Effect of emergency obstetric care and proximity to comprehensive facilities on facility-based delivery in Malawi and Haiti.

Proximity of households to comprehensive obstetric care is a key determinant for preventing maternal mortality due to obstetric emergencies. The relationship between proximity to comprehensive care and facility delivery is further complicated by the use of varied methods in measuring facility obstetric capacity-which may misrepresent the real scenario of obstetric care availability in a service environment. We investigated the joint effects of proximity and two emergency obstetric care assessment (EmOC) methods on women's place of delivery in Malawi and Haiti. Household level and health facility data were obtained from the 2013-2018 Demographic and Health Surveys and Service Provision Assessment surveys. Records of women aged 15 to 49 years who had a childbirth in the last 5 years were linked to obstetric facilities within 5km, 10km and 15km from their households using Kernel Density Estimation. Log-binomial models were fitted to estimate the joint effects of proximity to comprehensive facilities on place of delivery and two EmOC methods (1. the facility's recent performance of signal functions only, and 2. a composite index of obstetric care), and whether this varied by urban/rural setting. Proximity to comprehensive facilities was significantly associated with facility delivery in Malawi among women living 5km of a comprehensive facility (using EmOC method 2), in addition, living further (15km) from facilities with high capacity of EmOC was associated with reduced likelihood for facility delivery in urban settings in stratified analyses. In contrast, positive associations were present in Haiti in both urban and rural settings, with the likelihood of facility delivery being higher with greater proximity of women to comprehensive facilities, regardless of methods to define EmOC. Women living within 5km of a comprehensive facility in Haiti were the most likely to deliver in facilities based on EmOC method 1 (APR: 1.81, 95% CI 1.56, 2.09). Findings from Malawi elucidates the relevance of context and suggests the need for research in diverse settings.