Abstract
BACKGROUND: To elucidate the relative efficacy of diverse exercise modalities for migraine and quantify the optimal therapeutic dosage. METHOD: A systematic search was conducted across four electronic databases from their inception to May 2025. Two reviewers independently performed data extraction and risk of bias assessment. A multilevel network meta-analysis (ML-NMA) integrated with a dose-response analysis was employed to comprehensively compare these interventions. RESULTS: Twenty-seven randomized controlled trials (RCTs) (n = 1,611) were included. The most effective interventions were combined aerobic+resistance exercise (g = - 1.85, 95% credible interval (CrI): [-2.53 to -1.18]; surface under the cumulative ranking curve (SUCRA) = 0.91), followed by resistance exercise (g = - 1.45, 95% CrI [-1.79 to -1.10]; SUCRA = 0.81), yoga (g = - 0.35, 95% CrI [-0.63 to -0.06]; SUCRA = 0.49), and tai chi (g = - 0.48, 95% CrI [-0.91 to -0.05]; SUCRA = 0.46). The dose-response analysis identified an optimal therapeutic window of 300-600 metabolic equivalent of task (MET)-min/week, an intensity of 4.5-5.5 METs, and a duration of 8-10 weeks. The overall certainty of evidence was rated from very low to low. CONCLUSION: Combined aerobic+resistance exercise, resistance exercise, tai chi, and yoga represent promising therapeutic options for migraine. The optimal dose was identified as approximately 70-135 minutes of moderate-intensity or 45-90 minutes of vigorous-intensity activity weekly, for 8-10 weeks. These findings, however, must be interpreted with caution due to the low quality of the underlying evidence.