To develop and evaluate a new method for measuring headache severity within daily diary designs. 

Daily diary data play an important role in headache research and clinical trials. Headache frequency and pain severity outcomes are typically used for analyses, but these outcomes fail to fully capture the intricacies of headache as a multisymptomatic condition. The current work applies psychometric modeling to evaluate headache day severity in individuals with and without migraine.

Data came from N1-Headache®, a novel digital platform that tracks headache-related factors of individuals. The analyses focused on nine headache day categorical variables: pain intensity, duration, aura, pulsating/throbbing pain, unilateral pain, activity pain aggravation, nausea/vomiting, photophobia, and phonophobia. Latent variable models (factor analysis [FA] and item response theory [IRT] models) were fitted to one randomly selected headache day for each participant to account for non-independence of repeated measures. 

The analysis sample consisted of 4,489 individuals (mean age 36.3 and 87.9% female) of which 77.2% reported being diagnosed with migraine. FA models supported dropping the unilateral pain item (not informative) and using a bifactor structure with a general headache day severity factor (all 8 items) and a specific phobia factor (photophobia/phonophobia) that captures nuisance residual relations; RMSEA=.04, CFI=.98, TLI=.97. Items were calibrated using IRT, which demonstrated adequate reliability across wide range of severity (.7 reliability from roughly -1.5 to +2 standard deviations around mean). Differential item functioning analyses showed that severity was measured similarly regardless of whether individuals had a migraine diagnosis.

Daily diaries play a critical role in headache research. Findings suggest that researchers and trialists can integrating multiple correlated headache features into a single, robust score. Future research should focus on establishing validity evidence for these severity scores and show their utility in longitudinal applications.