Frontiers in Psychology · Oct 13, 2026

Effects of exercise on balance in people with Parkinson's disease: an umbrella review of systematic reviews and meta-analyses of randomized controlled trials

By Yuan Y, Ran H, Song Y et al.

BACKGROUND: Postural instability and impaired balance are disabling, levodopa-resistant features of Parkinson's disease (PD) and leading contributors to falls. Many systematic reviews and meta-analyses (SR/MAs) report that exercise improves balance, but they differ in modality, comparator, instrument and rigor and have not been synthesized and graded as a whole. We conducted an umbrella review of meta-analytic evidence on exercise and balance in PD. METHODS: Following PRISMA 2020 and umbrella-review methodology, PubMed and Web of Science were searched from inception to 8 June 2026. Pairwise SR/MAs of randomized controlled trials (RCTs) reporting balance performance (Berg Balance Scale [BBS], Mini-BESTest), mobility (Timed Up-and-Go [TUG]) or falls in PD were eligible; network meta-analyses were supporting evidence only. We extracted the pooled random-effects estimates as reported by each review (no re-pooling of individual-trial data), harmonized their direction, and interpreted standardized effects as small/moderate/large. Methodological quality, primary-study overlap and certainty were appraised with AMSTAR-2, the corrected covered area (CCA) and GRADE. RESULTS: From 161 unique records, 20 pairwise meta-analyses were included, spanning aerobic, resistance, aquatic, dance, VR/exergaming, traditional Chinese exercise, and home-based and group formats. Exercise improved balance in most modalities, though not universally (resistance training and one mixed-exercise review were non-significant). Because estimates came from separate meta-analyses, they are not directly comparable; standardized balance-performance effects ranged from small (g = 0.33, 95% CI 0.11-0.55) to large (aerobic, BBS g = 0.99, 0.76-1.23). Evidence on falls was inconsistent (proportion of fallers unchanged; fall rate reduced). Between-review overlap was very high at the BBS node (CCA = 16.8%). AMSTAR-2 confidence was moderate in 9, low in 10 and critically low in 1; GRADE certainty was moderate-to-very-low. CONCLUSION: Exercise is associated with small-to-large improvements in clinical balance in PD, but evidence on falls is inconsistent, and the underlying meta-analyses are mostly of low-to-moderate quality, highly overlapping, and of moderate-to-very-low certainty. Aerobic, mind-body and game-based modalities show relatively consistent signals, but this umbrella review cannot establish superiority of any modality. Larger, blinded, registered RCTs with standardized balance and prospectively ascertained fall outcomes are needed.

doi.org/10.3389/fpsyg.2026.1939213Open accessPubMed 42947838Interventions