Eleven studies were included in the review (1,928 participants) comprising five randomised controlled trials (RCTs, 1,412 participants), two non-RCTs (406 participants), and four pre-test post-test studies (110 participants). Six studies had population sizes based on statistical power calculations. The authors suggested that selection bias in participant recruitment was possible, Very few studies adjusted for confounders in analyses. Recall bias was potentially present in three studies. Five studies reported poor program adherence and significant loss of follow-up.
Mixed effects were reported in five studies, in which all participants were cognitively impaired or had dementia. One non-randomised controlled trial reported that mobility, balance and cognition significantly declined in the control group compared with the physical activity intervention group. One pre-test post-test study demonstrated significant improvement on one measure of physical performance using a resistance training program. Three studies (including an RCT) demonstrated no significant positive effects of the interventions on any of the fall-related outcomes assessed.
Six studies included both cognitively impaired and cognitively intact participants. None of these studies performed a subgroup analysis on the cognitively impaired participants, so no inferences could be made on the effects of the interventions on this subpopulation.