Efficacy of interventions used by physiotherapists for patients with headache and migraine-systematic review and meta-analysis

Aim: We aimed to conduct a systematic review evaluating the effectiveness of interventions used by physiotherapists on the intensity, frequency and duration of migraine, tension-type (TTH) and cervicogenic headache (CGH). Methods: We performed a systematic search of electronic databases and a hand search for controlled trials. A risk of bias analysis was conducted using the Cochrane risk of bias tool (RoB). Meta-analyses present the combined mean effects; sensitivity analyses evaluate the influence of methodological quality. Results: Of 77 eligible trials, 26 were included in the RoB assessment. Twenty trials were included in meta-analyses. Nineteen out of 26 trials had a high RoB in >1 domain. Meta-analyses of all trials indicated a reduction of TTH (p<0.0001; mean reduction -1.11 on a 0-10 visual analog scale (VAS); 95% CI -1.64 to -0.57) and CGH (p=0.0002; mean reduction -2.52 on a 0-10 VAS; 95% CI -3.86 to -1.19) pain intensity, CGH frequency (p<0.00001; mean reduction -1.34 days per month; 95% CI -1.40 to -1.28), and migraine (p=0.0001; mean reduction -22.39 hours without relief; 95% CI -33.90 to -10.88) and CGH (p<0.00001; mean reduction -1.68 hours per day; 95% CI -2.09 to -1.26) duration. Excluding high RoB trials increased the effect sizes and reached additional statistical significance for migraine pain intensity (p<0.00001; mean reduction -1.94 on a 0-10 VAS; 95% CI -2.61 to -1.27) and frequency (p<0.00001; mean reduction -9.07 days per month; 95% CI -9.52 to -8.62). Discussion: Results suggest a statistically significant reduction in the intensity, frequency and duration of migraine, TTH and CGH. Pain reduction and reduction in CGH frequency do not reach clinically relevant effect sizes. Small sample sizes, inadequate use of headache classification, and other methodological shortcomings reduce the confidence in these results. Methodologically sound, randomized controlled trials with adequate sample sizes are required to provide information on whether and which physiotherapy approach is effective. According to Grading of Recommendations Assessment, Development and Evaluation (GRADE), the current level of evidence is low.