Database search resulted in 822 studies (figure 1). Three duplicated studies were removed. After analyzing titles and abstracts, 761 of studies did not meet the inclusion criteria. from the remaining 58 studies, six were not available in full text. Review of the full texts of the remaining 52 studies resulted in the exclusion of additional 40 that also did not meet the inclusion criteria: 13 had an active control group (e.g. other type of exercise or therapy), 11 were not RCT, four did not assess balance or mobility, five were conducted in long term care facilities, four did not include older adults, one did not involve exergames, and two performed exergames combined with other interventions. We included the remaining 12 studies in the systematic review.
Participants and intervention characteristics
A total of 1520 older adults participated in the 12 studies included, 903 (61%) were women. One trial did not report the sex of the participants (29). The mean age was 76±6 for the exergame group and 76±5 for the no exercise or health education group.
Regarding the exergame type, most studies used the commercial non-immersive Nintendo Wii ® system (1,29–36). The remaining studies used the following serious games: the Balance Rehabilitation Unit (BRUTM) - a customized rehabilitation program that contains a immersive environment in which the user interact through three dimensional glasses (37); the LegSys™ (BioSensics LLC, MA, USA) - an interactive exergame interface with five wearable joint angle and position sensors (30); and two studies used Kinect-based exergames – the iStoppfalls system (38), and the exergame program with the following serious games: apple game, tightrope standing, balloon popping and one-leg stand (39).
The mean of time of exposure to exergames was 825 minutes (number of sessions x duration of each session), ranging from 360 (30,37) to 1440 minutes (1). The mean number of sessions was 21 varying from eight (30) to 48 (38), and duration varied from four (30) to 16 weeks(38).
Seven studies had no intervention as a comparison control group (1,29,30,33–35,39). In one study, the control group performed cognitive exercises (32); in another study, the control group wore E.V.A. (Ethylene Vinyl Acetate) insoles in their shoes everyday (31). In three studies, the control group received education on falls prevention and physical activity(36–38). The intervention characteristics are detailed in table 2.
The instruments used to assess balance and mobility varied among studies. Three studies used the Timed Up and Go (TUG) test (30,31,33), the Berg Balance Scale (BBS) – three studies (32,33,39), the 30-second stand test – three studies (1,31,39), and the 8 feet up and go test – three studies (1,30,35). The other instruments used were the Functional Reaching Test (33), the Activities-specific Balance Confidence Scale (32,35), the Tinetti balance test and the unipedal stance test (36), the MiniBESTest and the functional gait assessment (34). The center of pressure based balance parameters the were assessed using force plates were velocity (31), sway (29,30,37) and limits of stability (37). Table 3 shows detailed information about center of pressure parameters. The trials included in the systematic review did not have enough data collected using the same mobility and balance instruments/tests to allow pooling the data for the calculation of summary statistics in a meta-analysis.
[INSERT TABLE 02 HERE]
[INSERT TABLE 03 HERE]
Regarding secondary and descriptive outcomes, four studies reported adverse events (30–32,34), safety –two studies (30,34) and adherence – five studies (30,32,34,37,38). Other outcomes were motivation – two studies (31,34), user experience – two studies (30,34), quality of life – two studies (1,32) and physical activity enjoyment – one study (32).
Effects on balance
Considering the outcomes related to center of pressure (CoP) based variables, there was no significant effect of exergaming on CoP velocity (<0.23 mm²/s; CI = -4.1 to 4.6; p = 0.92) (31). However, Cho et al (29) found significant CoP excursion improvement (decrease) both with eyes open and closed after an exergame intervention (p<0.01). Significant balance improvements were also observed by Schwenk et al (30) in CoP sway area (p=0.007; effect size = 0.23), and on the limits of stability and CoP sway area (p<0.01) (37) Figure 04a shows the effects of exergames considering the CoP sway with eyes open and closed. BBS score. Data suggest an effect in favor to the exergames regarding CoP sway (SMD = -0.89; 95% CI = -1.26 to -0.51) and heterogeneity of 58%.
The effects of exergaming on the BBS was evaluated in three studies (32,33,39). Padala et al (32) reported a significant improvement in BBS scores after four and eight weeks of exergaming (CI = 2.3 to 4.8 after four weeks, and CI = 4.3 to 6.7 after eight weeks; p<0.001). Sato et al (39) also found significant improvement in BBS scores in the exergame group in comparison to a control group, but the effects were smaller (CI = 0.22 to 1.9; p<0.01). Similarly, Jung et al(33) found significant improvement in BBS scores in the Nintendo Wii exercise group compared to a control group (p<0.001).
Figure 04b shows the effects of exergames considering the BBS score. In a total of n = 51 participants in the experimental groups versus n = 51 in the control group, data suggest an effect in favor to the exergames regarding postural balance assessed by BBS (MD = 2.15; 95% CI = 1.77 to 2.53), although a substantial heterogeneity was observed (I² = 96%).
