OBJECTIVES: A cost-effectiveness model (CEM) for fremanezumab in migraine prevention has been developed from a Japanese public healthcare payer perspective. Our objective of this study was to assess the performance of three longitudinal regression models in fitting the monthly migraine days (MMD) distribution and to inform health-state distributions based on data from Japanese-Korean trials.

METHODS: The health states in the model are defined by the number of MMD, ranging from 0 to 28. Data from three randomized controlled Japanese-Korean trials (406-102-00001, 406-102-00002 and 406-102-00003) were used to estimate parametric distributions for MMD. Longitudinal models were fitted using generalized additive models for location, scale and shape (GAMLSS). Patients were separated into two groups, episodic migraine (EM: < 15 headache days per month including ≥ 4 migraine days) and chronic migraine ≥ 15 headache days per month including ≥ 8 migraine days), sorted by baseline headache measurements. For both groups, three different statistical distributions to describe the MMD were fitted: zero inflated beta-binomial (ZIBB), zero inflated negative binomial (ZINBI) and zero adjusted gamma distribution (ZAGA). All models underwent a stepwise Akaike information criterion (AIC) optimization and visual inspection to identify the model that best fitted the data.

RESULTS: In total, 1,495 observations in EM patients and 2,431 observations in CM patients from month 1 to month 3 were included in this study. The AIC and visual fit showed that the ZIBB fit the data best. The distribution shifts between fremanezumab and placebo demonstrated fremanezumab’s efficacy at reducing MMD for both EM and CM patient groups.

CONCLUSIONS: Health state distributions in the CEM were established by fitting longitudinal models using GAMLSS on 3,834 clinical trial observations. Both AIC and visual fit inspections revealed the ZIBB model as the best performing distribution. Therefore, demonstrating fremanezumab’s efficacy in both EM and CM patient groups.