Representativeness of obstetric patients who participate in perinatal depression research: findings from the Women's Mental Health and Infants Program (WMHIP) integrated dataset.

The aims of this study were to evaluate the feasibility of integrating archival datasets from depression projects involving pregnant women recruited from obstetric clinics and then assess the representativeness of the integrated dataset. Datasets from six studies were standardized and integrated. Chi-square, t-, and Wilcoxon rank-sum tests were used to compare characteristics between women who completed a depression screening questionnaire (DSQ) and were (1) eligible and ineligible for research participation and (2) eligible women who accepted and declined participation. The integrated dataset comprises 9,112 pregnant women, of whom 71.0 % (n = 6,472) were ineligible for participation because their DSQ scores indicated no-to-minimal depressive symptoms (NDS). Among the 23.9 % (2,176) of women identified as eligible, in part, because their DSQ scores indicated elevated levels of depressive symptoms (EDS), 29.6 % (644) of women participated (P-EDS) and 47.6 % (1,036) of women did not participate (D-EDS). While the NDS and EDS groups were significantly different on almost all variables, the P-EDS and D-EDS groups were significantly different on only a few variables. Compared to the D-EDS group, the P-EDS group was earlier in pregnancy and, on the Edinburgh Postnatal Depression Screen, was more likely to endorse impaired "ability to laugh" and "enjoy oneself", and endorse at greater severity "ability to laugh." It is a reasonable and feasible strategy to integrate thematically similar datasets to increase statistical power. Additionally, typical recruitment strategies for minimal risk perinatal depression research at obstetric clinics, during routine prenatal care visits, appear to produce an externally valid study cohort.