Dose-response relationship between physical activity and subjective sleep quality in patients with insomnia: a systematic review and Bayesian meta-analysis of randomized controlled trials.

Abstract

BACKGROUND: Insomnia is a prevalent public health concern with significant consequences for well-being. While physical activity (PA) is recognized as a beneficial non-pharmacological intervention, its precise dose-response relationship with subjective sleep quality remains insufficiently quantified. This study aimed to systematically evaluate the efficacy of PA on subjective sleep quality in patients with insomnia and to quantitatively delineate the dose–response relationship using Bayesian meta-analysis. METHODS: A systematic search was conducted across Web of Science, PubMed, Embase, PsycINFO, and the Cochrane Library from inception to February 2026. Randomized controlled trials (RCTs) investigating the effects of PA on subjective sleep quality in participants diagnosed with insomnia were included. A Bayesian random-effects meta-analysis was performed to calculate pooled effect sizes (Hedges’g) with 95% Credible Intervals (CrI). Dose–response relationships were modelled using restricted cubic splines to identify minimum and optimal dosages. Dosage was expressed as the weekly volume in MET-minutes (Metabolic Equivalent of Task × minutes/week). The certainty of evidence was assessed using the GRADE approach. RESULTS: 24 RCTs involving 1,591 participants were included. Bayesian synthesis indicated low evidence that PA significantly improves subjective sleep quality compared with inactive controls (Hedges’g = -0.56, 95% CrI: -0.68, -0.45). Dose–response analysis revealed a non-linear “L-shaped” relationship. Benefits accrued rapidly at lower volumes, with a minimum effective dose estimated at 100 METs-min/week. The therapeutic effect peaked at an optimal dose of approximately 650 METs-min/week (Hedges’g = -0.64, 95% CrI: -0.85, -0.44), beyond which additional volume yielded diminishing marginal returns. Moderation analyses indicated that the effect was more pronounced in adults (Hedges’ g = -1.00) than in older adults (Hedges’ g = -0.51), and was consistent across different diagnostic methods and measurement instruments. Various types of exercise, including aerobic, strength, mind-body, and multicomponent training, all demonstrated beneficial effects. CONCLUSIONS: Physical activity is an effective treatment for insomnia, characterized by an L-shaped dose–response curve. The relationship is characterized by a rapid initial benefit at low volumes, with a statistically significant response observed at 100 MET-min/week and a plateau in effect occurring at approximately 650 MET-min/week, beyond which additional volume yields diminishing returns. TRIAL REGISTRATION: CRD420251172873.