Measuring psychosocial stress during pregnancy: a multiple-group confirmatory factor analysis across race/ethnicity among Medicaid-covered pregnant women in the United States

Abstract

Background Differential experiences of psychosocial stress during pregnancy may contribute to racial inequities in adverse pregnancy outcomes in the US. Valid and unbiased measurement scales are needed to assess the effect of psychosocial stress on pregnancy outcomes, however, the numerous modified scales implemented to measure stress are not always validated. Methods The construct validity and measurement invariance of maternal stress among Medicaid-covered pregnant women (N = 1,632) were examined. Model fit estimates of three confirmatory factor analysis (CFA) models were compared to determine the appropriate measurement structure. Multiple-group CFA assessed measurement invariance across Black or African American women (51.7%) and women of all other races. Results Robust estimates of model fit supported a hierarchical CFA model composed of four latent domains of stress. Standardized factor loadings of three of these latent domains—external stress, perceived stress, and enhancers of stress- indicated positive correlations with a second-order latent factor for overall maternal stress, whereas the fourth domain, buffers of stress, had a negative association. Multiple-group CFA demonstrated strong measurement invariance. Conclusions Among Medicaid-covered pregnant women, measures for psychosocial stress were unbiased across two subgroups of maternal race/ethnicity. These findings support the construct validity of overall maternal stress underlying the common variability among four latent domains of stress.