Background

Fall-related injuries are the largest cause of accidental death in older people across Europe [1] and the second leading cause of accidental death amongst older people globally [2]. Over 30% of community-dwelling people aged 65 and older and 50% of people aged 80 and over fall at least once per year [3, 4]. People living in retirement villages/assisted living facilities fall frequently [5]. Those identified as frail fall more frequently than those who are classified as vigorous [6, 7]. Falls are associated with admission to residential care homes, reduced functioning, psychological problems such as fear of falling and loss of confidence leading to social isolation and increased dependency. The direct and indirect costs of falls are substantial, for example, estimated costs in the UK NHS are in excess of £2.3 billion per year [8].

There is strong evidence that strength and balance-based exercises reduce falls by up to 42% [3, 9, 10] and that strength and balance exercise as a stand-alone intervention may be the most cost-effective approach to fall prevention at a population level [11]. Although there are few studies in assisted living facilities, a cluster randomised controlled trial of group strength and balance exercise in retirement villages in Australia showed a reduction of 22% falls in the intervention group compared to the control [5]. Exercise as a means of fall prevention and for the promotion of independence has been welcomed by older people, as a positive step that individuals can take for themselves [12]. Reviews of community-based fall prevention indicates that strength and balance exercise need to be tailored, progressive and of adequate dose (50 h) [3, 13, 14]. Sherrington et al. reveal that exercise programmes that are of a higher dose (more than 3 h/week) have larger effects [14]. Such training can be costly and inaccessible to older adults [15]. The repetitive nature of these exercises may also discourage older adults to exercise in the home setting, thereby rendering the intervention ineffective [16]. Uptake and adherence to exercise programmes is low, and appropriate levels and progression are often not adequately prescribed [5, 17]. Fear of falling can lead to restriction or avoidance of daily activities, loss of independence, depression and a reduction in quality of life [18]. In frailer older adults and for those living in institutional settings, there is also the risk that unsupervised exercise can increase risk of falls [16, 19]. A number of reviews of exercise to improve function in frailer older adults recommend supervision to ensure progression and effectiveness [20, 21]. Thus, there are compelling reasons to find interventions that are both effective and safe for this population.

Exergaming (active video games which combine gameplay with physical exercise and may also incorporate types of virtual reality simulations) may be a feasible tool for older people to improve exercise uptake, challenge and progression [22]. There is growing evidence that Exergaming may also improve function and adherence and provide other health outcomes [4, 23] and that such technology-based approaches can be attractive to older people [24]. The advantages of using gaming systems to deliver exercise are that they can be immersive, entertaining and enjoyable, potentially improving adherence and frequency/duration of the exercise programme [25, 26]. The gamified elements of Exergames (levels, points, progress) may also encourage uptake and adherence to exercise [27, 28]. The feedback on progress alongside comparison or competition with other players may be persuasive and motivate longer-term use [29, 30]. A recent meta-analysis of effectiveness of virtual reality (VR) games for fall prevention in older people found positive effects on balance and fear of falling compared with no intervention; VR games were also concluded to be superior to conventional treatment [31]. But a high risk of bias, small sample sizes and large variability between methods and interventions mean evidence remains inconclusive and further research is needed. Other systematic reviews of virtual reality exercise programmes, Exergames and technology-based exercise interventions for older people [32,33,34] present similar conclusions that these interventions have potential, but larger and more rigorous studies are required to make more definitive conclusions regarding the effectiveness of Exergames to improve outcomes such as fall rate and fall risk.

The development and design of the Exergames used in this study were previously tested in a feasibility study [25, 35]. We followed the UK Medical Research Council (MRC) Guidance for Developing and Evaluating Complex Interventions [36] that recommends: identifying the evidence base and relevant theory; modelling process and outcomes; and then testing through feasibility or pilot studies prior to a full-scale evaluation [36].

