Antidepressant adherence using group-based trajectory modeling among postpartum women with Texas Medicaid

Mental health conditions are among the most frequent underlying causes of pregnancy-related death, and antidepressants may have a positive impact. However, adherence is suboptimal, and little is known regarding antidepressant adherence trajectories among postpartum women in the United States.

To describe antidepressant use among postpartum women with Texas Medicaid and determine factors associated with adherence trajectories.

This retrospective analysis of Texas Medicaid claims (January 1, 2018, to June 30, 2022) included women aged 12 to 55 years with at least 1 delivery, who were continuously enrolled 84 days before and 12 months after delivery, and who received an antidepressant within 90 days after delivery. The index date was the first dispensing of an antidepressant after delivery. The dependent variable was antidepressant adherence, defined as the proportion of days covered (PDC) and measured in 30-day increments for 270 days after antidepressant initiation. The independent variables were guided by the Andersen Behavioral Model and included predisposing (age and race and ethnicity), enabling (urbanicity, prenatal care, and postpartum care), and need (baseline depression/anxiety, baseline substance use disorder [SUD], cesarean delivery, preterm birth, and pregnancy complications) factors. Group-based trajectory modeling (GBTM) was used to identify antidepressant adherence trajectory groups. Multinomial logistic regression was used to identify factors associated with adherence trajectory group membership.

The included patients (N = 15,667) had a mean ± SD age of 27.4 ± 5.9 years, and 41.7% were White. Most resided in urban counties (78.0%) and had 6.4 ± 3.5 prenatal visits, 3.1 ± 2.8 postpartum visits, and 1.4 ± 0.9 pregnancy complications. Nearly half (49.8%) had baseline depression/anxiety, 17.2% had baseline SUD, 37.4% had cesarean delivery, and 13.9% had preterm birth. At 270 days after antidepressant initiation, mean ± SD adherence was 43.9 ± 29.5, and the adherence rate (PDC ≥ 80) was 15.9%. During the 270 days follow-up, mean ± SD persistence without a 30-day gap was 103 ± 85.2 days, and the persistence rate (proportion persisting 180 days without a 30-day gap) was 22.1%. GBTM revealed 5 membership groups: consistent high (19.0%), fluctuating (22.5%), slowly decreasing (13.3%), and rapidly decreasing (21.8%) adherence and early and consistent nonadherence (23.4%). Patterns emerged with decreasing adherence at 2, 3, and 6 months after initiation. Increasing age, non-Black race, urban residence, increasing postpartum care visits, and baseline depression/anxiety were associated with the consistent high-adherence trajectory compared with most lower-adherence trajectories. However, baseline SUD and preterm birth were associated with membership in the less-adherent compared with the consistently adherent trajectory.

Overall adherence and persistence were suboptimal, and GBTM revealed unique patterns of postpartum antidepressant adherence behaviors. To help impact maternal morbidity and mortality, adherence interventions should be tailored to women who are younger, Black, live in rural counties, have SUD, or had a preterm birth.