteria were used in a study, the most commonly used Fried phenotype criteria or its modified versions were selected if available or criteria less modified from the original were selected Methodological Quality Assessment Eligible studies were further examined for methodological quality using the Newcastle-Ottawa scale for cohort studies This scale has criteria to examine the methodological quality of cohort studies Each of the included studies was assessed using this scale and considered to have adequate quality to be included for meta-analysis if it met or more items out of Statistical Analysis OR, RR, and HR along with 95% confidence interval (95% CI) of future fall risk for frailty or prefrailty compared with nonfrailty/ robust were extracted directly from the articles or calculated based on raw numbers shown in the articles All analyses were performed using StataIC 13 (Stata Corp, College Station, TX), Review Manager (Computer program, Version 5.2; The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark), and Comprehensive Meta-Analysis version 3.3 (Biostat, Englewood, NJ) OR, RR, and HR were log-transformed SEs of the log-transformed OR, RR, and HR were calculated by dividing the difference between log-transformed upper and lower limits of 95% CI by 3.92 These data of each study were entered into the Review Manager and Comprehensive Meta-Analysis to perform meta-analysis and meta-regression analysis The c2 test was used to assess heterogeneity across the studies, and heterogeneity was considered present when P value was less than 0.10 I2 statistic was used to quantify the degree of heterogeneity and I2 values of 25%, 50%, and 75% were considered as low, moderate, and high heterogeneity, respectively.22 When high heterogeneity was observed, subgroup analyses and random-effects meta-regression were performed to identify possible causes of heterogeneity Publication bias was assessed by visually inspecting the funnel plots Results Selection Processes Figure shows a flow chart of the literature search and study selection with numbers of studies at each stage Of 2245 citations identified by the systematic review of the literature using electronic databases, 1306 duplicated articles were excluded and 920 articles were excluded through review of titles and abstracts One additional article18 was found by manual search and added, leaving 20 articles for full-text review Of these, articles were excluded because they were review articles (n ¼ 2), did not classify frailty and nonfrailty status (n ¼ 2),23,24 included nonecommunity-dwelling populations (n ¼ 2), and used the same cohort (n ¼ 1) Neither abstracts nor full texts were able to be obtained for studies Eleven articles were left and confirmed that they met the inclusion criteria.10e20 Eleven articles provided data for 68,723 community-dwelling older people and these were included in this systematic review These studies were then assessed for methodological quality using the Newcastle-Ottawa G Kojima / JAMDA xxx (2015) 1e7 Osteoporosis in Women, consisting of 48,636 women from multiple countries in Europe, North America, and Australia,16 and studies involved fewer than 1000 individuals.14,15,17e20 Three studies had allfemale cohorts,13,16,18 had an all-male cohort,12 and the rest had cohorts including 30.3% to 53.5% male individuals.10,11,14,15,17,19,20 The mean age of the included studies with available data ranged from 72.1 to 82.0 years old Original or modified Fried phenotype criteria were most commonly used by of the 11 studies.10,12e16,18,20 The other criteria used were Study of Osteoporotic Fractures frailty index,12,14,19 Longitudinal Aging Study Amsterdam frailty instrument,11 and Chinese-Canadian Study of Health and Aging Clinical Frailty Scale.17 Three studies used different kinds of frailty criteria.12,14,19 Recurrent falls was most frequently used as an outcome11e14,16,17,19,20 and first fall or any falls were used by studies.10,15,17,18 Follow-up periods ranged from year to years Although studies monitored falls every to 18 months,10e14,18 the other studies identified falls by asking individuals if they had or more falls at the end of the followup period.15e17,19,20 Cox proportional hazard models were used in studies10,11,14,18 presenting HRs Seven studies presented ORs calculated using logistic regression models or from raw numbers of  tables.12,13,15e17,19,20 No study used RR Frailty as a Predictor of Future Falls Fig Flow chart of study selection quality assessment scale for cohort studies All of the 11 studies met at least criteria of and were included for the meta-analyses (Table 1) Study Characteristics Characteristics of the 11 studies included in this study are summarized in Table 1.10e20 More than half of the included studies were from the United States,10,12e15,18 were from Europe,11,19,20 was from Taiwan,17 and included cohorts from multiple countries.16 The largest study used a cohort