Exergame programmes were planned to be carried out on a one-to-one basis (i.e. one person playing the Exergames at a time, supervised for research purposes by a physiotherapist or a physiotherapist assistant) either as part of a group setting or individually. Tailoring the Exergame programme and setting realistic, person-centred goals that are continuously reassessed to ensure personalisation and progression were also deemed important [37,38,39]. Understanding the health benefits of exercise and having self-belief in one’s ability to exercise (self-efficacy) are also necessary [41, 42]. There are also practical considerations for the delivery of the intervention such as ease of access to the exercise, particularly for frailer older adults unwilling or unable to travel [15, 43].

This trial investigated the effectiveness of a suite of Exergames that were developed with, and for, older people to improve function and reduce the risk of falls [25]. A series of games were co-created with older adults, therapists and software designers [25], which are based on OTAGO and FaME exercises for older people (strength and balance exercises with demonstrated effectiveness to reduce falls) [44,45,46]. The Exergames draw on self-determination theory and gamification to aid uptake, motivation and adherence to the Exergame programme [40]. Evaluation of this intervention is particularly timely given the current emphasis on healthy ageing and prevention [46, 47]. A cluster randomised trial design was adopted as individual randomisation within assisted living facilities would have been open to contamination bias. The research hypothesis was that a 12-week tailored programme of strength and balance Exergames will improve balance in people aged 55 years and older, in assisted living facilities.

Study objectives

The primary objective of this study was to evaluate the effectiveness of a tailored 12-week fall prevention Exergame programme on balance in adults aged 55 years or older as assessed by the Berg Balance Scale [48].

The secondary objectives were to investigate the effectiveness of the Exergame programme on fear of falling, lower limb function, self-reported physical activity, fall risk, pain, mood, fatigue, cognition, healthcare utilisation, health-related quality of life (HRQoL) and fall rates during a 3-month follow-up.

Methods

Design

This was a multi-centre, cluster randomised (1:1 ratio) controlled trial comparing fall prevention Exergames plus standard care (physiotherapist advice and leaflet) against standard care only in adults aged 55 years or older, dwelling in assisted living facilities (a cluster comprised one facility).

Study setting and participants

Clusters were assisted living facilities (also known as sheltered housing facilities or specialist housing) for people in Greater Manchester or Glasgow, identified via a national housing website [49]. Clusters were eligible if they housed residents aged ≥ 55 years, were willing to take part as agreed by their managers and had sufficient communal space (i.e. > 10 m2) to exercise without obstacles. The facilities comprised individual flats and bungalows (ranging from 28 to 80) with residents aged between 45 to 102 years. Each facility had a communal lounge where residents would meet to undertake the intervention. Between January and May 2016, 18 assisted living facilities were invited to participate in geographic areas that included a range of levels of deprivation in urban and semi-urban areas. Eligibility criteria for participants are presented in Table 1.

Table 1 Participant eligibility criteria for the study
Full size table

Managers at the assisted living facilities invited residents to attend an information session about the Exergames and the study, presented by ES or AM. Trained research staff and research physiotherapists initially identified residents who appeared to meet inclusion criteria, and ascertained if they were willing to receive information about the study, before providing them with participant information sheets and consent forms. After at least 24 h, participants were consented, fully assessed for inclusion as per protocol and underwent physical assessment by a research physiotherapist. All participants’ general practitioners (GP) were notified of study participation and eligibility criteria were verified using GP records.

Cluster randomisation and blinding

To reduce the potential for selection bias, baseline assessments took place before randomisation. Cluster randomisation of the units was based on blocks of two in each location (Manchester; Glasgow), matched according to readiness to deliver the intervention, and the number of consented participants per facility. Recruitment was staggered, with Manchester sites recruiting first. Randomisation was computer-generated (using Stata 14) [50] and performed by the Lancashire Clinical Trials Unit (CTU). Clusters were enrolled and assigned by ES and DS. Blinding of the assessors at the 12-week assessment was not possible for two out of the three assessors due to participants revealing the group allocation of their assisted living facility/cluster. Data analysis was performed by two unblinded researchers at the CTU (CJS, VB) based on an a priori statistical analysis plan. Recruitment commenced in Manchester in January 2016 and ended in Glasgow in June 2016, with the final follow-up being completed in November 2016.

