77

This article is a qualitative review of the Edinburgh Postnatal Depression Scale (EPDS) validation studies as a screening tool. A research on MEDLINE, PASCAL-BIOMED and PSYCINFO databases from the period 1987-2004, identified forty-seven papers. The main judgment criterion was the use of a gold-standard for depression diagnosis. Nineteen studies were selected.Sensitivity ranged 0.65 to 1.00, and specificity 0.71 to 0.97. Wide confidence intervals for valuable data did not affect the performances of this self-questionnary, even if the EPDS cut-off scores ranged 8.5 to 13.5 in these studies. Positive predictive value, for a 13% postnatal depression (PND) prevalence, was calculated at 51%. A meta-analytic approach was not realised because of heterogeneous study design and cultural specificities. The use of larger sample size and of united diagnostic criterion for PND would be helpful in future research. Multicentric study would have to include some qualitative and quantitative variables.