Exercise for falls prevention in aged care: systematic review and trial endpoint meta-analyses

Abstract

Background

There is strong evidence that exercise reduces falls in older people living in the community, but its effectiveness in residential aged care is less clear. This systematic review examines the effectiveness of exercise for falls prevention in residential aged care, meta-analysing outcomes measured immediately after exercise or after post-intervention follow-up.

Methods

Systematic review and meta-analysis, including randomised controlled trials from a Cochrane review and additional trials, published to December 2022. Trials of exercise as a single intervention compared to usual care, reporting data suitable for meta-analysis of rate or risk of falls, were included. Meta-analyses were conducted according to Cochrane Collaboration methods and quality of evidence rated using the Grading of Recommendations Assessment, Development and Evaluation approach.

Results

12 trials from the Cochrane review plus 7 new trials were included. At the end of the intervention period, exercise probably reduces the number of falls (13 trials, rate ratio [RaR] = 0.68, 95% confidence interval [CI] = 0.49–0.95), but after post-intervention follow-up exercise had little or no effect (8 trials, RaR = 1.01, 95% CI = 0.80–1.28). The effect on the risk of falling was similar (end of intervention risk ratio (RR) = 0.84, 95% CI = 0.72–0.98, 12 trials; post-intervention follow-up RR = 1.05, 95% CI = 0.92–1.20, 8 trials). There were no significant subgroup differences according to cognitive impairment.

Conclusions

Exercise is recommended as a fall prevention strategy for older people living in aged care who are willing and able to participate (moderate certainty evidence), but exercise has little or no lasting effect on falls after the end of a programme (high certainty evidence).

Key Points

• Exercise programmes can reduce falls in older people living in residential aged care.

• Exercise has little or no lasting effect on falls after the end of a programme so it should be incorporated into routine ongoing care.

• Exercises should include a combination of exercise types and be tailored, considering individual abilities and preferences.

Background

World Health Organisation guidelines recommend all older adults undertake physical activity for falls prevention and a range of other health benefits, including reduced mortality, improved cardiovascular health, mental and cognitive health and sleep [1]. In community-dwelling older adults, there is high certainty evidence that exercise reduces falls and that balance and functional exercises plus resistance training, and possibly Tai Chi, successfully prevent falls [2]. The consequences of falls have a significant impact on the quality of life and health of older people and contribute major costs to the health system, including for falls occurring in residential care [3].

However, the most recent Cochrane Collaboration review update for falls prevention in residential aged care settings concluded that the effectiveness of exercise as a single intervention to reduce the rate of falls is unclear, and there may be little or no effect on the risk of falling [4]. The high heterogeneity between trial findings was not explained by predefined subgroup analyses on provision of exercise type or level of care. Nevertheless, some of the trials reduced falls and newer trials have been published since [5–8]. Other recent meta-analyses have had both similar and conflicting findings, thus the value of exercise for falls prevention in residential aged care remains uncertain [9–12].

Exercise needs to be maintained for ongoing effects on muscle strength and fitness [13, 14]. In both highly trained athletes, and older adults discharged from hospital, deconditioning occurs when training ceases [15–17]. However, previous meta-analyses examining the effectiveness of providing exercise training programs for older people living in residential aged care settings have not examined falls outcomes according to the timing of recording outcomes in relation to delivery of the programme [4, 10, 18].

The current study therefore used systematic review and meta-analysis to address the research question ‘What is the effectiveness of exercise as a single falls prevention intervention to reduce the rate or risk of falls in older people living in residential aged care

• as measured at the end of an intervention period or

• following a period of post-intervention follow-up?’

Methods

Search strategy and selection criteria

A systematic review of randomised controlled trials (RCTs) of falls prevention exercise programs delivered in residential aged care settings was conducted. This work provides interim findings for a single intervention from an in-progress Cochrane review update of all intervention types in two settings. It includes, updates and re-examines eligible trials from the most recent Cochrane Collaboration review using the same analysis and synthesis methods; however, conference abstracts or trial records were not systematically searched or included [4, 19].