Description of the interventions

The Control and Exergame group interventions are summarised in Table 2.

Table 2 Summary of the control and Exergame group interventions
Full size table

Control

A physiotherapist gave members of the control group standard community fall prevention advice comprising the Age UK Staying Steady leaflet [51] and the OTAGO strength and balance home exercise programme leaflet [52]. Control participants were encouraged to do three preselected (by the physiotherapist) exercises from the OTAGO list over the 12-week period. After 12 weeks, control participants were offered the opportunity to use the Exergame platform to reduce drop out [53].

Exergame intervention

In the intervention group, the same standard care as the control group was given. In addition, Exergames were offered three times a week (under the supervision of a physiotherapist or physiotherapist assistant) for 12 consecutive weeks in the assisted living facility communal rooms. Each participant was given a prescribed programme of standardised Exergames that suited the participant’s starting level of ability, with tailored progression (e.g. more Exergames within a session, greater challenge, longer duration) over the 12 weeks.

The physiotherapists and physiotherapist assistants were trained for approximately 30 min in the use of the Exergame system. This consisted of being shown the different games and corresponding exercises and how to set up the laptop and Kinect sensor (version 2; Microsoft Corp., Redmond, WA) and set up a programme for users (Fig. 1). The Kinect sensor requires placement 1 m above the ground and can be manually tilted so that it tracks the participant’s entire body. The participant needs to face the Kinect sensor, with no obstacles in front or to the side to allow free movements, and stand at a distance of 2–3 m.

Fig. 1
figure 1

Set-up and positioning of the sensor for Exergame intervention

Full size image

Physiotherapists and assistants familiarised themselves with the system prior to commencing the trial. Remote technical assistance was provided by MIRA Rehab during the trial. The Exergame system used in this trial was designed to engage and motivate individuals to participate in ‘game’-driven physical activities [54]. In an earlier feasibility study, this Exergame system showed potential to improve balance and increased engagement through motivational design [35]. The Exergame system enables interaction with the user by giving verbal and subtitled feedback on correct movements and any adjustments that may be required. It utilises the Microsoft Kinect (Microsoft Corp., Redmond, WA), a 3D motion tracking device that does not require handheld controls. This tracks the user’s performance and records parameters such as frequency and duration of use. Individual exercise programmes can be tailored using a choice of games for lower or upper limb exercises (see Additional file 1: Tables S1, S2 and Figure S2). The system also supports remote monitoring, whereby the user’s exercise programme (and progress) can be remotely viewed and adjusted by a remote supervisor, but this was not evaluated during this study.

Outcome measures

Assessments were completed with individual participants at baseline and 12 weeks by trained physiotherapists and included a series of standardised tests and questionnaires (Table 3).

Table 3 Outcome measures
Full size table

Balance at 12 weeks post-baseline, measured by the Berg Balance Scale [55], was the primary outcome measure.

A 3-month follow-up of self-reported participant falls (baseline assessment) was conducted using daily fall calendars that were posted monthly to the researchers. Participants who reported a fall during the previous month were contacted by telephone to record details of the fall [56].

The trial protocol is registered on ClinicalTrials.gov reference: NCT02634736.

Sample size

A sample size calculation (5% two-sided significance level, 90% power) to detect a between-group difference in mean clinically important change in the Berg Balance Scale (BBS) of 8 points [68], indicated a sample size of 23 participants in each group was needed. Assuming a common standard deviation (SD) of 8.0 (conservative estimate from a pilot trial) [35], using an independent-samples t-test and assuming a conservative intraclass correlation coefficient (ICC) of 0.05 and a cluster size of 9 participants, this leads to a design effect of 1.4 and sample size of 33 per arm. In our pilot study, 22/24 (92%) provided outcome data at week 6. Conservatively, we assumed a 75% retention rate at week 12. Therefore, at least 5 assisted living facilities in each arm (a total of 10 assisted living facilities), with a site average of 9 participants recruited, were required. However, to allow for unit attrition or limited participation, we recruited a total of 18 assisted living facilities.

