Validity of self-reported migraine in adolescents and children.

Abstract

To assess agreement for migraine day between self-report and diagnostic guidelines for children and adolescents using a headache diary. Trial guidelines recommend prospective collection of headache features and adoption of migraine day as an outcome measure, but there is no clear consensus on the definition of migraine day. This is a secondary analysis of data from two projects-a prospective cohort study validating a pediatric scale of treatment expectancy and a clinical trial of occipital nerve blocks to treat status migrainosus. Participants completed a text message-based diary for 4 or 12 weeks (depending on treatment), and a detailed headache assessment on a random 20% of headache days. Using this assessment, we determined whether a headache day qualified for migraine or probable migraine, based on the International Classification of Headache Disorders, 3rd edition (ICHD-3). Of 122 enrolled children and adolescents, 106 (86.9%) completed ≥1 detailed headache assessment (438 entries). We found moderate agreement between self-reported and ICHD-derived migraine day, with a Cohen's Kappa of 0.50 (positive predictive value [PPV]: 0.66; negative predictive value [NPV]: 0.85; correlation: 0.51). Allowing for ICHD-derived probable migraine significantly increased PPV (0.66 vs. 0.94; 95% CI: 0.57-0.74 vs. 0.90-0.97), but decreased NPV (0.85 vs. 0.293; CI: 0.77-0.90 vs. 0.199-0.40), Cohen's Kappa (0.50 vs. 0.237; CI: 0. 389-0.60 vs. 0.139-0.352), and correlation (r = 0.51 vs. 0.302; CI: 0.41-0.61 vs. 0.192-0.41). Pain severity (OR: 5.7; CI: 2.39-13.8), photophobia (OR: 4.1; CI: 1.02-16.6), and phonophobia (OR: 7.5; CI: 1.95-29.3) were significantly associated with participants' perception of migraine. We found only moderate agreement between self-reported and ICHD-derived migraine day, suggesting both measures are not equal but may represent overlapping aspects of migraine as a disease. This highlights the difficulty of applying ICHD criteria to individual attacks. We recommend greater methodological transparency in future research to avoid readers conflating both measures.