New trials published from 2017 to December 2022 were identified by searching CENTRAL, MEDLINE, Embase and CINAHL databases using the Cochrane Collaboration review search strategies [4]. Reference lists of included studies and relevant systematic reviews were examined for potentially eligible trials. Title and abstract screening and full-text review were conducted independently by two reviewers using Covidence systematic review software [20]. Disagreements were resolved by discussion or involvement of a third author. Eligible trials from the 2018 Cochrane review were also included [4].

Inclusion criteria were RCTs of exercise as a single intervention with a comparison to usual care, or inactive control that reported data suitable or obtained from authors to determine the rate of falls (number of falls per unit time), or risk of falling (number of participants falling one or more times) [4]. Included studies were conducted in care facilities with most participants aged over 65 (or mean age over 65). Excluded trials were conducted in residential settings that do not provide nursing care (e.g. retirement villages) or in community settings; only reported comparisons of alternate forms of exercise without a comparison to usual care (or a non-exercise group) or reported only a subset of falls (e.g. injurious falls). Included interventions were those designed to reduce falls, thus trials that recorded falls as a potential adverse event of the intervention were excluded.

Data extraction

Trial characteristics and outcomes data were extracted into a proforma word document or Covidence (variables reported in Appendix S2 are available in Age and Ageing online) [20]. Participant and trial characteristics were extracted by a single reviewer and checked for completeness and accuracy by a second reviewer. Outcomes data were extracted independently by two reviewers. Trial authors were contacted to clarify details as necessary [7, 8, 21, 22]. Falls outcomes data were extracted at the end of intervention delivery period, at the end of the longest post-intervention follow-up timepoint or both. Data for fall-related fractures were extracted as a secondary outcome. Original trial reports for eligible trials included in the Cochrane review were examined for additional falls endpoints independently by two reviewers [4]. Any additional data were extracted by a single reviewer and checked for accuracy by a second.

Risk of bias assessment

Two reviewers independently assessed methodological quality according to risk of bias items as outlined in the Cochrane Handbook [23] and Cameron et al. [4] using Covidence [20] (see Appendix S2, available in Age and Ageing online). Disagreements were resolved by discussion or involvement of a third reviewer as necessary.

Data analysis and synthesis of results

Outcomes data were pooled using generic inverse variance, random effects meta-analysis in Cochrane RevMan web (or fixed effects where study heterogeneity was negligible) for the ratio of the rate of falls, risk of falling or risk of fall-related fracture. Outcomes measured at the end of the intervention period, or after a period of post-intervention follow-up were pooled separately. Subgroup analyses were conducted according to (a) exercise type following the Prevention of Falls Network Europe (ProFaNE) taxonomy, with trials classified by primary type of exercise provided and (b) whether or not trials enrolled participants with cognitive impairment (part or whole of sample) [24]. Primary exercise type was agreed between two reviewers with expertise in physiotherapy and consultation with a third reviewer as necessary. Sensitivity analyses were conducted excluding any trials rated as having a high risk of bias for the items of sequence generation, allocation concealment, attrition bias, baseline imbalance or method of ascertaining falls. Where trials examined more than one type of exercise, data for the different exercise arms were combined to provide a single comparison to usual care for the main analyses, but arms were considered separately in the subgroup analysis by exercise type. We adjusted for clustering where this was not done in the published report using published intra-cluster correlation coefficients (falls rates 0.100, falls risk 0.071 and fracture risk 0.026) [4, 25]. Funnel plots were constructed to examine the potential presence of publication bias/small study effects. Outcomes data were calculated using excel or RevMan, entered by one reviewer and checked for accuracy of calculations and data entry by a second reviewer.

Confidence in conclusions was rated independently by two reviewers using the Grading of Recommendation Assessment, Development and Evaluation (GRADE) approach [26]. Conclusions were formulated according to standardised statements for communicating findings of systematic reviews of interventions [27].