Data collection and analysis

Participant characteristics assessed at baseline included age, sex, ethnicity, employment, marital status, socioeconomic status (Index of multiple deprivation of participants’ assisted living facility postcode), fall history and self-reported vision. Outcome measures (Table 3) were assessed at baseline and 12 weeks.

Statistical analysis

Our primary analysis was intention-to-treat; therefore, we included all participants willing to provide outcome data irrespective of their engagement with the Exergame or control intervention. We implemented this using a complete case analysis, and no imputation of missing outcome data was performed. However, a sensitivity analysis on the primary outcome variable was performed using multiple imputation (using a linear regression model, with age, BBS at baseline, location and group as predictors) [69] if there was more than 10% missing data or a between-group difference of more than 10% in the percentages with missing data. To account for the cluster randomisation, linear mixed effects modelling for the primary and secondary outcomes was used to compare the two groups. The group indicator was included in each outcome model as the focus of the analysis and adjustment was performed for the following variables: assisted living facility unit (random effect), baseline measures of the corresponding outcome variable (fixed effect), location indicator (Manchester; Glasgow; fixed effect). To investigate potential differences in effectiveness between locations, we added a location-by-intervention interaction term to the model for the primary outcome measure.

Fall data were summarised as recommended by ProFaNE [70] using the number of falls, number of non-fallers/single fallers/multiple fallers and fall rate per person-year. We investigated the effect of the Exergame intervention on fall rate by estimating the incidence rate ratio (IRR) using Mantel-Haenszel methods, stratified by match pair [71].

A significance level of 5% was used, and effectiveness estimation included both point and 95% confidence interval estimates. All analyses were performed using Stata 14 [50].

Economic analysis

The primary objective of the economic analysis was to assess the incremental cost-effectiveness of Exergames compared to treatment as usual (TAU). In this case, TAU was a visit from a physiotherapist to explain the Otago Exercise Programme (OEP) and to talk through a leaflet on fall prevention and an additional leaflet explaining the OEP recommended exercises. The control arm was asked to undertake their individually recommended exercises three times a week in their leisure time. However, TAU was not the full OEP (which may be the case in a non-trial setting) that is delivered by trained personnel at home [72] or in group setting [73]. The analysis was conducted alongside the Exergame trial from the perspective of the English National Health Service (NHS).

Quality-adjusted life years (QALYs) were calculated based on EQ-5D-5L using the area under the curve method assuming linear extrapolation of utility between time points. Following Hunter et al., we calculated the QALYs at the individual level (rather than at the group level) [74]. Health care services resource-use data were collected during the study and combined with relevant unit cost data for the financial year 2015–2016 to calculate the total health service costs incurred over the study period.

Multiple-imputation techniques were used for the main analysis presented here. Following White et al. and Faria et al., we used 15 imputations (as the missing data percentage was approximately 15%) [75, 76].

The incremental cost-effectiveness ratio (ICER) was calculated, adjusting for baseline EQ-5D-5L index score following Manca et al. [77]. We further adjusted for gender and age at baseline as well as identifiers for each of the residential homes.

Results

Study participants

Figure 2 shows the flow of participants through the study. Eighteen assisted living facilities gave permission at a managerial level to take part in the study and were randomised to control (n = 9) or to Exergame intervention (n = 9). The assisted living facilities ranged from smaller facilities with 19 occupied flats to larger facilities with 80 occupied flats. Following invitation by the managers of the facilities, 137 adults aged 55 years or older expressed an interest in participating in the study. Of these, 31 did not meet the inclusion criteria mostly related to participants not having the mental capacity to consent (Fig. 2). All assisted living facilities randomised were retained in the trial.

Fig. 2
figure 2

CONSORT flow diagram for cluster RCT

Full size image

The 106 participants’ baseline characteristics are presented in Table 4. Almost 80% of the participants were female. Mean ages of control and Exergame groups were similar (77.8 (SD = 10.2) vs 77.9 (SD = 8.9) years respectively), and there was a good balance between the groups in terms of ethnicity, employment and marital status. More participants from the Exergame group reported a fall in the previous year (58%) compared to control (42.9%) and Glasgow had higher index of multiple deprivation scores than Manchester sites, particularly in the Glasgow Exergame group.

