Participant demographics are shown in Table 1. Descriptive statistics are presented at Table 2. Subscales from the UTAUT [85] and FSS [86] questionnaires demonstrated high internal-consistency reliability exceeding the minimum Cronbach’s alpha of 0.7 [93]. Seventeen clusters from the MAPS questionnaires were deemed reliable having exceeded 0.7. Clusters that did not show internal reliability were temporal pain, faint pain, muscle/joint pain, mental distress, respiratory distress, cold pain, pain restriction, anxiety, emotional avoidance, treatable illness and mentally engaging, with Cronbach’s alpha values ranging from 0.16 to 0.63, respectively.
Primary outcomes
Pain intensity
No effect of intervention was found on self-reported pain intensity experienced within 30 days before and after the intervention, and pain intensity at the time of testing, as determined by the ANCOVA with pre-measures as the covariate (see Table 3). Although the mixed ANOVA did not show any significant differences in pain intensity over time, the interaction effect between time and intervention was significant for pain intensity experienced at the time of testing in favour of exergaming (F [1. 52] = 3.98, p = 0.05, ε2 = 0.46, large effect). The 30% significant reduction in perceived pain intensity in the exergaming exceeded the appropriate cut-off point for determining the minimal clinically important differences (MCID) of changes in pain intensity of 15%, where a numerical rating change score of -2.0 and a percent change score of -33% are best associated with the concept of “much better improvement” [94].
Multidimensional affect and pain variables (MAPS)
ANCOVA revealed that the variable physically engaged; (F [1,48] = 3.76, p = 0.06, ε2 = 0.01, small effect) from the well-being subcluster approached significance in favour of exergaming (see Table 4). This suggests meaningful increases in older people’s feelings of being active and vigorous after exergaming. Thermal pain (F [1,48] = 14.43, p = 0.00, ε2 = 0.09, medium effect) showed a significant effect of intervention in favour of exergaming.
The mixed ANOVA revealed significant effects of time on depressed mood (F [1, 50] = 9.09, p = 0.004, ε2 = 0.01, small effect) and affiliative feelings (F [1, 50] = 6.92, p = 0.01, ε2 = 0.03, small effect) in favour of exergaming. Moreover, three variables approached significance also in favour of exergaming. They were thermal pain (F [1,50] = 3.85, p = 0.06, ε2 = 0.01, small effect), anger (F [1,50] = 3.76, p = 0.06, ε2 = 0.01, small effect) and physically engaged (F [1,50] = 3.82, p = 0.06, ε2 = 0.01, small effect).
Postural control
Although the ANCOVA did not reveal any effect of intervention on postural control, the mixed ANOVA showed that there were significant reductions over time for AP SD (F [1, 46]= 8.29, p = 0.01, ε2 = 0.09, medium effect), ML SD (F [1, 46]= 8.37 p = 0.01, ε2 = 0.05, nearly medium effect), AP range (F [1, 45]= 9.91, p = 0.003, ε2 = 0.16, large effect) and ML range (F [1, 45]= 4.12, p = 0.05, ε2 = 0.06, medium effect) during bipedal standing with vision, and for CoP excursion in the medio-lateral direction (F [1, 47]= 5.43, p = 0.03, ε2 = 0.08, medium effect) during pedal standing without vision (see Table 5).
Technology acceptance
UTAUT scores increased in both groups which indicated moderate-to-high acceptance for both exergaming and TGB. The ANCOVA found significant effects of intervention in favour of TGB for social influence (F [1, 44] = 5.16, p = 0.03, ε2 = 0.06, medium effect) and behavioural intention (F [1, 44] = 4.99, p = 0.03, ε2 = 0.08, medium effect) (see Table 6). Higher mean values occurred in the control group indicating a greater level of acceptance towards TGB rather than exergaming.
The mixed ANOVA revealed a statistically significant increases over time for all the UTAUT measures – performance expectancy (F [1, 46] = 45.04, p < 0.001, ε2= 0.36, large effect), effort expectancy (F [1, 46] = 49.40, p < 0.001, ε2= 0.37, large effect), social influence (F [1, 46] = 42.69, p < 0.001, ε2= 0.34, large effect), facilitating conditions (F [1, 46] = 28.07, p < 0.001, ε2= 0.27, large effect), self-efficacy (F [1, 46] = 26.27, p < 0.001, ε2= 0.27, large effect) and behavioural intention (F [1, 46] = 43.96, p < 0.001, ε2= 0.38, large effect). A significant interaction effect was between time and intervention was found for social influence (F [1, 46] = 6.73, p = 0.01, ε2= 0.05, almost medium effect) in favour of TGB.
Flow
The ANCOVA (shown in Table 7) revealed a significant effect of intervention on concentration of task (F [1, 44] = 5.67, p = 0.02, ε2 = 0.09, medium effect) favouring TGB whereas autotelic experience (F [1, 44] = 4.06, p = 0.05, ε2 = 0.04, small effect) and paradox of control (F [1,44] = 3.63, p = 0.06, ε2 = 0.05, medium effect) approached significance, also in favour of TGB. Nevertheless, the results showed a direction of increase in post-intervention scores for these variables. No effect of intervention was found on the other variables: challenge-skill-balance (F [1,44] = 3.21, p = 0.08, ε2 = 0.04), transformation of time (F [1, 44] = 2.09, p = 0.16, ε2 = 0.02), loss of consciousness (f [1, 44] = 1.29, p = 0.26, ε2 = 0.01), feedback (F [1, 44] = 1.96, p = 0.17, ε2 = 0.02). The same was found for variables that did not meet homogeneity of regression as determined by mixed ANOVA by blocking: clear goals (F [1, 44] = 1.25, p = 0.27, ε2 = 0.01) and action-awareness-merging (F [1, 44] = 0.47, p = 0.50, ε2 = 0.00).
Mixed ANOVA found significant increases over time in all of the flow variables – autotelic experience (F [1, 46] = 40.20, p < 0.001, ε2 = 0.23, large effect), clear goals (F [1, 46] = 13.57, p < 0.001, ε2 = 0.16, large effect), challenge-skill-balance (F [1, 46] = 57.69, p < 0.001, ε2 = 0.32, large effect), concentration at task (F [1, 46] = 49.27, p < 0.001, ε2 = 0.32, large effect), paradox of control (F [1, 46] = 47.46, p < 0.001, ε2 = 0.33, large effect), unambiguous feedback (F [1, 46] = 63.12, p < 0.001, ε2 = 0.37, large effect), loss of consciousness (F [1, 46] = 56.01, p < 0.001, ε2 = 0.35, large effect), transformation of time (F [1, 46] = 21.96, p < 0.001, ε2 = 0.16, large effect) and loss of self-consciousness (F [1, 46] = 41.39, p < 0.001, ε2 = 0.29, large effect).
Physiological outcomes
The ANCOVA did not reveal significant effect of interventions on the physiological measures, although heart rate was higher in the exergaming group. The mixed ANOVA however, revealed significant increases over time in perceived physical effort (F [1, 46] = 11.24, p = 0.002, ε2 = 0.07, medium effect) and perceived subjective mental effort (F [1, 46] = 15.12, p = 0.00, ε2 = 0.09, medium effect) (see Table 8). Post-intervention scores for perceived physical exertion and mental effort were lower in the exergaming group compared to the control group (as earlier shown in Table 2). These scores suggest that exergaming may be less strenuous and require less effort compared to TGB.