Results

Search and study selection

Nineteen RCTs of exercise in comparison to usual care or an inactive control were included; these comprised 12 trials from the Cochrane Collaboration review plus seven newly identified trials [7, 8, 21, 22, 28–30] from screening 3,927 citations from the update search (see Appendix S1, available in Age and Ageing online: study selection flowchart and excluded studies) [4, 31]. Five trials reported falls data at both the end of the intervention period and after post-intervention follow-up.

Characteristics of included studies

The included studies were conducted in 13 different countries (Table 1). The participants ranged from 73 to 86 years of age [5, 7, 8]. Two trials were conducted specifically in participants with dementia [8, 21], one in residents with cognitive impairment [32] and four excluded participants with dementia [6, 29, 30, 33], while others applied exclusion criteria for severe dementia [34–36] (see Appendix S3, available in Age and Ageing online). Two trials only enrolled women [5, 6], the remainder included approximately three-quarters women, ranging from 47 to 89% [30, 37]. Five trials were inclusive in terms of baseline mobility and function [5, 7, 8, 38, 39], but the majority applied inclusion or exclusion criteria related to mobility and/or dependence and two studies enrolled only participants that had relatively good mobility and no dementia [30, 33] (see Appendix S3, available in Age and Ageing online). Thirteen studies reported the baseline falls risk of participants [6, 7, 21, 28–30, 32–34, 37–40]. Most studies were small (range 16–553 participants); five randomised 60 or fewer participants [5, 6, 8, 34, 37]. Six studies were cluster randomised [7, 21, 33, 34, 38, 40].

Table 1.

Summary characteristics of included trials of exercise for falls prevention for older people in residential aged care facilities^a

