How many patients do we need? Sample size calculation in headache research

An adequately planned sample size is a fundamental component of clinical research. Studies that enroll too few participants may fail to detect clinically important effects, whereas unnecessarily large studies consume resources and may expose more participants than needed. This narrative review provides a practical introduction to sample size calculation for clinicians and investigators conducting headache research. We review the null and alternative hypotheses, type I and type II errors, statistical power, target difference (effect size), dropout, pilot studies, clustered designs, and repeated measurements, using examples from migraine, cluster headache, acute-treatment trials, preventive trials, observational studies, and interventional headache research. Particular attention is given to clinically meaningful differences in monthly migraine or headache days, responder outcomes, 2-hour pain-freedom endpoints, variability and missing data in headache diaries, and recruitment constraints in uncommon headache disorders. We emphasize that the calculation should be aligned with the prespecified primary outcome and planned analysis, that every assumption should be justified and reported transparently, and that feasibility should not be used retrospectively to legitimize an underpowered efficacy study. Early collaboration with a biostatistician remains strongly recommended.