Meta-analysis finds non-linear exercise dose-response for mobility in stroke patients aged 45+

OBJECTIVE: This study systematically evaluated the effects of both overall and modality-specific exercise doses on lower-limb functional mobility in middle-aged and older adults with stroke, and identified optimal dose-response relationships using Bayesian non-linear modeling. METHODS: Twenty-five randomized controlled trials involving 788 stroke patients aged ≥45 years were included. Lower-limb function was assessed using the Timed Up and Go test. Exercise dose was standardized as MET·min/week, incorporating frequency, duration, and intensity. Bayesian model-based network meta-analysis was used to estimate non-linear dose-response curves across six intervention modalities: gait training, motor control training, resistance training, integrated aerobic training, combined cognitive-exercise training, and virtual reality training. RESULTS: A non-linear association was observed between overall exercise dose and functional mobility improvement, with a peak effect (SMD ≈ 0.45) at approximately 1400 MET·min/week. Optimal dose zones varied across modalities: resistance and gait training showed strong effects at low doses (∼560 MET·min/week), integrated aerobic and combined cognitive-exercise training peaked at higher doses (∼1700-1800 MET·min/week), while motor control training achieved significant effects (∼SMD 0.50) at moderate doses (∼890 MET·min/week). Virtual reality training exhibited wide credible intervals and non-significant effects. CONCLUSION: This study demonstrates that lower-limb functional mobility in stroke patients follows a clear moderate-dose optimal pattern, with improvements peaking at around 1400 MET·min/week. Modality-specific analyses show that exercise approaches do not differ in inherent superiority but instead present distinct dose-response profiles. Evidence for virtual reality training remains inconclusive due to wide credible intervals. These findings provide quantitative guidance for dose-informed, individualized exercise prescriptions in neurorehabilitation.