Study b Country Design	Number randomised Recruitment period	Description of intervention. Duration intervention, length post-intervention follow-up	Falls outcomes
Arrieta 2019 [22] Spain RCT	112 October 2016–July 2017	Intervention: supervised progressive group resistance training and balance exercise; 60 minutes/session, 2 sessions/week 6 months, 6 months follow-up	Rate of falls Risk of falling
Brett 2021 [8] Australia RCT	60 NR	Supervised group with variety of exercises to target strength, balance, endurance and flexibility. Intervention A: 45 minutes/session and 1 session/week, Intervention B: 15 minutes/session and 3 sessions/week 3 months, no follow-up	Rate of falls Risk of falling
Buckinx 2014 [39] Belgium RCT	62 NR	Individual supervised whole-body vibration performed in standing position, knees flexed, using vibration platform; 3 times/week; roughly 3 minutes/session 6 months, 6 months follow-up	Rate of falls Risk of falling
Choi 2005 [34] Korea Quasi Cluster RCT	59 NR	Supervised Tai Chi group exercise programs; 3 times/week; 35 minutes/session 3 months, no follow-up	Risk of falling
Dhargrave, 2020 [28] India RCT	163 NR	Partially individually supervised exercise at geriatric homes (initially supervised exercise programme for the first week and visit by therapists once every 15 days for 3 months) with education programme about awareness and prevention of falls. Recommended exercise: 7 times/week; 30 minutes/day plus walking 30 minutes/day 3 months, no follow-up	Rate of falls Risk of falling
Faber, 2006 [44] Netherlands RCT	278 March 2002–September 2003	Intervention A: Supervised group balance exercise derived from the principles of Tai Chi. Intervention B: Supervised group functional exercise programme including standing up from a chair, reaching and stepping, staircase walking and single-limb standing 5 months, 8 months follow-up	Rate of falls Risk of falling
Hewitt, 2018 [7] Australia Cluster RCT	221 (16 clusters) NR	Supervised individually prescribed, group progressive resistance and balance program; 2 sessions/week, 60 minutes/session for the first 6 months; 2 sessions/week, 30 minutes/session for another 6 months 12 months (6 months +6 months maintenance), no follow-up	Rate ratio Risk of falling Risk of fracture
Irez, 2011 [5] Turkey RCT	60 NR	Supervised group-based pilates exercise; 3 sessions/week, 60 minutes/session 12 weeks, no follow-up	Rate of falls
Jahanpeyma, 2021 [29] Turkey RCT	72 September 2016–June 2017	Individual Otago program and walking exercise (supervised in the first months and unsupervised for the second and third month); Otago exercise 3 sessions/week, 45 minutes/session plus ≥30 minutes/day walking for 3 times/week 3 months, no follow-up	Rate of falls
Kerse, 2008 [38] New Zealand Cluster RCT	682 February–November 2004	Intervention to promote independence in residential care including goal setting, functional assessment and activity programme design and implementation; implementation supervised by healthcare assistants or usual nursing staff; frequency and the duration of each session NR 6 months, 6 months follow-up	Rate of falls Risk of falling
Kovacs, 2013 [32] Hungary, RCT	86 February 2011–April 2012	Supervised group-based multimodal exercise based on Otago program; 2 sessions/week, 30 minutes/session 12 months, no follow-up	Rate of falls Risk of falling
Mulrow, 1994 [35] USA, RCT	194 NR	Individually supervised exercise session tailored to individual needs and included balance, gait, coordination, strength and flexibility exercise; 3 sessions/week, 30–45 minutes/session 4 months, no follow-up	Rate of falls Risk of falling
Rosendahl, 2008 [40] Sweden Cluster RCT	191 January–February 2002	Supervised high intensity functional group exercise programme including balance, functional and strength exercises; 2.5 sessions/week; 45 minutes/session 3 months, 6 months follow-up	Rate of falls Risk of falling
Sakamoto, 2006 [36] Japan RCT	553 NR	Individual unsupervised unipedal standing balance exercise; daily exercise, 6 minutes/day 6 months, no follow-up	Rate of falls Risk of falling
Schoenfelder, 2000 [37] USA RCT	16 NR	Supervised ankle strengthening and walking program; 3 sessions/week, 30 minutes/session (20 minutes for exercise and 10 minutes for walking) 3 months, 3 months follow-up	Rate of falls
Sihvonen, 2004 [6] Finland RCT	28 NR	Individualised dynamic balance exercise sessions on force platform, with visual feedback of the movement of centre of pressure; 3 sessions/week, 20–30 minutes/session 4 weeks, 11 months follow-up	Rate of falls Risk of falling
Toots, 2019 [21] Sweden Cluster RCT	186 NR	Supervised individual high-intensity functional exercise programme; 2.5 sessions/week. 45 minutes/session 4 months, 6 and 12 months follow-up	Rate of falls Risk of falling Risk of fracture
Varela, 2018 [30] Spain RCT	74 NR	Self-paced cycling; 7 sessions/week, ≥15 minutes/session 15 months, no follow-up	Rate of falls
Yokoi, 2015 [33] Japan Cluster RCT	105 NR	Supervised seated short stick exercise including throwing the stick while sitting; 2 sessions/week, 25 minutes/session 6 months, 6 months post-intervention follow-up	Risk of falling

NR, not reported; RCT, randomised controlled trial.

^aFull details of study characteristics are provided in Appendix S3, available in Age and Ageing online.

^bStudies identified by search updating Cameron et al. [4] are indicated by bold text.

There was wide variability in the length, intensity and type of exercise programs examined (Table 1). The length of the intervention period ranged from 4 weeks [6] to 15 months [30], with most intervention programs lasting 3–6 months and two for 12 months [7, 32]. The majority of programs provided a combination of exercise types, but some primarily focused on strength and resistance training [5], cycling [30], walking [37], short stick exercises [33], gait, balance and functional training [6, 29, 32, 36, 38], Tai Chi [34] or whole body vibration [39].

Risk of bias

The risk of bias of the included trials is provided in Appendix S4, available in Age and Ageing online. In general, blinding participants and personnel to provision of an exercise intervention are not feasible; however, one trial lowered the risk of bias by using a non-exercise activity control and not revealing the study hypothesis to the participants, relatives or staff [21]. Seven included trials had an imbalance of baseline characteristics that were not adequately accounted for in the analysis. Four trials were considered at high risk of bias with regard to the reported methods for ascertaining falls, due to recording falls weekly [34], relying on participant recording of falls collected monthly [5] or not providing a definition of falls [29, 35]. Randomisation and allocation concealment were considered at low or unclear risk of bias in 18 of 19 trials for each item. Three trials were considered at high risk of reporting bias and four at high risk of attrition bias due to high loss to follow-up or imbalance between study arms.

