Can we identify mothers at risk for postpartum depression in the immediate postpartum period using the Edinburgh Postnatal Depression Scale?

Background: Postpartum depression is a major health issue for many women around the world with well-documented negative health consequences for the mother, child and family. While research has demonstrated the amenability of postpartum depression to treatment, there is preliminary evidence suggesting maternal mood in the immediate postpartum period may be predictive of postpartum depression such that secondary preventive interventions may be implemented. Methods: A population-based sample of 594 mothers completed the Edinburgh Postnatal Depression Scale (EPDS) at 1, 4 and 8 weeks postpartum. The sensitivity, specificity and predictive power of the 1-week EPDS in relation to identifying mothers with elevated EPDS scores at 4 and 8 weeks was determined. The predictive power of the 1-week EPDS was further assessed using odds ratios and receiver operator characteristic (ROC) curves. Results: At 1 week postpartum, 29.5% of mothers scored >9 on the EPDS, decreasing to 23% at 4 weeks and 20.5% at 8 weeks. Using the cut-off score of 9/10, the 1-week EPDS accurately classified 85.4% mothers at 4 weeks and 82.5% mothers at 8 weeks with or without postpartum depression symptomatology. The 1-week EPDS was significantly correlated to the 4-week (r=0.72, P<0.001) and 8-week (r=0.65, P<0.001) EPDS. Mothers with a 1-week EPDS score >9 were 30.3 times more likely at 4 weeks (95% CI = 17.5-42.3) and 19.1 times more likely at 8 weeks (95% CI = 11.0-32.9) to exhibit postpartum depression symptomatology. Limitations: Psychiatric interviews were not completed in collaboration with the EPDS. Conclusion: The EPDS administered in the 1st week postpartum was predictive of maternal mood at 4 and 8 weeks postpartum. To identify mothers at high risk for postpartum depression, health care professionals could consider screening all new mothers in the immediate postpartum period such that secondary preventive interventions may be implemented. (C) 2004 Elsevier B.V. All rights reserved.