Table 4 Baseline characteristics
Full size table

Study retention and adherence

Retention levels at 12 weeks was 86.8%. The most reported reasons for withdrawing from the study were ‘being medically (physically/mentally) unfit’ (n = 9), ‘family issues’ (n = 3) and ‘loss of interest’ (n = 2) (Fig. 2).

The Exergame participants attended a mean number of 25 (SD = 8.5) of a total of 36 sessions offered over the 12-week study period (a mean of 2 out of the 3 weekly offered sessions). Attendance at 12 weeks was 87.5% for the intervention group. The mean Exergame total exercise time at the end of the 12 weeks was 359 min (SD 151.2).

Effect of intervention

Primary outcome

Using ITT analysis, over 12 weeks, the Exergame intervention had a significant positive impact on balance as measured by BBS, relative to control [6.2 (95% CI 2.4 to 10.0; p = 0.003)]. The mean positive change of BBS from baseline was 2.9 points (SD 8.5) for the Exergame group, representing a 4.3% improvement from baseline, whilst the control group deteriorated by a mean of 2.8 (SD 6.5) points.

The estimated intracluster correlation coefficient for the BBS at 12 weeks was 0.08. There was no evidence that the effect of Exergames differed between the two locations (p = 0.39). As there was more than 10% missing outcome data, a sensitivity analysis was performed using multiple imputation of the missing values of BBS at week 12. Findings were not sensitive to ignoring the missing data and performing a complete case analysis, with the estimated impact of Exergames on BBS at week 12 being almost identical when using multiple imputation (6.2, 95% CI 2.7 to 9.8; p = 0.002), thus supporting the ITT results.

Secondary outcomes

Relative to controls, at 12 weeks the Exergames had a positive impact on fear of falling measured by Short FES-I (adjusted mean difference = − 2.7, 95% CI − 4.5 to − 0.8, p = 0.007) and VAS pain scale (− 12.1, 95% CI − 22.3 to − 1.8, p = 0.024). No statistically significant impact of the Exergame intervention was found on any other secondary outcomes (Table 5).

Table 5 Primary and secondary outcome measures at baseline and 12 weeks
Full size table

Follow-up fall incidence

A total of 55 falls were self-reported by the 106 participants during the 3-month follow-up. Of these, 38 falls were reported by the control group (12 fallers (24%) of whom 5 were single fallers and 7 were multiple fallers), an incident rate of 3.11 falls per person-year. Seventeen falls were self-reported by the Exergame group (11 fallers (20%) of whom 8 were single fallers and 3 multiple fallers), an incident rate of 1.26 falls per person-year. The incident rate ratio (IRR) of falls between groups was 0.31 (95% CI 0.16 to 0.62, p = 0.001) in favour of Exergames.

Health economic outcomes

Exergames were associated with a mean incremental total cost increase of £101.84 (95% CI − £7.42 to £211.11) and a mean incremental QALY gain of 0.007 (95% CI − 0.003 to 0.016) (Table 6). Whilst there were no statistically significant differences in costs or QALYs between the control and treatment arms during bootstrapping, the point estimate of the ICER was £15,209.80 per QALY.

Table 6 Cost-effectiveness analysis comparing Exergames with OEP leaflet control
Full size table

The cost-effectiveness plane (Fig. 3) plots the 10,000 bootstrap replications of incremental cost and QALY estimates, to help illustrate the uncertainty surrounding the point estimates in probabilistic terms. The replications were clustered predominantly in the north-east quadrant, reflecting the point estimates that Exergames resulted in a positive health gain, but at an increased cost. Exergames resulted in an incremental QALY gain in 91.5% (9151 out of 10,000) of bootstrap replications, and a higher cost than controls in 95% (9490 out of 10,000) of replications.

