Differential Predictors of Postpartum Depression and Anxiety: EPDS Two Factor Structure Construct Validity

Purpose and Background:  The postpartum period is associated with a range of emotional difficulties; however, the focus of attention is commonly given to postpartum depression (PPD). Postpartum anxiety (PPA) is often comorbid with depression but in many cases appears as an independent condition, and is at least as prevalent and disruptive. The Edinburgh Postnatal Depression Scale (EPDS) was originally created as a uni-dimensional scale to screen for postpartum depression (PPD); however evidence suggests that it includes an anxiety sub-scale as well as a depression sub-scale. This study examined the construct validity of the PPD and PPA subscales of the EPDS by examining differential predictors of both subscales in a structural regression model.

Method:  Women (n = 650) were recruited from a maternity ward in Jerusalem, Israel, where they completed a baseline survey providing data on maternal age in years, number of previous pregnancies, family income, religion (1 = non-orthodox; 2 = orthodox), history of depression (1 = yes; 2 = no), and family history of depression (1 = yes; 2 = no).  Women were screened for PPD and PPA at 6 weeks postpartum using the EPDS; higher scores indicate more symptoms of PPD or PPA.  All data were collected in Hebrew.   The baseline characteristics were entered into a structural regression model as predictors of PPD and PPA; the analysis was conducted in Mplus 7.11 using the WLSMV estimator.

Results:  The overall structural regression model fit the data well (RMSEA = .049, 90% CI RMSEA = .040-.056, p RMSEA <= .05 = .581; CFI = .92; TLI = .90).  The EPDS indictors of PPD and PPA were significant and the measurement model fit well; PPD and PPA were moderately correlated (r = .59).  History of depression was a significant predictor of both PPD and PPA (B = -.54, p < .005 and B = -.57, p < .005, respectively); women with a history of depression reported more symptoms of both PPD and PPA.  No other predictors were significant for PPD.  However, income (B = -.07, p < .005) and number of past pregnancies (B = -.06, p < .005) were also significant predictors for PPA.  Women with lower incomes and fewer past pregnancies reported more symptoms of anxiety.

Implications:  Findings show that PPD and PPA have different predictors, suggesting that the EPDS depression and anxiety sub-scales should be considered separately rather than only focusing on the total depression score.  Researchers and clinicians should bear in mind that an elevated total score on the EPDS may be indicative of increased symptoms of anxiety rather than just symptoms of depression.  Further assessment and treatment should be tailored to the specific symptoms of depression and/or anxiety reported in order to achieve the best treatment results. Finally, the generalizability of these findings may be limited; future research should consider the variations in cultural meanings of PPD and PPA in order to determine whether these findings generalize to other languages and cultures.