In respect to other types of balance assessment, is was observed inconsistent findings on the effect of exergames based on changes in the Activity-specific Balance Confidence Scale (ABC) scores (32,35) Also, no between groups’ difference was observed using the Tinetti’s balance test (36) the MiniBEST Test (34)
Effects on mobility
The effects of exergames on TUG time were reported by three studies. Jung et al(33) found a significant difference between groups with better TUG performance in the exergame group than in the control group (p<0.001). Similarly, Jorgensen et al(31) reported a between group difference in TUG time of -1.4 seconds (CI = -2.5 to -0.4; p= 0.01), and Schwenk et al (30) also found a significant better performance in TUG test in the exergame group. Figure 04c shows the effects of exergames considering the TUG test. A total of n = 50 participants in the experimental groups versus n = 53 in the control group. Data suggest an effect in favor to the exergames regarding TUG (MD = -2.48; 95% CI = -3.83 to -1.12) with no heterogeneity (I² = 0%.).
The other studies that assessed mobility used the following instruments: 8-foot up and go test (1,30,35), 30-second chair stand test (1,31,39) and MiniBEST test (34), Functional Reach Test (39,40), the 6-meter Walking Test (1) and the Functional Gait Assessment (34) They all found significant better mobility for the exergame groups than for the control groups (p<0.05).
The effects of exergaming on the 8-foot up and go test was evaluated in three studies (1,30,35), Rendon et al (35) reported a significant improvement in 8-foot up and go test after six weeks of exergaming (median = 8.8; min = 5.1; max = 23.44; p = 0.045). Significant improvement in 8-foot up and go test in the exergame group in comparison to control (change = -1.07±0.74; p<0.01) was found (1). Schwenk et al (30) found significant improvement in 8-foot up and go test (19% of change; p=0.037). Figure 04d shows the combined effect of exergames considering the 8-foot up and go test. In a total of n = 39 participants in the experimental groups versus n = 41 in the control group, data suggest an effect in favor to the exergames (MD = -1.88; 95% CI = -2.40 to -1.38), with heterogeneity of I² = 53%.
Four studies that used Nintendo Wii® reported no adverse effects (30–32,34). The remaining studies did not mention adverse effects.
The studies using Nintendo Wii® reported good adherence: 100% (32), 93% (30) and 80% (34). The participants who did not adhere reported transportation issues, back pain and unrelated medical. Similarly, the studies that used the BRUtm (37) and the iStoppFalls system (38) reported a 97% and 81% of adherence. The causes for no adherence were transportation issues (37), motivation, personal, health and system-related issues (32).
Quality of life
Quality of life was investigated in two trials. Padala et al(32) found no difference between experimental and control groups on SF-36 scores. Maillot et al(1) found significant improvements in the social functioning (p<0.05) and global mental health (p<0.01) domains of the SF-36 in the experimental group that played Nintendo Wii® games.
Safety information was extracted from the user experience questionnaire (30). The older adults could “completely disagree” (0), “moderately disagree” (1), “neutral” (2), “moderately agree” (3) or “absolutely agree” (4) with 10 statements. The following six were safety-related: “I never lost my balance while using the exercise technology” (4±1). “I was afraid to tumble or to fall during the exercise” (0.2±0.6). “I required balance support while conducting the exercises” (0.5±1). “I feel that the exercises were going too fast for me” (0.2±0.4). “Some of the movements were difficult to perform” (1±0.9). “I felt safe using the exercise technology” (4±0.4). Gomes et al(34) used a “Game Satisfaction Questionnaire”. One of the questions was “Did you feel safe playing the games? If not, why?”. All participants from the experimental group stated that they felt safe.
Motivation and enjoyment
One study investigated enjoyment (32) and two assessed motivation (31,34) of the participants who played exergames in the Nintendo Wii®. Based on the Physical Activity Enjoyment Scale (PACES), 83% of the participants rated the Wii-Fit to be high on the measure of pleasure, 75% considered the Nintendo Wii® as fun, 75% considered pleasant, 67% rated it as invigorating, 83% as gratifying, 83% as exhilarating, 92% as stimulating and 92% as refreshing. Motivation was assessed using a Likert scale for the sentence (31): “I find the Nintendo Wii training both fun and motivating”, and 70% of the participants strongly agreed; 25% agreed, and 5% were undecided. The “Game Satisfaction Questionnaire” assessed motivation using two questions (34): “Did you feel motivated to play the games?” and “Would you like to play the games with someone?”, and 83% of the participants said they were “very motivated” and would like to play the games with someone, and 17% said they were “motivated” and would not play with someone.
Risk of bias in the included studies
The risk of selection bias was low in five of the twelve studies (31),24,26,28,32 for both sequence generation and for allocation concealment. Two studies showed low risk of bias for sequence generation, and unclear for allocation concealment (1,40). The level of risk of bias was unclear in four studies (29,33,35,37), and high in one study (36).
High risk for performance bias was observed in two studies (34,36). The performance bias risk was unclear for six studies (29,32,33,35,38,39), and it was low risk for four trials (1,30,31,37). The risk of detection bias was low risk in six of the twelve studies (30,31,34,35,37,38); unclear in four trials 1,23,27,33, and high in two study (32,36). Attrition bias was low in six trials (31–35,39), unclear in four trials (1,29,30,38), and high in two (36,37). For reporting bias, five of the twelve studies had low risk (30–32,34,36), six trials were unclear 1,23,27,29,32,33, and the risk of reporting bias was high for one trial (37). Other source of risk of bias was considered high in one study (Nintendo lent the equipment for the training) (36).