Society for Social Work and Research

Sixteenth Annual Conference Research That Makes A Difference: Advancing Practice and Shaping Public Policy
11-15 January 2012 I Grand Hyatt Washington I Washington, DC

16402 Confirmatory Factor Analysis of the Edinburgh Postnatal Depression Scale: A Uni- or Multi-Dimensional Scale?

Thursday, January 12, 2012: 2:00 PM
McPherson Square (Grand Hyatt Washington)
* noted as presenting author
Rena Bina, PhD, Bar Ilan University/Ezer Mizion Organization, Givat Shmuel, Israel
Donna Harrington, PhD, Professor, University of Maryland at Baltimore, Baltimore, MD
The 10-item Edinburgh Postnatal Depression Scale (EPDS) is a postpartum depression (PPD) screening tool. The EPDS Hebrew version is used clinically in several Israeli mother-infant health clinics to detect women with PPD symptomatology; use will be mandatory in all clinics in Israel within a year or two. Although Katzenelson (2004) did some initial validation work on the Hebrew version, no examination of the factor structure has been done. Method. We tested three confirmatory factor analysis (CFA) models with ML estimation for the EPDS Hebrew version using Mplus 6. The sample included 805 Jewish Israeli women who completed telephone interviews 6 weeks postpartum. Their mean age was 29, they had an average of 3.3 children, 98.5% were married, and 12% had postpartum depression symptomatology (EPDS>=9). Results. Model one with all 10 items loading on one latent variable (postpartum depression) did not fit well (n = 805; ÷2 = 316.98, df = 35, p < .005; CFI = .82; RMSEA = .10); all standardized loadings were significant (p < .005) and ranged from .32 (item 10) to .71 (item 8). Examination of the modification indices (MI) suggested allowing items 4 and 5 to covary; adding the error covariance for model two resulted in improved model fit (÷2 = 177.28, df = 34, p < .005; CFI = .91; RMSEA = .07); standardized loadings were all significant (p < .005), ranging from .33 (item 10) to .73 (item 8). Examination of the MI for model two suggested adding a covariance between items 1 and 2, resulting in model three with two error covariances. Model three fit well (÷2 = 131.28, df = 33, p < .005; CFI = .94; RMSEA = .06); standardized loadings were all significant (p < .005), ranging from .34 (item 10) to .73 (item 8); the covariance between items 4 and 5 was .42 and between items 1 and 2 was .25 (p < .005). Discussion. Both covariances in model three make sense conceptually and are consistent with the literature (e.g. Tuohy & McVey, 2008). Items 4 and 5 relate to anxiety (as opposed to depression) and may constitute an anxiety subscale along with item 3. However preliminary work on this subscale has resulted in mixed findings (Phillips, Charles, Sharpe & Matthey, 2004). Items 1 and 2 are the only items that address positive emotions (i.e., “I have been able to laugh and see the funny side of things” and “I have looked forward with enjoyment to things”, respectively), which loaded on a separate factor in some EPDS exploratory factor analysis studies (e.g. Chabrol & Teissedre, 2004). Items 4 and 5 may reflect anxiety, whereas the covariance for items 1 and 2 is probably due to the shared positive nature of the items. Implications. The factor structure of the Hebrew version EPDS should be further examined in order to validate the possible presence of an anxiety sub-scale. Policy makers and social workers need to understand that high scores on the EPDS may reflect anxiety as well as depression symptoms.