Inter-regional child mortality, programme efficiency, and throughput: an evaluation of the Ethiopian health extension programme.

Abstract

Background: Despite the remarkable improvement Ethiopia had made in the overall health outcomes, such as reducing under-five child mortality rate, there were substantial variations in the rate of progress across its administrative regions over different periods of time. Moreover, compared to many other developing countries, the progress that Ethiopia made in child mortality reduction remained low, and accounted for three percent of the share of global under-five child deaths in 2015. While the community-based health extension programme contributed to improving health outcomes of the population, such as reducing child mortality, access to and use of basic healthcare services are limited, with significant variations across regions of the country. Much less is known about the factors affecting the inter-regional variations in under-five child mortality: how efficient the health extension programme is in delivering basic healthcare services to its rural societies across regions and the determinant factors affecting the health extension programme beneficiary households' graduation in rural settings of Ethiopia. This study is therefore the first attempt to explore determinants of inter-regional differentials in under-five child mortality, and to evaluate the efficiency and productivity changes of the community-based health extension programme in rural areas of Ethiopia at the national level. This study addresses three specific objectives. These are: i) to examine the determinant factors affecting the inter-regional differentials in under-five child mortality. ii) to evaluate the efficiency and productivity growth (changes) of the community-based health extension programme. iii) to identify the determinant factors influencing the health extension programme beneficiary households' graduation. Methodology: This study employed cross-sectional secondary data from the Ethiopian demographic and health survey, 2016, for a total of 4,200 deaths of under-five children. It also utilised data from the regional health bureaus of Ethiopia, constituting a sample of 1,552 health posts and 4,244 rural households for the years 2013 and 2014. The statistical methods employed include the extended Oaxaca-Blinder decomposition to count data model, Data Envelopment Analysis, Tobit, ordinary least square, and the multiple logistic regressions. Results: The main findings, which addressed the first objective of the study, revealed that the regional differentials in under-five child mortality were due to socio-economic factors (such as mother's age at first birth, antenatal healthcare services, parental education, households' wealth status, and household size), proximate factors (such as child's birth spacing, child's birth order, and size of the child at birth), and environmental factors (such as place of delivery). However, their relative contributions in explaining the regional differences varied significantly within and across groups in the regional comparisons. The main findings, which addressed the second objective of the study, indicated that there was a substantial variation in technical and scale efficiency estimates among health posts, both across the regions and over periods of time. The results indicated that about 5.67 percent of health posts were a variable return to scale (VRS) technical efficient, with an average technical efficiency estimate of 79.6 percent in 2014. Moreover, most of the health posts (91.24 percent) were operating below their optimal scale size, indicating a potential for improving the efficiency of the health extension programme by improving the scale size, the efficiency of the scale as well as technically inefficient health posts. Furthermore, the overall productivity change increased by about 6.7 percent due to the technological progress. In a subsequent study, results of the regression analysis indicated that households' travel distance to the nearest health posts, provision of supportive supervision to the health extension workers, religion, and region of residence of the health extension workers affected the disparities in technical efficiency estimates among the health posts. The main findings addressing the third objective of the study explained the reason behind the rate of graduating households as model households. The results indicated that family size, head of the household head, parental level of education, households' access to the agricultural extension programme, mothers' age, and the professional level of the health extension workers were the major determinant factors affecting the health extension programme beneficiary households' graduation. Conclusion: The explained part of the regional differentials in under-five child mortality was due to differences in socio-economic, proximate, and environmental factors among the regions, with significant differences in the magnitude of the effect. Most of the health posts were operating below their optimal scale size, with substantial variations in technical and scale efficiency estimates, suggesting potential room for improving the efficiency of the health extension programme. Therefore, this study suggests the need for sustained efforts with a due focus on improving households' economic status, maternal education, sustained in-service training and supportive supervision provision to the health extension workers across regions of the country.