Abstract

Objective: To develop and validate a brief screening instrument for postpartum depression in resource-constrained primary care settings. Method: Secondary data analysis of a cohort of 305 mothers (Mdnage = 26) attending well-child check-ups in six primary care centers in Santiago, Chile, answered the Edinburgh Postnatal Depression Scale (EPDS), the 36-Item Short Form Health Survey, and the Mini International Neuropsychiatric Interview depression module. A predictive model for postpartum depression was built using logistic and least absolute shrinkage and selection operator regressions, with bootstrap validation. Results: A three-item version of the EPDS exhibited excellent discriminative capacity (c statistic = 0.95) and showed no significant differences versus the full version of the EPDS (χ2(1) = 1.75, p =.187). The best trade-off between sensitivity (92.86%) and specificity (86.70%) was achieved at a cut-off score of 8/9. Conclusions: The three-item version of the EPDS can save clinicians valuable time, which might potentially improve communication of results to patients.