of the Global Longitudinal Study of Meta-analysis of studies presenting OR ORs from studies, including a total of 60,577 individuals, were combined to calculate a pooled OR and 95% CI using a random-effects model due to high heterogeneity Frailty was significantly associated with higher future fall risk (pooled OR ¼ 1.84, 95% CI ¼ 1.43e2.38, P < 001, c2 ¼ 26.41, df ¼ 6, I2 ¼ 77%) Prefrailty was examined by of these studies and was also found to be associated with significantly higher future fall risk (pooled OR ¼ 1.25, 95% CI ¼ 1.01e1.53, P ¼ 04, c2 ¼ 12.83, df ¼ 3, I2 ¼ 77%) (Figure 2A) Meta-analysis of studies presenting HR Four studies with 8146 individuals presented HRs for frailty, among which studies presented HRs for prefrailty Fixed-effects models were used to calculate pooled HR and 95% CI, as heterogeneity was low for frailty and prefrailty Although frailty was significantly associated with higher future fall risk (pooled HR ¼ 1.24, 95% CI ¼ 1.10e1.41, P < 001, c2 ¼ 5.11, df ¼ 3, I2 ¼ 41%), an association between prefrailty and fall risk did not reach a statistical significance Table Summary of Included Studies on Future Fall Risk Associated With Frailty Among Community-Dwelling Older People Author Year Location Fried et al10 Bandeen-Roche et al18 Ensrud et al13 Ensrud et al12 Kiely et al14 Samper-Ternent et al15 Forti et al19 de Vries et al11 Sheehan et al20 Wu et al17 Tom et al16 2001 2006 2007 2009 2009 2012 2012 2013 2013 2013 2013 Sample Size* Men, % Age, yy Frailty Criteria USA 5317 USA 560 USA 6543 USA 3118 USA 760 USA 847 Italy 741 Netherlands 1509 Ireland 521 Taiwan 653 USA, Europe, 48154 Australia 42.1 0 100 36.1 35.3 44.6 48.2 30.3 53.5 >65 >65 76.7 76.4 78.1 82.0 74.7 75.6 72.1 75.4 >55 Outcome Follow-up Frequency of Period, y Monitoring Phenotype First fall Modified phenotype First fall Modified phenotype Recurrent falls Modified phenotype SOF Recurrent falls Modified phenotype SOF Recurrent falls 1.5 Modified phenotype Any falls CSBA modified SOF Recurrent falls LASA Second fall Modified phenotype Recurrent falls CSHA-CFS Any falls Modified phenotype Recurrent falls Every mo Every 18 mo Every mo Every mo Every mo Once at follow-up Once at follow-up Every mo Once at follow-up Once at follow-up Once at follow-up Effect Qualityz Measure aHR aHR aOR aOR aHR OR OR aHR aOR OR aOR 7 8 7 aOR, adjusted OR; aHR, adjusted HR; CSHA-CFS, Chinese-Canadian Study of Health and Aging Clinical Frailty Scale; CSBA index, Conselice Study of Brain Aging index; LASA, Longitudinal Aging Study Amsterdam frailty instrument; Recurrent falls, or more falls; SOF, Study of Osteoporotic Fractures frailty index *Sample size of cohort actually used for frailty and fall analysis, or of entire cohort if not available y Mean age of analytic sample if available, otherwise mean age of entire sample or age criterion for inclusion z Number of methodological quality criteria met using the Newcastle-Ottawa scale for cohort studies (range: 0e9) 4 G Kojima / JAMDA xxx (2015) 1e7 Fig Forest plots presenting effect of frailty and prefrailty on future fall risk according to OR (A: studies, C: studies including women) and HR (B: studies) df, degrees of freedom; fixed, fixed-effects model; IV, inverse variance; Random, random-effects model (pooled HR ¼ 1.14, 95% CI ¼ 0.95e1.36, P ¼ 15, c2 ¼ 2.16, df ¼ 2, I2 ¼ 7%) (Figure 2B) Subgroup analysis and meta-regression analysis High heterogeneity was noted among the studies presenting ORs.12,13,15e17,19,20 Subgroup analyses were attempted based on age and gender, well-known factors associated with frailty,2,3 as well as other factors including location (United States versus others), sample size (n ! 1000 versus 0.5) and showed that a higher level of frailty was significantly associated with future fall risk (adjusted OR ¼ 1.54 per increase in frailty level, 95% CI ¼ 1.34e1.76).23 The findings of these studies using a different approach of the frailty Fig Bubble plot with fitted meta-regression line for the association between OR of future fall risk for frailty and proportion of male individuals 6 G Kojima / JAMDA xxx (2015) 1e7 Fig Funnel plots for future fall risk according to frailty (A: studies presenting ORs, B: studies presenting HRs) index further support the association between frailty and higher future fall risk shown in this meta-analysis Higher fall risk associated with frailty was observed in studies including more men in this systematic review Although women are reported as more likely to be frail3 and more likely to fall27 than men, mechanisms underlying this finding are not clear Gender disparity in frailty-associated fall risk could be related to differences in health conditions, physical components, lifestyle factors, behavioral patterns, or mixed Among these, gender difference in physical activity may explain the higher fall risk in frail men Compared with women, men are more physically active28 and therefore may be more