Associations of predictor variables with place of delivery (unadjusted, accounting for clustering by site) (n = 407).

407 women had non-missing data for both baseline and postpartum follow-up variables.aEstimates account for clustering by site (health facility), using random effects logistic regression.bAs compared to those working in housework, selling things (including fish), or other occupations (combined).cMean anticipated stigma score was computed for women who had non-missing data for at least six of the nine items.dMean perceived community stigma sub-scale scores were computed for women who had no more than one missing item in that sub-scale, and the mean total score was computed for those who do not have more than one item missing on any sub-scale.eAs compared to women who responded “no” to all intimate partner violence (IPV) questions (physical, psychological, or sexual abuse) regarding IPV during pregnancy and postpartum. Women who refused to answer or had missing data were placed in the unknown category.OR, odds ratio.