(152) To whom results apply? Assessing the impact of extended continuous enrollment inclusion criteria in pharmacoepidemiologic studies of medication use in pregnancy
Assistant Professor Department of Health Policy, VUMC Nashville, United States
Background: Pharmacoepidemiologic research in pregnant populations commonly requires continuous enrollment (CE) during a baseline period, yet there is no standard CE period duration. Few studies examine how requirements for CE during vs. prior to pregnancy affect ascertainment of baseline characteristics, and whether these design considerations differentially affect some groups.
Objectives: To compare baseline and follow-up characteristics between patients with CE >180 days before delivery versus CE >365 days before delivery using data from a mother-child linked birth cohort of Medicaid enrollees.
Methods: We used a retrospective cohort of opioid-naïve pregnant patients with evidence of livebirth in Tennessee (TN) Medicaid (2007-2014) with CE >180 days before delivery. Among this population, we further identified the subgroup of patients with CE >365 days before delivery. Patients were followed from the day after delivery through the earliest of 1) a serious opioid-related event, 2) loss of enrollment prior to 1 year after delivery, or 3) 1-year after delivery. We compared baseline factors measured within 180 days before delivery between two mutually exclusive groups (CE 180-365 days vs. CE >365 days before delivery) using χ-square for categorical variables and t-tests for continuous variables. Secondary analysis of the group with CE >365 days before delivery examined the impact of variable baseline ascertainment periods (180 vs. 365 days) on population characteristics.
Results: We identified 234,791 pregnant patients with CE >180 days before delivery, of which 51.2% also had CE >365 days before delivery. Compared to those with CE >365 days, those with only CE >180 days (180 < CE < 365 days) had a higher proportion of White patients (75% vs. 53%; p< 0.001) and nulliparous delivery (50% vs. 29%; p< 0.001), and a higher mean yearly income ($37,746 vs. $33,977; p< 0.001). Patients with only CE >180 days were also more likely to lose enrollment within the 1 year after delivery (39% vs. 17%; p< 0.001). In the group with CE >365 days, the largest differences in covariate ascertainment using a 365 vs. 180-day baseline period were predelivery opioid (26% vs. 9%; p< 0.001) and antidepressants use (10% vs. 4%; p< 0.001).
Conclusions: Requiring a longer period of CE prior to delivery in a cohort of pregnant patients enrolled in Medicaid substantially changed the demographic composition and disproportionately excluded patients more likely to lose enrollment after delivery. Among the subgroup with CE >365 days, increasing the baseline ascertainment period resulted in differential ascertainment of clinical characteristics, especially for prior medication use. The benefits and consequences of varying CE inclusion criteria should be considered when designing future studies.