Effectiveness in preventing falls

Rate of falls

The pooled meta-analysis of 13 RCTs indicated that exercise significantly reduced falls in comparison to control, as measured at the end of the intervention period (rate ratio [RaR] = 0.68, 95% confidence interval [CI] 0.49–0.95; I² = 85%; Figure 1a). Substantial heterogeneity between trial outcomes remained which was explored further by subgroup analyses. There was no reduction in the rate of falls after a period of post-intervention follow-up (RaR = 1.01, 95%CI 0.80–1.28; I² = 58%; 8 trials; Figure 1b).

Figure 1.

Meta-analyses of rate of falls of exercise interventions compared to usual care in included randomised controlled trials in care facilities at (a) the end of the intervention period or (b) after a period of post-intervention follow-up

Risk of falling

The meta-analysis of 12 RCTs reporting risk of falling showed risk was reduced for participants in an exercise programme in comparison to control at the end of the intervention period (risk ratio [RR] 0.84, 95%CI 0.72–0.98; Figure 2a). However, there was no reduction in the risk of falling after a period of post-intervention follow-up (RR = 1.05, 95%CI 0.92–1.20; 8 trials; Figure 2b). Between study heterogeneity was relatively low at both endpoints (I² = 10–18%).

Figure 2.

Meta-analyses of the risk of falling of exercise interventions compared to usual care in included randomised controlled trials in care facilities at (a) the end of the intervention period or (b) after a period of post-intervention follow-up

Risk of fractures

Three trials reported the outcome of risk of fracture (Appendix S5 and Figure S1 are available in Age and Ageing online) [7, 21, 40]. The pooled estimate for the risk of any fracture for participants in an exercise programme, in comparison to control, was 0.87 at the end of the intervention period (95%CI 0.34–2.20; I² = 0%; 2 trials, 407 participants). After a period of post-intervention follow-up, the RR was 0.53 (95%CI 0.24–1.14; I² = 0%; 2 trials, 359 participants).

Subgroup analyses

Type of exercise programme

A subgroup analysis as measured at the end of the intervention period, by primary type of exercise programme, showed significant differences between subgroups for the rate of falls (P = 0.01) but not the risk of falling (P = 0.67; Figure 3). Within both outcome measures, falls were reduced in the subgroup of trials categorised as providing a combination of exercises, but the 95% confidence interval included the possibility of an increase in falls (RaR 0.62, 95%CI 0.36–1.07, I² = 91%; RR 0.84, 95%CI 0.68–1.03, I² = 42%). For the rate of falls, heterogeneity remained high for this subgroup. All other subgroups of exercise type included only one to three trials (Figure 3).

Figure 3.

Subgroup analyses by type of exercise (according to the ProFaNE taxonomy) of the (a) rate of falls and (b) risk of falling for exercise interventions to reduce falls compared to usual care in randomised controlled trials in care facilities as measured at the end of the intervention period.

Cognitive impairment

A subgroup analysis of whether trials enrolled participants with cognitive impairment or not did not find significant differences between subgroups for the rate of falls (P = 0.89; Appendix S5 and Figure S2a are available in Age and Ageing online) or risk of falling (P = 0.23; Appendix S5 and Figure S2b are available in Age and Ageing online) at the end of the intervention period. There was high between-study heterogeneity for the rate of falls in all subgroups (four trials enrolling participants with no cognitive impairment, I² = 84%; six trials of participants with mixed cognition, I² = 84%; three trials enrolling only participants with cognitive impairment, I² = 93%; Appendix S5 and Figure S2 are available in Age and Ageing online).

