Methods: The study sample consisted of 899 women in Wisconsin who received services within a network of evidence-based home visiting programs. Eligible participants were screened by home visitors using a validated depression screen, the Edinburgh Postnatal Depression Scale (EPDS; range 0-30), at least three times across four time periods: (1) first and second trimester, (2) third trimester, (3) postpartum month 1, (4) postpartum months 2-6. Descriptive analyses and paired t-tests were performed for the full sample, rendering mean depression scores, sample proportions above the cutoff (> 12) for likely clinical depression, and mean differences by time. Latent class growth modeling (LCGM) was applied to identify distinct trajectory subgroups while controlling for primiparity (i.e., first pregnancy) and any prior miscarriage.
Results: Results showed a significant decrease over time in mean depression scores (M(t1)=8.08, M(t2)=8.05, M(t3)=7.09, M(t4)=6.82) and sample proportions above the cutoff (t1=24.5%, t2=23.3%, t3=20.7%, t4=21.0%). Paired t-tests revealed significant mean differences between time 1 and 2 (p<.001) and between time 2 and 3 (p<.001). A 6-class LCGM confirmed this trend (BIC=15593.827; Entropy=.852; V-LMRL p<.01). Three classes, representing 84.3% of the sample, had estimated means below the clinical cutoff at baseline. Participants in class 1 (24.9%) presented with consistently low EPDS scores (intercept=2.12, slope=.04, n.s.). Participants in class 2 (34.4%) also had low depression scores at baseline that decreased significantly over time (intercept=6.39, slope=-.49, p<.001). Class 3 (25.0%) presented with a subclinical EPDS score at baseline that decreased over time (intercept=11.01, slope=-.80, p<.001). Among participants in the clinical range at baseline, three groups were classified. Class 4 (8.0%) maintained consistently high scores (intercept=15.24, slope=.27, n.s.), class 5 (2.7%) showed a gradual decrease (intercept=23.33, slope=-1.21, p<.05), and class 6 (5.0%) showed a precipitous decrease (intercept=19.96, slope=-4.08, p<.001), ultimately presenting with a non-clinical mean score (7.91) at time 4. Notably, primiparity and prior miscarriage were significantly associated with being classified to class 6.
Conclusions and Implications: Results showed that there were higher levels of prenatal depression than postpartum depression. Approximately two-thirds (67.1%) of participants belonged to a class whose depression scores improved over time, and no classes were identified that presented with an increasing trajectory. Generalizability of the findings beyond low-income women who received home visiting services is uncertain, though the findings imply that interventions are needed prior to the postpartum period. Home visiting programs may have a special niche in the intervention landscape because there are few comparable strategies that serve families throughout the perinatal period. Broader implications for two-generation intervention programs will be discussed.