(119) Factors Associated with Health Plan Disenrollment During Pregnancy Among US Commercially Insured Women

Publication Number: 1114

Background: In studies of drug safety in pregnancy that utilize claims databases, loss to follow-up due to health plan disenrollment during pregnancy may lead to selection bias if pregnancies which are lost are differential with respect to the exposure and outcome under study.

Objectives: To describe the frequency of disenrollment and the characteristics of those who do and do not disenroll during pregnancy among women enrolled in a large United States claims database.

Methods: This study used data from Optum’s Dynamic Assessment of Pregnancies and Infants (DAPI), a proprietary pregnancy identification process within the Optum Research Database, a health insurance claims data source. All pregnancies were required to have 183 days of health plan enrollment prior to the start of the pregnancy, defined as the estimated last menstrual period (LMP). The prevalence of select demographics, comorbidities, and previous adverse pregnancy outcomes (spontaneous abortion [SA], stillbirth, and preterm birth) were assessed and compared between those with an observed pregnancy outcome and those who disenrolled during the pregnancy (prior to an observed outcome or at 42 weeks following LMP).

Results: Between 01 January 2016 and 31 December 2021, a total of 713,117 pregnancies with 183 days of pre-LMP enrollment were observed. Of these, 592,708 (83.1%) had an observed pregnancy outcome, while 88,962 (12.5%) disenrolled during pregnancy (the remainder were enrolled through 42 weeks without an observed pregnancy outcome). Age was skewed younger for those who disenrolled during pregnancy (47.4% of those who disenrolled were < 30 years at LMP as compared to 35.4% for those with an observed outcome). Pregnancies with an observed outcome were enrolled for a longer period of time prior to pregnancy (median days of prior enrollment was 597 for those who disenrolled as compared to 759 for those with an observed outcome). There were few differences in comorbidities between the two groups: the prevalence of diabetes, obesity and hypertension all differed by < 1%. Of prior adverse pregnancy outcomes, only prevalence of prior SA differed slightly, with 6.5% and 8.2% among disenrolled and completed pregnancies, respectively.

Conclusions: Pregnancies which disenrolled during the pregnancy were slightly younger, had less pre-LMP enrollment, and may have had slightly lower prevalence of previous SA than those with an observed outcome, but did not greatly differ along other select pre-LMP comorbidities. Future exploration will assess early-pregnancy comorbidities (i.e., infection during the first trimester).