(SLA-033) Integration of Two Large Databases for an Observational Cohort Study of Exposure to Galcanezumab during Pregnancy among Women with Migraine

Background: Assessing the safety of galcanezumab during pregnancy in women with migraine requires robust maternal and neonatal outcomes data. Single payer databases such as the Healthcare Integrated Research Database (HIRD) are essential tools for post-marketing drug utilization studies. However, uptake of drugs is often limited in the pregnant population, necessitating the use of multiple data sources to achieve target study size. The Maternal Outcomes Masterset (MOM) is a large, multi-payer database with infant linkage. Integrating single- and multi-payer databases like HIRD and MOM is expected to significantly expand study size. However, there remain challenges to the feasibility of integration.

Objectives: To integrate HIRD and MOM datasets for the evaluation of safety outcomes among galcanezumab-exposed pregnancies. Specific objectives included overcoming data harmonization challenges and projecting cohort size increase to achieve target study size for comparative analysis.

Methods: A list of HIRD patients was sent to the MOM data vendor to remove HIRD patients and prevent duplication of person-time. In both HIRD and MOM, the study included pregnancies exposed to galcanezumab in 28Sept2018-30Sept2024 with at least 6 months of continuous enrollment prior to last menstrual period (LMP) and no pre-LMP diagnosis of epilepsy or severe cardiovascular disease. Age inclusion criteria (15-49 years on LMP) were applied to the HIRD but could not be applied in MOM data due to current privacy restrictions. Progress toward the target study size was assessed by adding the number of qualifying patients between the two data sources.

Results: The target study size for a comparative analysis of safety outcomes of women with migraine during pregnancy exposed to galcanezumab vs. comparator groups is 430. During the study period, 227 HIRD pregnancies exposed to galcanezumab met study criteria, with 151 successfully linked to infants. MOM integration is predicted to increase the cohort size by approximately 200 linked pregnancies. However, comparative analyses using the combined data require strategic alignment across the datasets and clear documentation of limitations.

Conclusions: Integrating single- and multi-payer databases like the HIRD and MOM represents a potential advancement in evaluating the safety of galcanezumab during pregnancy. However, logistical challenges persist, including administrative lead time and differences in privacy restrictions leading to some variables being available in one dataset but unavailable in another. This has implications for how the data will be integrated, such as the ability to run a single propensity score model on combined data, and how the integrated results will be interpretated.