Psychometric Properties of the Edinburgh Postnatal Depression Scale (EPDS) for Use in Digital Perinatal Mental Health Research: A Longitudinal Study Among Spanish Pregnant and Postpartum Women

Background The Edinburgh Postnatal Depression Scale (EPDS) is the most frequently used paper‐and‐pencil instrument for assessing perinatal depression in digital perinatal mental health research. However, the psychometric properties of the EPDS to be used in this context need to be established, and more evidence is needed regarding its internal factorial structure. Objective To examine the psychometric properties of the EPDS for its use in digital perinatal mental health research. Methods The sample, recruited online, consisted of 1222 pregnant and 1772 postpartum Spanish women. Three follow‐up assessments were conducted over a 6‐month period. Evidence of validity based on internal structure and relationships with other variables was provided. Reliability of the EPDS was evaluated through Cronbach’s alpha (α) and McDonald’s Omega (ω) coefficients. In addition, EPDS item analysis was performed. Results Exploratory factor analysis (EFA) revealed a three‐correlated‐factor model ( anhedonia, depression , and anxiety ) that showed a good data fit by confirmatory factor analysis (CFA) for pregnant (CFI = 0.995; NNFI = 0.993; RMSEA [95% CI] = 0.047 [0.032; 0.062]) and postpartum (CFI = 0.996; NNFI = 0.994; RMSEA [95% CI] = 0.039 [0.027; 0.051]) women. The three correlated‐factor model showed measurement invariance across pregnant and postpartum women. EPDS total score and the three EPDS factors showed positive ( r > 0.500; p < 0.001) correlation with the GAD‐7 ( anxiety symptoms ) and the PTSD‐checklist ( post-traumatic stress disorder symptoms ). Reliability coefficients ( α and ω ) exceeded 0.87 for both the total EPDS score and its three factors. Conclusions The EPDS appears to be a suitable tool for assessing anhedonia, anxiety, and depression symptoms in digital perinatal mental health research. However, there is considerable variation among studies in item loadings and factors across the two‐ and three‐factor models, indicating that more in‐depth mixed‐methods analyses are needed.