Effects of Multicomponent Digital Health Interventions on Multidimensional Physical Activity in Older Adults: Systematic Review, Meta-Analysis, and Meta-Regression of Randomized Controlled Trials

Background: The comprehensive effects of multicomponent digital health interventions (DHIs) on multidimensional physical activity indicators and sedentary behavior (SB) remain controversial. Objective: This systematic review aimed to evaluate the impact of multicomponent DHIs on daily steps, moderate-to-vigorous physical activity (MVPA), light physical activity, total physical activity, and SB in older adults. Methods: PubMed, Web of Science, Embase, The Cochrane Library, and CINAHL were searched up to February 20, 2026. Randomized controlled trials concerning multicomponent DHIs for promoting exercise behavior in older adults were included. RoB 2.0 was used to evaluate study quality. Meta-analyses were performed using the Hartung-Knapp-Sidik-Jonkman random-effects model, and 95% prediction intervals (PIs) were calculated via Nagashima adjustment to evaluate effect dispersion. The GRADE (Grading of Recommendations, Assessment, Development, and Evaluation) system was used to evaluate evidence certainty. Results: A total of 26 randomized controlled trials (n=4129) were included. The results showed that multicomponent DHIs significantly improved daily steps (mean difference [MD] 822.8, 95% CI 198.3 to 1447.3 steps/d; 95% PI -1452.4 to 3098.0) and MVPA (MD 45.9, 95% CI 23.9 to 67.9 min/wk; 95% PI -9.4 to 101.2). However, the improvements in SB (MD -283.7, 95% CI -610.8 to 43.5 min/wk; 95% PI -984.5 to 417.1), total physical activity (MD 104.4; 95% CI -109.2 to 318.0 min/wk; 95% PI -444.4 to 653.2), and light physical activity (MD 39.3, 95% CI -96.2 to 174.7 min/wk; 95% PI -227.6 to 306.2) did not reach statistical significance. As some included studies combined digital tools with human support, the independent contribution of digital technology remains uncertain. PIs indicated a certain degree of dispersion across different clinical contexts. Subgroup analysis showed higher effect sizes for standalone wearables, human-assisted interventions, and populations with chronic disease risks. Meta-regression showed that effect sizes remained stable across different ages and durations. The trim-and-fill method confirmed the robustness of MVPA results. GRADE assessment indicated "moderate" certainty for MVPA and "low" for daily steps and other indicators. Conclusions: This systematic review suggests that multicomponent DHIs may serve as an effective means for enhancing daily steps and MVPA in older adults. The innovation lies in evaluating the true effect distribution of multicomponent DHIs through Hartung-Knapp-Sidik-Jonkman random-effects models and Nagashima PIs. Compared with previous studies, this review identified the impact of population characteristics and control group differences on effect estimates using PI and subgroup models, confirming that advanced age did not significantly diminish the good adaptability of older adults to DHIs. Evidence limitations include high heterogeneity, lack of long-term follow-up, and differences between objective and subjective measurement tools. In practice, priority should be given to hardware carriers with simplified interaction and integrated human support, with tailored strategies developed for different risk subgroups.