Measuring headache day severity using multiple features in daily diary designs

Abstract

Objective Though migraine is thought of as a symptom complex, symptoms are typically assessed one at a time. For use in clinical research, we developed a composite measure of headache day severity by combining eight well-known symptoms captured in daily diaries. Subjects and Methods Data came from adults with a self-reported diagnosis of migraine (n = 4380) who provided daily diary information assessed using a novel digital platform. Nine observed features theoretically linked to headache day severity were analyzed using latent variable modeling to create a psychometrically robust headache day severity score. Logistic regression was used to assess the cross-sectional relationships of headache day severity scores against an array of clinically-relevant outcomes. Results Participants were largely females (90%), approaching middle age (mean age of 37.3) and living in the United States (49%) or United Kingdom (23%). Findings supported a single latent headache day severity construct based on eight observed headache features. Headache day severity scores were associated with an increased odds of physician visits (Odds ratio[95% CI]: 1.71[1.32–2.21]), emergency department visits (4.12[2.23-7.60]), missed school/work (2.90[2.56–3.29]), missed household work (3.37[3.06–3.72]), and missed other activities (3.29[2.97–3.64]) (p < .0001 for all). Conclusions Modern measurement techniques support a single headache day severity construct that reflects migraine is a symptom complex. The headache day severity scores were associated with external validators and initial visualizations showed how headache day severity scores can be applied broadly in clinical practice and research.

Evaluating therapeutic benefits of ubrogepant via latent class models: A post hoc exploratory analysis of the ACHIEVE I and ACHIEVE II trials

“…As an alternative, one could explore trajectories of latent severity scores. McGinley et al used eight common migraine symptoms to create a latent severity score and demonstrated that it can predict productivity losses and health‐care resource use 40 . The LCM‐based approach did not incorporate the data on participant‐designated MBS in the model but included the presence/absence of each migraine‐associated symptom as input.…”