Dissertations and Theses

Date of Degree


Document Type


Degree Name

Doctor of Philosophy (Ph.D.)


Epidemiology and Biostatistics



Committee Members





Subject Categories

Epidemiology | Health Services Research | International Public Health | Maternal and Child Health | Public Health | Women's Health


Obstetric Care; Signal functions; EmOC; Travel distance; Facilities


Background: More than 90% of maternal deaths occur in low-and-middle-income countries (LMICs) and are largely attributed to preventable pregnancy-related causes. Comprehensive emergency obstetric care (EmOC), also known as signal functions, is the most effective life-saving intervention for managing obstetric and newborn emergencies. Health facilities offering delivery services are generally classified as having either comprehensive, basic or less than basic EmOC capacity based on their obstetric resources. Multiple EmOC methods are regularly utilized in characterizing facility obstetric capacity and this introduces inconsistencies that pose significant public health and policy implications for access to delivery care among women within a service environment. In particular, the distribution of comprehensive obstetric facilities within a service environment varies with the use of different EmOC methods to measure their obstetric capacity. These inconsistencies in the distribution of comprehensive facilities ultimately impacts the association between proximity to comprehensive obstetric care and facility utilization for delivery in diverse settings. Hence, consistency in EmOC methods is needed in order to accurately estimate the association between proximity to comprehensive obstetric facilities and women’s place of delivery, as well as to characterize the role of individual and contextual predictors of place of delivery in diverse settings. The specific aims of this dissertation included: to compare the performance of four established EmOC assessment methods across six LMICs in other to identify an optimal EmOC method; examine the joint effect of proximity to comprehensive obstetric care and EmOC methods on women’s place of delivery; and investigate the individual and contextual predictors of facility utilization for delivery.

Methods: This cross-sectional study utilized health facility assessment data from the 2013-2016 Service Provision Assessment surveys (SPA) and household-level data from the 2015-2017 Demographic Health Surveys (DHS) conducted in Malawi, Haiti, Tanzania, Nepal, Bangladesh, and Senegal. For Aim 1: to compare the capacities of health facilities to provide emergency obstetric care using four common EmOC methods across six countries, and identify the best EmOC method that predicts facility volume of deliveries; health facilities offering delivery services across the six countries were classified into comprehensive, basic and less than basic capacities based on each EmOC method, and then Poisson regression models were fit to compare the performance of the EmOC methods in predicting facility volume of deliveries. These EmOC methods included: Method 1—performance of signal function in the past 3 months based on facility self-report; Method 2—the interviewer-observed availability of the facility’s structural capacity to perform the signal functions; Method 3—recent or previous performance of signal functions, and Method 4—a composite index of 53 indicators of obstetric care. For Aim 2: to investigate the joint effect of proximity to comprehensive care with two emergency obstetric care assessment methods on women’s place of delivery in Malawi and Haiti; records of women between 15-49 years of age who had a childbirth in the last 5 years from the DHS survey period were linked to obstetric facilities (from the SPA survey) within 5km, 10km and 15km from their household clusters using Kernel Density Estimation, a geo-spatial technique. Multivariable log-binomial models were fitted to estimate the joint effect of EmOC methods 1 and 4 (listed above) with proximity to comprehensive obstetric facilities on women’s place of delivery, and whether this association varied by geographic segmentation (urban/rural residence). For Aim 3: to examine the individual and contextual predictors of facility-based delivery services in Malawi and Haiti, using the EmOC method that best captures facility obstetric capacity; multilevel logistic regression models were fitted using the linked DHS-SPA data to examine various individual and contextual predictors of facility utilization for delivery care.

Results: Findings from Aim 1 showed that all four EmOC methods were significantly associated with facility volume of deliveries in at least one of the six countries, however, EmOC methods 1 and 4 were associated with facility volume of deliveries in at least 3 countries, and across both comprehensive and basic facilities. Findings from Aim 2 showed that proximity to comprehensive obstetric facilities was significantly associated with place of delivery using both EmOC methods 1 and 4, in Haiti but in Malawi, the association only present in urban settings within 15km of households (APR: 0.53, 95% CI 0.28, 0.98) based on EmOC method 1. Specifically, in Haiti, living within 5km of a comprehensive EmOC facility was significantly associated with a greater likelihood of facility delivery—based on both EmOC method 1 (APR: 1.81, 95% CI 1.56, 2.09) and method 4 (APR: 1.27, 95% CI 1.12, 1.44). Also, living within 15km of a comprehensive EmOC facility was significantly associated with a greater likelihood of facility delivery (only when a basic EmOC facility was available within 5km or 10km), based on both EmOC methods 1(APR: 1.44, 95% CI 1.15, 1.82) and method 4 (APR: 1.24, 95% CI 1.03, 1.48). Overall, the magnitude of association was stronger in rural compared with urban settings. Findings from Aim 3 showed that 92% of women had their most recent childbirth in a health facility in Malawi, compared with 42% in Haiti. The magnitude of the Intra Cluster Correlation in both Malawi (0.27) and Haiti (0.34) were considerable, indicating that context contributed significantly to the variation of household clusters with respect to facility utilization for delivery. In Malawi, after adjusting for both individual and contextual predictors: younger age between 15-24 (aOR: 2.31, 95% CI 1.86, 2.87) and 25-34 (aOR: 1.82, 95% CI 1.49, 2.22), being married (aOR: 1.49, 95% CI 1.09, 2.04), attending up to four or more antenatal visits (aOR: 11.95, 95% CI 7.77, 18.37), and being a Christian (OR: 3.89, 95% CI 1.62, 9.30) or Muslim (aOR: 5.19, 95% CI 2.07, 13.04) remained significantly associated with greater odds of facility utilization for delivery, whereas wealth quintile was associated with lesser odds. In Haiti, younger age between 15-24 (aOR: 1.44, 95% CI 1.14, 1.82) and 25-39 (aOR: 1.55, 95% CI 1.27, 1.89), having health insurance (aOR: 5.91, 95% CI 2.97, 11.77), attending up to four or more antenatal visits (aOR: 5.51, 95% CI 3.66, 8.32), proximity to comprehensive obstetric care at 5km (aOR: 1.84, 95% CI 1.34, 2.61) or 15km (aOR: 2.03, 95% CI 1.23, 3.36) when a basic facility was closer, and urban residence (OR: 1.83, 95%CI 1.36, 2.45) remained significantly associated with facility utilization in Haiti. However, region remained significantly associated with lesser odds of facility utilization in Haiti.

Conclusion: EmOC method 4 emerged as the best method for characterizing facility obstetric capacity based on its consistency in characterizing obstetric facilities, its coverage of multiple domains of obstetric care, and the association of those domains with facility volume of deliveries in all six countries examined. Nevertheless, multiple methods are encouraged as a sensitivity approach to guide stakeholders on what approach most closely reflects the true obstetric capacity in a given setting. The findings further demonstrate that the association between proximity to comprehensive obstetric care and place of delivery differs depending on the EmOC method utilized, as well as the country setting. This finding underscores the relevance of EmOC method and context to this investigation and suggests the need for further research in diverse settings. Proximity to comprehensive obstetric care emerged as an important contextual predictor of facility utilization and should receive critical priority as part of health system strengthening efforts to mitigate maternal mortality across LMICs. Findings from Haiti showing that living near comprehensive facilities was associated with facility delivery only when there were basic facilities nearby—may suggest a stronger preference among women for proximity to facilities over facility EmOC capacity. Hence, interventions are needed to improve health-seeking behaviors for comprehensive obstetric care, as well as to promote proximity in geo-location of comprehensive obstetric facilities.



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.