Adverse outcomes

Adverse events were generally poorly reported and often not systematically addressed [4]. Three trials reported no differences in adverse events between study arms, four reported no adverse events and two reported no serious adverse events (Appendix S5 and Table S3 are available in Age and Ageing online). Two trials reported deaths for which association with exercise could not be definitively ruled out; one due to a ruptured abdominal aortic aneurysm 1 week after the exercise trial [40] and the other 1 day after participating in a session from causes attributed to circulatory failure and general atherosclerosis [21].

Sensitivity analyses and funnel plot

Sensitivity analyses excluding trials at a high risk of bias showed pooled estimates of effect similar to that for the main analyses, but with wider confidence intervals due to the reduced number of participants contributing data (Table 2). The reduction in the risk of falling at the end of the intervention nevertheless remained statistically significant (RR 0.75, 95%CI 0.59–0.97, P = 0.03; Table 2).

Table 2.

Sensitivity analyses, excluding trials at high risk of bias in the domains of sequence generation, allocation concealment, attrition bias, baseline imbalance or method of ascertaining falls (see Supplementary Materials for forest plots)

Outcome measure	Outcome endpoint	No. of trials	Estimate of effect	95% CI	I 2
Rate of falls	End of intervention	4	0.55	0.29–1.07	89%
Rate of falls	Post-intervention follow-up	4	0.87	0.59–1.26	54%
Risk of falling	End of intervention	6	0.75	0.59–0.97	37%
Risk of falling	Post-intervention follow-up	5	1.02	0.81–1.27	34%

CI, confidence interval.

Funnel plots for the main analyses pooling the rate of falls did not indicate major asymmetry on visual inspection (Appendix S5 and Figure S4 are available in Age and Ageing online). However, asymmetry was observed for the risk of falling, indicating some concerns about publication bias or small study effects (Appendix S5 and Figure S5 are available in Age and Ageing online).

GRADE ratings of confidence in conclusions

The findings of the impact of exercise to prevent falls in residential aged care were considered of moderate to high confidence when assessed according to the GRADE criteria (Appendix S2 and Table S2 are available in Age and Ageing online). Exercise is recommended as a fall prevention strategy for older people living in aged care settings (moderate certainty evidence), but exercise has little or no lasting effect on falls after the end of a programme (high certainty evidence). We are uncertain of the impact exercise on fractures due to the small number of events captured within the trials (very low certainty evidence).

Discussion

This systematic review and meta-analysis of RCTs demonstrates that exercise interventions can effectively prevent falls in older people living in residential aged care facilities based on moderate certainty evidence. However, this effect is not maintained beyond cessation of the exercise programme (high certainty evidence), which is not surprising in this frail population who are unlikely to continue exercising when a specific interventional programme is withdrawn. This analysis differs to that in the most recent Cochrane Collaboration review, which did not report a significant effect of the intervention overall, by the addition of several trials published since that time [4, 7, 8, 21, 22, 28–30]. In addition, the current analysis focusses on whether the falls outcomes are measured at the end of the intervention period or not, capturing additional data by examining falls outcomes at different follow-up times. The Cochrane review pooled outcome data from the final endpoint of the included trials and therefore included a mix of outcomes from the end of the intervention period and after a period of post-intervention follow-up, an approach that would have contributed to the observed high heterogeneity and diluted any pooled measure of effectiveness.

Further research is required to inform what type of exercise programmes are most effective. Whilst the pre-defined subgroup analysis of exercise type (using ProFaNE criteria) demonstrated significant subgroup differences, the findings do not provide clear guidance (Figure 3) [24]. The analysis indicates that a combination of exercise types is likely to be effective at reducing falls; however, the residual heterogeneity and possibility of increased falls shows further research is required before definitive guidance is provided. In contrast, one recent meta-analysis concluded that more complex exercises do not reduce falls, contradicting this finding, but that review includes some trials not conducted in a care facility and some unit of analysis errors [12, 39, 41–46]. Another recent meta-analysis concluded that exercises including balance training reduced falls, but the review also included multifactorial trials with exercise as a component and one study not conducted in residential care [10, 47].

