Thirteen studies (5,866 patients, range 37 to 1,816) were included in the review. The overall quality of the included studies was judged to be moderate to good. However, interpretation of identification of migraine test blind to the reference standard was unclear in 12 studies, and interpretation of the reference standard blind to the migraine tool was unclear in five studies. In addition, 10 studies did not clearly report the time between application of the identification of migraine test and the reference standard. Finally, possible spectrum bias was identified in seven studies that did not include patients who were representative of those who should receive the identification of migraine test in primary care, neurology departments, or headache clinics.
The pooled estimate of sensitivity for the identification of migraine tool was 84% (95% CI 75 to 90); pooled specificity was 76% (95% CI 69 to 83). The positive likelihood ratio pooled estimate was 3.55 (95% CI 2.76 to 4.57) and pooled negative likelihood ratio was 0.21 (95% CI 0.14 to 0.32).
A sensitivity analysis, which included the six studies conducted in primary care, headache clinics and neurology departments, gave similar results.