Fig. 3
figure 3

Cost-effectiveness plane for Exergames vs OEP leaflet control

Full size image

The cost-effectiveness acceptability curve (CEAC; Fig. 4) demonstrates how the probability that Exergames are cost-effective increases with the decision-maker’s willingness to pay. At the lower bound of the NICE threshold [78] of £20,000 per QALY, there was a 61% probability of Exergames being cost-effective. This rose to 73% at the upper bound of £30,000. Compared with controls, Exergames were likely to be cost-effective in 50% or more cases if decision makers are willing to pay approximately £15,500 for one additional QALY.

Fig. 4
figure 4

Cost-effectiveness acceptability curve of willingness to pay for one additional quality-adjusted life year (£)

Full size image

Exergame usability and acceptance

Participants scored the Exergame system (set up by the physiotherapist or assistant) on the technology assessment model (1 = strongly disagree to 7 = strongly agree) as easy to use (mean = 6.3, SD = 1.4) and useful (mean = 5.9, SD = 1.9). Participants had a favourable attitude (mean = 6.6, SD = 1.2) and indicated they intended to use the Exergame system in the future if it were to become available (mean = 5.7, SD = 2.2).

The mean system usability scale (range 0–100) at 12-week assessment was 82.4 (SD = 15.5) indicating high usability [79].

No major protocol deviations or unexpected adverse events occurred during the study period.

Discussion

This is the first cluster RCT of Exergames in community-based, assisted living facilities. Provision of Exergames for strength and balance exercises over 12 weeks to residents aged 55 years and older of these facilities significantly improved balance (as measured by the BBS), pain (VAS) and fear of falling (Short FES-I). Furthermore, an IRR = 0.31 for falls over a 3-month follow-up indicates an unexpected significant effect for this fall reduction.

The trial was powered on the minimum detectable change (MDC) of 8 points on the BBS from a trial of dependent older people (living in residential care homes) of somewhat older age than those in this trial [68]. However, other studies have found that the MDC for more physically independent older people receiving physiotherapy was rather less (3.2 points [80] and 4.9 points [81]). Furthermore, it has recently been advised that studies should not be powered on distribution-based statistics, such as the MDC [82], but should primarily use anchor-based methods and/or expert opinion to estimate the minimally important difference (MID).

There has been no research to determine the MID for the BBS on a population similar to ours, but Saso et al. [81] found an MID of 6 points for people early after stroke and Godi et al. [83] found an MID of 7 points for a groups of older people with balance deficits undergoing rehabilitation. Given the nature of the population in our trial, the MID would be expected to be not greater than 6 and perhaps a little smaller. The estimated difference in change between the groups for balance (BBS 6.2 points) is therefore plausible as an important difference, if only slightly exceeding the likely MID. However, the 95% confidence interval for this effect is relatively wide (2.4 to 10.0) so further research is recommended to provide further evidence as to the magnitude of the effect.

The results of our economic evaluation imply that Exergames are likely to be cost-effective compared to control (physio advice and the Otago Exercise Programme leaflet). These findings were robust, controlling for baseline characteristics using multiple imputation or complete case analysis, and choice of methodology to derive utility values from the EQ-5D-5L instrument. In the primary analysis, the ICER was £15,209.80 per QALY, and there was a 61–73% probability of Exergames being cost-effective at NICE thresholds [78] of £20,000–£30,000 per QALY. It should be noted that the full Otago Exercise Programme consists of a set of three times per week progressive exercises, over a 12-month period and a walking plan with multiple home visits by an instructor [45]. Therefore, traditional Otago Exercise Programme costs would be much higher than those proposed in this current 12-week study due to the additional labour costs, on-going training, travel, telephone calls and overhead costs.

This study demonstrated consistent findings between the locations indicating the potential transferability of the Exergame system. Retention of the Exergame participants was high (87%) compared to other exercise programmes for older people [84], with a mean attendance at Exergame sessions of around two thirds (69%). Although there is no current consensus on the cut-off points for levels of high, moderate or low adherence [85], this may be considered a beneficial level of attendance that resulted in improvements in balance, the primary outcome measure and also falls. Participants also reportedly liked the Exergames and found them easy to use (albeit, after it had been set up by the therapists, so this may refer to their understanding and playing of the Exergames). These are important factors when considering not just the effectiveness of the intervention but also whether people will actually want to use it in the long term [86]. We recorded no adverse events related to the 12-week intervention.