The trial features of the included studies vary widely including regarding the length of the intervention period, intensity of the exercise programme and types of participants enrolled. Many studies are also small; 47% randomised fewer than 100 participants (Table 1) and 26% were considered as unclear or high risk of attrition bias (see Appendix S4, available in Age and Ageing online). Many are therefore underpowered to detect a significant reduction in falls (particularly considering the highly fluctuating nature of falls rates) and are also of limited methodological quality (see Appendix S4, available in Age and Ageing online). The studies were also often poorly reported, in terms of the baseline characteristics of participants and detail of interventions provided, limiting the ability for further examination of these factors. Only 42% of trials were considered to have adequately measured and accounted for baseline characteristics of the participants and 37% were considered as low risk of bias for the method of ascertaining falls (see Appendix S4, available in Age and Ageing online). Thus, although further subgroup analyses of various trial features may seem appealing in an attempt to determine the drivers of success, in the absence of a distinct theory or pre-defined subgroup analyses, and considering the poor reporting, any findings from such non-randomised comparisons may be misleading [48–50]. It is also possible that the interactions between the types of exercise delivered, the characteristics of the participants, the setting and trial intervention features are too complex to be adequately captured within unidimensional subgroup analyses, and require more sophisticated analyses [51, 52].

Consideration must be made about the generalisability of the findings from this meta-analysis to the broader residential aged care population. Many trials applied specific inclusion criteria, often in relation to baseline mobility and function (such as only including those who could walk independently) and most did not include residents with severe cognitive impairment or who were bedridden (see Appendix S3, available in Age and Ageing online). Several trials only included residents with intact cognition [6, 29, 30, 33]. Many did not provide a complete description of the baseline characteristics of the participants in terms of cognitive status, dependency, falls risk and medical status (see Appendix S3, available in Age and Ageing online). It is likely that residents require a minimum level of cognitive ability and mobility to be able to participate in exercise programs. Also, given the multimorbidity, reduced cognitive status and frailty of many residents, professional supervision and consideration of the suitability of any exercise programme to individual abilities and preferences will be important, with a particular focus on exercising safely for this high-risk group.

Limitations of this review include that trial registry records and conference abstracts have not been searched. Excluding trials that record falls as adverse events (e.g. Rolland et al, 2007) provides a smaller set of trials than including all trials of exercise, but ensures consistency of the trialists’ intent and that the outcomes data for all falls are adequately captured [53]. This analysis includes only 5 of 19 trials reporting outcomes both at the end of the intervention and after a follow-up period, so future studies should measure falls outcomes at the end of the intervention and extend beyond the period of implementation.

Trials appear to be trending towards becoming more effective over time, which is unsurprising given advances in understanding of exercise physiology and falls prevention approaches in the community (Figures 1a and 2a). However, there remains a need for further high-quality trials in residential aged care to better understand effective interventions in this setting. Future trials should be carefully planned, powered and reported, following CONSORT (Consolidated Standards of Reporting Trials) and TIDieR (Template for Intervention Description and Replication) guidelines, with detailed information on the exercise programme provided, implementation and characteristics of the enrolled participants [49, 50].

Conclusions

Exercise is recommended as a fall prevention strategy for older people living in aged care facilities who have sufficient physical and mental capacity and agree to participate (moderate certainty evidence). Continuation of exercise is required for effective fall prevention, as exercise has little or no lasting effect on falls after the end of a programme (high certainty evidence). Additional analyses and research are needed to determine the most effective types of exercise programs to be provided and how to best deliver them to people in this setting. Nevertheless, this meta-analysis provides justification for the provision and continuation of exercise programs to prevent falls in older people living in residential aged care settings; however, programs should consider individual abilities and preferences.

Declaration of Conflicts of Interest

Rik Dawson, Catherine Sherrington and Keith Hill are authors of trials of exercise in older people, which are funded by national grant agencies.

Declaration of Sources of Funding

The Centre of Research Excellence for Prevention of Falls Injuries funded by the Australian National Health and Medical Research Council provides salary support for authors Dyer and Suen. The funder played no role in the conduct or interpretation of the review.