An economic analysis of maternal health care in Zambia

Abstract

This thesis investigates the utilisation of maternal health care in Zambia, where despite being a signatory to the Safe Motherhood Initiative and Millennium Development Goals, which are aimed at improving maternal health, indicators of maternal health continue to perform poorly. The need to understand crucial factors in improving maternal health motivated the current research, especially since there is a dearth of literature in this area in Zambia. The thesis focuses on two aspects of maternal health care: antenatal care (ANC) and facility-based deliveries, to answer two broad questions. Firstly, what factors determine the use of ANC in Zambia? Secondly, to what extent has the abolition of user fees affected facility-based deliveries? An assessment of the factors, which explain the utilisation of ANC in Zambia, using three sets of comparable datasets reveals that, while there are differences in the factors explaining the decision to use ANC and the frequency of visits over time, the decision to seek ANC and the frequency of use is low among the poor and less educated, and there are marked regional differences in utilisation. The most appropriate econometric specification for antenatal visits, according to different performance indicators, was the two-part model, which differs from recent research favouring more complex methodologies. The analysis is further extended through the inclusion of supply-side factors and the examination of individual and community level factors associated with inadequate and non-use of ANC, following the adoption of the focused ANC approach in Zambia. To incorporate the supply side factors, the 2007 Zambia Demographic and Health Survey was linked to administrative and health facility census data using geo-referenced data. To assess the factors associated with (1) the inadequate use of ANC (defined as three or less visits), and (2) the non-use of ANC in the first trimester of pregnancy, we specify two multilevel logistic models. At the individual level, the woman’s employment status, quality of ANC received and the husband’s educational attainment are negatively associated, while parity, the household childcare burden and wealth are positively associated with inadequate utilisation of ANC. Both individual and community level characteristics influence inadequate use and non-use of ANC in the first trimester; however, community level factors are relatively stronger in rural areas. Although ANC is an important facet of maternal care, it occurs before delivery, but does not necessarily provide much information with respect to delivery decisions. Therefore, the thesis investigates delivery decisions, as well, in particular, the effect of user fee removal in rural areas of Zambia on facility-based deliveries. To account for regional differences, we employ a Seemingly Unrelated Regression model incorporating an Interrupted Time Series design. The analysis uses quarterly longitudinal data covering 2003q1-2008q4. When unobserved heterogeneity, spatial dependence and quantitative supply-side factors are controlled for, user fee removal is found to immediately increase aggregate facility-based deliveries, although the national trend was unaffected. Drug availability and the presence of traditional birth attendants also influence facility-based deliveries at the national level, such that, in the short-term, strengthening and improving community-based interventions could increase facility-based deliveries. However, there is significant variation and spatial dependence masked in the aggregate analysis. The results highlight the importance of service quality in promoting facility-based deliveries, and also suggest that social and cultural factors, especially in rural areas, influence the use of health facilities for delivery. These factors are not easily addressed, through an adjustment to the cost of delivery in health facilities. Additionally, we analyse the effect of user fee abolition on the location of childbirth, focussing on deliveries that occur in public health facilities using household survey data. To elicit the causal relationship, we exploit the relative change in fees across health districts within a difference-in-differences framework. Surprisingly, although reductions in home deliveries were observed, as expected, reductions in public health facility-based deliveries were also uncovered, along with increases in deliveries at private health facilities. However, these findings were statistically insignificant; suggesting that the abolition of user fees had little, if any, impact on the choice of location for childbirth. The statistically insignificant, but unexpected, causal effects further suggest that the removal of user fees have unintended consequences, possibly the transference of facility costs to the client, which would deter the utilisation of delivery services. Therefore, abolishing user fees, alone, may not be sufficient to affect changes in outcomes; instead, other efforts, such as improving service quality, could have a greater impact.