likely to encounter situations in which frailty is influential to their maintaining balance, stability, and coordination In such situations, men’s relatively higher center of gravity and heavier weight may predispose them further to higher risk of falling associated with frailty Falling as a research outcome is difficult to investigate among older people because it mostly relies on self-report information and therefore its accuracy may be compromised by memory disorders, especially when a fall monitoring covers a long period of time Therefore, it is important to recognize how falls are identified Various methodologies were used across the studies to obtain fall information from participants Three studies interviewed participants for incident falls,15,19,20 and other methods included post card,12,13 calendar,11,14 and telephone.13,17 Three studies did not provide clear explanation of how falls were reported.10,16,18 Among the studies using original or modified Fried phenotype criteria, all but the original study modified or more of the original criteria components according to data availability or study designs However, it is not the case only with these included studies in this review but also with most other published studies.29 The modification of the original criteria can potentially result in biasing the study findings.28 Four studies examined any falls or incident first fall10,15,17,18 and studies examined recurrent falls or incident second falls.11e14,16,19,20 They are theoretically different outcomes although these outcomes were treated as the same fall outcome when metaanalyzed in this review because of the relatively small number of studies if stratified Especially results based on recurrent falls may be affected by nature and consequences of the first fall For example, a serious or injurious first fall may cause disabilities or fear of falling, which can lead to less mobility or physical activities, eventually limiting chances of second falls Pooled risk estimates among studies examining any or incident first fall and those examining recurrent or incident second falls were calculated in the metaanalysis; pooled OR ¼ 2.05 (95% CI ¼ 1.46e2.89, P < 0001, from studies15,17 using fixed-effects model) and pooled OR ¼ 1.77 (95% CI ¼ 1.28e2.43, P ¼ 0005, from studies12,13,16,19,20 using randomeffects model), respectively This study has some limitations First, a few of the included studies15,17,19 did not present adjusted OR but only crude ORs, which were incorporated into the meta-analyses However, because of possible confounding effects on frailty and falls, especially by age and gender as well as other factors, adjusted ORs would better describe the true associations Excluding these studies in the meta-analysis did not change pooled OR or I2 statistics substantially (pooled OR ¼ 1.80, 95% CI ¼ 1.27e2.57, P ¼ 001, df ¼ 3, I2 ¼ 88%) Second, publication bias favoring studies with positive results was suggested by visual inspection of funnel plots and therefore pooled estimates may be overestimated One of this study’s strengths is that it is the first systematic review and meta-analysis on associations between frailty and future fall risk among community-dwelling older people Another strength is its robust methodology, including extensive and reproducible systematic reviews using databases and assessments of methodological quality, publication bias, and heterogeneity across the included studies In addition, subgroup analyses and meta-regression analyses were performed to investigate possible causes of the heterogeneity Despite different frailty criteria and methodology used by the included studies, the meta-analyses consistently demonstrated frailty as a significant predictor of future falling In summary, this study provides the first evidence of the association between frailty and higher future fall risk among community-dwelling older people based on the comprehensive systematic review and meta-analyses Subgroup and meta-regression analyses suggest a gender disparity in future fall risk associated with frailty Given the detrimental effects of falls in older people, it is important for health care providers, especially geriatricians and those who treat elderly individuals, to recognize frailty as a risk factor for future falling Future research should be directed to whether treating or reversing frailty should prevent falling among frail elderly and also to investigation of mechanisms underlying the gender disparity in the future fall risk according to frailty, which may lead to further understanding frailty in relation to falls and developing more effective interventions for both frailty and falls References Mitnitski AB, Graham JE, Mogilner AJ, et al Frailty, fitness and late-life mortality in relation to chronological and biological age BMC Geriatr 2002;2:1 Clegg A, Young J, Iliffe S, et al Frailty in elderly people Lancet 2013;381: 752e762 Collard RM, Boter