Abstract
BACKGROUND: Traditional meta-analyses with pairwise direct comparison revealed that a variety of exercise-based training interventions can prevent falls in community-dwelling older adults. This network meta-analysis adds value by comparing and ranking different exercise training strategies based on their effects on fall risk reductions determined by analysis of direct and indirect comparisons. METHODS: The studies included in this network meta-analysis were identified through a comprehensive search in five biomedical databases (PubMed, SportDiscus, CINAHL, Web of Science and EMBASE). We included (randomized) controlled trials (RCTs) that compared the occurance of fall events in older adults who received different interventional treatments. RESULTS: Seventy six comparisons from 66 RCTs with 47.420 (61% male / 39% female) participants aged 77 ± 4 (68 - 88) years were included in this network meta-analysis. The network model revealed low heterogeneity (I(2) = 28.0, 95%CI 1.0 to 47.7%) and inconsistency (Q between designs = 15.1, p = 0.37). Postural control training was found to be most effective in preventing falls (Postural Control Training: (home): Risk Ratio (RR) = 0.66, 95%-CI [0.49; 0.88], P-score = 0.97;Postural Control Training: RR = 0.82, 95%-CI [0.75; 0.91], P-score = 0.82). Combined and multifactorial interventions also display a robust but smaller effect (RR = 0.88-0.93, P-score = 0.65-0.47). CONCLUSION: Physical activity that includes balance training presents itself to be the most effective. Multifactorial approaches are well investigated but could be slightly less effective than isolated postural control training.