A more rational, theory-driven approach to analysing the factor structure of the Edinburgh Postnatal Depression Scale

Abstract

A more rational, theory-driven approach to analysing the factor structure of the Edinburgh Postnatal Depression Scale, Psychiatry Research, http://dx.doi.org/10.1016/j.psychres. 2017.01.059 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. AbstractWe endeavoured to analyze the factor structure of the Edinburgh Postnatal Depression Scale (EPDS) during a screening programme in Hungary, using exploratory (EFA) and confirmatory factor analysis (CFA), testing both previously published models and newly developed theory-driven ones, after a critical analysis of the literature.Between April 2011 and January 2015, a sample of 2,967 pregnant women (between 12th and 30th weeks of gestation) and 714 women 6 weeks after delivery completed the Hungarian version of the EPDS in South-East Hungary. EFAs suggested unidimensionality in both samples. 33 out of 42 previously published models showed good and 6 acceptable fit with our antepartum data in CFAs, whilst 10 of them showed 1 The last two authors contributed equally 2 good and 28 acceptable fit in our postpartum sample. Using multiple fit indices, our theory-driven anhedonia (items 1,2) -anxiety (items 4,5) -low mood (items 8,9) model provided the best fit in the antepartum sample. In the postpartum sample, our theorydriven models were again among the best performing models, including an anhedonia and an anxiety factor together with either a low mood or a suicidal risk factor (items 3,6,10). The EPDS showed moderate within-and between-culture invariability, although this would also need to be re-examined with a theory-driven approach.

Establishing a coherent and replicable measurement model of the Edinburgh Postnatal Depression Scale

“…The findings from each CFA evaluation at each time point and combined for each model are summarised in Table 3 . The models of Tuohy and McVey (2008) , the modified version of this model and the three-factor model of Kozinszky et al. (2017) were consistently observed to offer the best fit to data.…”

“…(2011) 3 10 All data 79.82 32 <0.001 0.06 (0.04–0.07) 0.80 0.99 0.98 Tuohy and McVey (2008) 3 9 All data 34.10 24 0.08 0.03 (0.01–0.05) 0.54 0.99 0.99 Modified three-factor 3 7 All data 18.30 11 0.08 0.04 (0.01–0.07) 0.46 0.99 0.99 Kozinszky et al., (2017) 3 6 All data 6.71 6 0.35 0.02 (0.01–0.06) 0.28 0.99 0.99 Kozinszky et al., (2017) 1 6 All data 168.23 9 <0.001 0.20 (0.17–0.22) 2.10 0.92 0.87 Coates et al., (2016) 3 10 All data 138.71 32 <0.001 0.09 (0.07–0.10) 1.06 0.97 0.96 Cox et al., (1987) 1 10 3 month 145.07 35 <0.001 0.11 (0.10–0.13) 1.16 0.94 0.92 Gollan et al., (2017) 1 7 3 month 55.76 14 <0.001 0.11 (0.08–0.14) 0.98 0.96 0.94 …”

“…(2017) 2 10 6 month 74.64 34 <0.001 0.07 (0.05–0.10) 0.83 0.98 0.97 Reichenheim et al. (2011) 3 10 6 month 61.83 32 0.001 0.07 (0.04–0.09) 0.73 0.98 0.98 Tuohy and McVey (2008) 3 9 6 month 30.94 24 0.16 0.04 (0.01–0.07) 0.53 0.99 0.99 Modified three-factor 3 7 6 month 19.57 11 0.05 0.06 (0.01–0.10) 0.49 0.99 0.99 Kozinszky et al. (2017) 3 6 6 month 8.25 6 0.21 0.04 (0.01–0.10) 0.31 0.99 0.99 Kozinszky et al.…”

“…(2011) ; # Tuohy and McVey (2008) ; *Modified model based on Tuohy and McVey (2008) ; † Coates et al. (2016) ; ^ Kozinszky et al. (2017) ; EPDS-7 ( Gollan et al.…”

“…The seventh model was a three-factor model, based on a critique of the EPDS comprised a 7-item tri-dimensional structure ( Matthey and Agostini, 2017 ), similar to that of Tuohy and McVey (2008) but with potentially ambiguous items removed (items 7 and 10). The eighth model evaluated was the three-factor 6-item model of Kozinszky et al., (2017) comprising anhedonia (items 1 and 2), anxiety (items 4 and 5) and low mood (items 8 and 9). A unidimensional version of this model (model 9) was also evaluated to determine the relative contribution of multi-dimensionality over selection of what might be simply better performing items.…”