Strengths and limitations

Baseline assessment was undertaken pre-randomisation, thus limiting the potential for selection bias. Sample sizes were achieved and high retention rates suggest little effect of attrition bias. Recruitment across multiple locations provides some evidence of generalisability across assisted living facilities and should reduce risks of performance bias. The Exergames were tested with people aged 55 years or older using rigorous methods in a ‘real-world’ context in communal areas in assisted living facilities rather than in a controlled laboratory-type setting, and the physiotherapists and assistants had no prior experience of this type of technology. Benefits of the intervention were that the Exergames were co-designed with older people and therapists and are based on evidence-based exercises (FaME and OTAGO) and gamification health psychology [27] and can be tailored to a participant’s level of ability, preference and progression (i.e. different upper and lower limb exercises could be matched to different games and changed throughout the intervention period to maintain interest).

Cluster randomisation was essential to ensure that intervention contamination did not occur, but this also made blinding of the assessors difficult. Therefore, despite the use of objective measures and standardised participant instructions, it is possible that some of the improvements at follow up may have resulted from detection bias and so the results should be interpreted with caution. Blinding of the data analysts would also have been difficult, although there is sparse evidence that this may have affected outcomes [87]. An a priori analysis plan was agreed and executed to minimise risk of reporting bias.

The incidence rate ratio for falls over the 3-month follow up (IRR = 0.31) significantly favours the Exergame group and this is in line with community-based RCTs testing the exercises that the Exergames were based on [3, 9, 52]. But falls were only followed up for 3 months, and falls were self-reported by the participants. Recall bias could be present, and it is recommended in future studies that falls are prospectively followed up for 1 year as per the ProFaNE consensus guidelines [70]. A further limitation is that the majority of the participants were women and of white, British ethnic origin. Future studies need to target ethnic groups and more men. Changes in physiotherapy staffing (one change in Manchester, two changes in Glasgow) could have also affected the delivery of the intervention, as it was observed that it took around 2 weeks for the therapists to become confident in setting up and delivering the Exergames to the intervention group. There tended to be more remote technology support required during these early weeks to ensure the sessions ran smoothly. These support issues caused short delays in the delivery of the Exergames which could have reduced both the duration of the session and the enjoyment level of the participants.

Attendance at the Exergame sessions was generally good, but the amount of time spent exercising (mean 33 min per week) was well below the recommended level of 150 min per week of moderate aerobic activity for adults and older people [46]. This appears to be an issue, even in well-designed community exercise RCTs [88]. However, in one recent cluster RCT, increased levels of physical activity were self-reported, albeit below the recommended levels, and the fall rate was reduced [89].

Future studies are needed to investigate the optimum intensity, progression and long-term adherence to Exergame programmes. Moreover, such studies should investigate improvements in balance using clinical measures (e.g. BBS) with also include an instrumental measurement of static and dynamic balance (e.g. posturography) and include other outcomes such as lower limb muscle strength [14], fall rates, fall risk, fatigue and habitual physical activity. In particular, a definitive trial is recommended to evaluate the cost-effectiveness of the implementation of the Exergames with a primary outcome of fall rates, and costs per QALY along with a process evaluation comparing traditional delivery of exercise [86]. More research is also required to test the effectiveness of Exergames in other settings such as care homes, within early hospital discharge schemes, or to top-up or continue longer-term rehabilitation at home under remote monitoring by clinicians. This is particularly important as, in frailer older adults, unsupervised home exercise may be risky [19].

Conclusions

The use of technology to reduce falls and to support exercise uptake and adherence in younger and older adults is increasing. This study has demonstrated that an evidence-based programme of Exergames is an acceptable, effective and potentially cost-effective way to improve balance, pain and fear of falling. Exergames may be a scalable intervention to reduce falls (at a personal and societal level).