LAUSR

AIMS AND OBJECTIVES We aimed to identify postpartum depression (PPD) trajectories and examine relevant predictors amongst smoking women. BACKGROUND PPD can adversely affect families. Predictors of PPD trajectories amongst smoking women are understudied. DESIGN Longitudinal cohort study. METHODS A cohort of 49 U.S. women (current or ex-smoking) completed the Edinburgh Postnatal Depression Scale from birth to 24 months postpartum. Latent class growth modelling was used to identify PPD trajectories. Predictors of PPD trajectories were identified, adjusting for confounders. Effect modification by prenatal Patient Health Questionnaire (PHQ) depression score was also assessed. STROBE guidelines were followed in reporting results. RESULTS Three PPD trajectories were identified: non-PPD, transient PPD and chronic PPD. In multinomial logistic regression, social support was associated with lower odds of membership in the chronic PPD trajectory compared to non-PPD trajectory: being married or having a partner sharing resources (odds ratio OR = .14 [.02, .85], p-value = .03), greater partner support (OR = .87 [.77, .98], p-value = .02) and greater family/friends support (OR = .53 [.34, .82], p-value = .004). Transient PPD showed no differences with non-PPD on any predictors. In ordinal logistic regression models, social support was associated with lower odds of membership in a more severe PPD depression trajectory when prenatal PHQ depression score was in the low range (being married or having a partner sharing resources: p for effect modification = .06; partner support: p for effect modification = .05; and family/friends support: p for effect modification = .005). RELEVANCE TO CLINICAL PRACTICE Compared to the general population, chronic PPD trajectories were more common amongst smoking women. Social support was an important predictor of more severe PPD trajectories, especially when prenatal depression is low. CONCLUSION Our findings indicated that social support might decrease likelihood of severe PPD trajectories, especially when prenatal depression was low. Relevant predictors of transient PPD remained elusive.