Multilevel Network Meta-Analysis of Nonpharmacological Interventions for Migraine: Focusing on the Dose-Effect of Physical Exercise and Its Moderators

To compare the relative efficacy of common nonpharmacological treatments for migraine and to determine the optimal dosage for physical exercise.

We searched 4 databases up to January 2025 for randomized controlled trials of nonpharmacological interventions for migraine. A multilevel network meta-analysis, integrated with a dose-response analysis, was conducted to compare intervention efficacy and determine the optimal exercise dosage. Treatments were ranked by the Surface Under the Cumulative Ranking curve. Two independent reviewers extracted data and assessed the risk of bias.

Fifty-nine randomized controlled trials involving 10,020 participants (78.1%, female) were included. Neuromodulation techniques were most effective (Hedges' g =-0.61, 95% credible interval: -0.89 to -0.33), followed by physical exercise (Hedges' g =-0.42, 95% credible interval: -0.67 to -0.18), and mindfulness meditation (Hedges' g =-0.38, 95% credible interval: -0.63 to -0.12). The dose-response analysis for exercise indicated that while 100 metabolic equivalent (MET)-minutes per session was statistically effective, a minimum of 110 MET-minutes per session was required to achieve the Minimal Clinically Important Difference. Efficacy reached an optimal therapeutic plateau at 250 to 300 MET-minutes per session, achievable with 3 to 5 weekly sessions of 30 to 40 minutes.

Neuromodulation, physical exercise, and mindfulness meditation are promising nonpharmacological therapies for migraine. For physical exercise, a minimum dose of 110 MET-minutes per session is needed for clinically significant effects, with an optimal therapeutic window at 250 to 300 MET-minutes per session. Due to the low quality of primary evidence, these findings warrant cautious interpretation and require future validation.