H, Schoevers RA, et al Prevalence of frailty in communitydwelling older persons: A systematic review J Am Geriatr Soc 2012;60: 1487e1492 G Kojima / JAMDA xxx (2015) 1e7 Centers for Disease Control and Prevention Falls among older adults: An overview 2013 Available at: http://www.cdc.gov/homeandrecreationalsafety/ falls/adultfalls.html Accessed April 29, 2015 Parry SW, Deary V, Finch T, et al The STRIDE (Strategies to Increase confidence, InDependence and Energy) study: Cognitive behavioural therapybased intervention to reduce fear of falling in older fallers living in the communitydstudy protocol for a randomised controlled trial Trials 2014; 15:210 Centers for Disease Control and Prevention Costs of falls among older adults 2015 Available at: http://www.cdc.gov/HomeandRecreationalSafety/Falls/ fallcost.html Accessed April 29, 2015 Tinetti ME, Speechley M, Ginter SF Risk factors for falls among elderly persons living in the community N Engl J Med 1988;319:1701e1707 National Institute for Health and Care Excellence (UK) NICE clinical guideline, falls: Assessment and prevention of falls in older people 2013 Available at: https://www.nice.org.uk/guidance/cg161 Accessed April 29, 2015 Public Health Agency of Canada (Canada) You CAN Prevent Falls! 2011 Available at: http://www.phac-aspc.gc.ca/seniors-aines/publications/public/ injury-blessure/prevent-eviter/index-eng.php Accessed April 29, 2015 10 Fried LP, Tangen CM, Walston J, et al Frailty in older adults: Evidence for a phenotype J Gerontol A Biol Sci Med Sci 2001;56:M146eM156 11 de Vries OJ, Peeters GM, Lips P, et al Does frailty predict increased risk of falls and fractures? A prospective population-based study Osteoporos Int 2013;24: 2397e2403 12 Ensrud KE, Ewing SK, Cawthon PM, et al A comparison of frailty indexes for the prediction of falls, disability, fractures, and mortality in older men J Am Geriatr Soc 2009;57:492e498 13 Ensrud KE, Ewing SK, Taylor BC, et al Frailty and risk of falls, fracture, and mortality in older women: The study of osteoporotic fractures J Gerontol A Biol Sci Med Sci 2007;62:744e751 14 Kiely DK, Cupples LA, Lipsitz LA Validation and comparison of two frailty indexes: The MOBILIZE Boston Study J Am Geriatr Soc 2009;57: 1532e1539 15 Samper-Ternent R, Karmarkar A, Graham J, et al Frailty as a predictor of falls in older Mexican Americans J Aging Health 2012;24:641e653 16 Tom SE, Adachi JD, Anderson FA Jr, et al Frailty and fracture, disability, and falls: A multiple country study from the global longitudinal study of osteoporosis in women J Am Geriatr Soc 2013;61:327e334 17 Wu TY, Chie WC, Yang RS, et al Risk factors for single and recurrent falls: A prospective study of falls in community dwelling seniors without cognitive impairment Prev Med 2013;57:511e517 18 Bandeen-Roche K, Xue QL, Ferrucci L, et al Phenotype of frailty: Characterization in the women’s health and aging studies J Gerontol A Biol Sci Med Sci 2006;61:262e266 19 Forti P, Rietti E, Pisacane N, et al A comparison of frailty indexes for prediction of adverse health outcomes in an elderly cohort Arch Gerontol Geriatr 2012; 54:16e20 20 Sheehan KJ, O’Connell MD, Cunningham C, et al The relationship between increased body mass index and frailty on falls in community dwelling older adults BMC Geriatr 2013;13:132 21 Stroup DF, Berlin JA, Morton SC, et al Meta-analysis of observational studies in epidemiology: A proposal for reporting Meta-analysis of observational studies in epidemiology (MOOSE) group JAMA 2000;283:2008e2012 22 Higgins JP, Thompson SG, Deeks JJ, et al Measuring inconsistency in metaanalyses BMJ 2003;327:557e560 23 Fang X, Shi J, Song X, et al Frailty in relation to the risk of falls, fractures, and mortality in older Chinese adults: Results from the Beijing Longitudinal Study of Aging J Nutr Health Aging 2012;16:903e907 24 Li G, Ioannidis G, Pickard L, et al Frailty index of deficit accumulation and falls: Data from the Global Longitudinal Study of Osteoporosis in Women (GLOW) Hamilton cohort BMC Musculoskelet Disord 2014;15:185 25 Stevens JA, Sogolow ED Gender differences for non-fatal unintentional fall related injuries among older adults Inj Prev 2005;11:115e119 26 Rockwood K, Andrew M, Mitnitski A A comparison of two approaches to measuring frailty in elderly people J Gerontol A Biol Sci Med Sci 2007;62: 738e743 27 Deandrea S, Lucenteforte E, Bravi F, et al Risk factors for falls in communitydwelling older people: A systematic review and meta-analysis Epidemiology 2010;21:658e668 28 Centers for Disease Control and Prevention Facts about physical activity 2014 Available at: http://www.cdc.gov/physicalactivity/data/facts.html Accessed April 29, 2015 29 Theou O, Cann L, Blodgett J, et al Modifications to the frailty phenotype criteria: Systematic review of the current literature and investigation of 262 frailty phenotypes in the Survey of Health, Ageing, and Retirement in Europe Ageing Res Rev 2015;21:78e94