Determinants of child mortality in LDCs: Empirical findings from demographic and health surveys
Introduction
To improve health outcomes for poor people in low income countries, efforts have been directed in two areas. First, improving the understanding of key determinants of health outcomes and identifying the principle causes of health gap between the poor and the better off. Second, translating empirical findings into policy interventions as reflected in the recent development strategies supported by the international agencies [1]. Strong advocacy for defining poverty in terms of human development and broadening of the traditional income/consumption definition of poverty has renewed the debate on health inequality. The World Bank (2000) compilation of the socioeconomic differences in health, nutrition, and population based on the Demographic and Health Survey (DHS) data, serves as the most comprehensive set of indicators on health inequality. Using survey data, Ref. [2] reviews trends in health inequality and identifies the causes of health inequalities using data from both developed and developing countries, and proposes approaches for evaluating the impact of anti-inequality polices.
However, to better understand the determinants of health outcomes, it is essential to measure health indicators and health inequality using reliable data sources. A major concern in deriving policy implications from evidence based on cross-country studies is the reliability and comparability of data sources. Ref. [3] has critically reviewed the potential problems associated with cross-county data. This problem is particularly acute in the estimates of child mortality as they are sensitive to data sources and estimation methods. Ref. [4] study on child mortality estimates complied by the United Nations shows substantial discrepancies among different estimates for the same county and same period, depending on the data source and the estimation method used. Therefore, to provide credible empirical evidence on health issues using cross county data efforts should focus on data comparability. In this regard, demographic and health surveys conducted in over 60 low-income countries are the superior data source. The DHS surveys use the same methodology to estimate health and other socioeconomic indicators and are comparable across countries and over time. Therefore, empirical studies based on the DHS data are expected to generate more reliable results (Table A1).
Translating research findings into operational priorities requires designing policies with strong poverty focus. These interventions should be effective in improving the average level of health (efficiency dimension) and in narrowing health inequalities (equity dimension). In reality, policy design needs to take account of the trade-offs between efficiency and equity concerns. However, to provide more informative policy recommendations, empirical analysis of health needs to be conducted at a disaggregate level, by socio-economic group, or by geographic location, as data permits. The emphasis of rural/urban separation is particularly useful from the policy perspective as geographical distinction is often more useful as targeting indicator than the income quintile. Also household access to safe water and sanitation, infrastructure, and health care, often varies sharply between urban and rural areas. Given that the poor are mainly concentrated in rural areas, it is likely that the determinants of health differ between the rural and urban population.
Another important health issue concerns the health outcomes of poor neighborhoods in urban areas [5]. A growing number of empirical studies highlight the importance of neighborhood effects on individual health among developed and developing countries.1 Among low-income countries, findings from the China National Child Health Survey [6] show that in urban areas, children living in neighborhoods with poor access to flush toilets have significantly high mortality risks, controlling for household socio-environmental conditions. Similarly, in urban Peru, children who live in neighborhoods with poor access to sanitation have low nutritional status, measured by height for age [7]. There also exists some evidence on widening disparity in urban access to basic services in developing countries. Hence, social benefits from public investment in poor neighborhoods in urban areas are likely to be high, based on efficiency and equity criteria. Unfortunately, the DHS data do not allow the construction of more disaggregate indicators on child mortality in poor urban neighborhoods for analyzing health outcomes in poor urban localities.
This study aims to identify key determinants of health outcomes in poor countries at the national level, and for rural and urban areas separately. We focus on two health indicators, infant mortality rate (IMR) and under-five mortality rate (U5MR). The primary data sources are the DHS and World Development Indicators (WDI). There exists a large body of empirical literature that focuses on the determinants of health outcomes based on data sources of various forms. These include: (1) cross-country data sources [4]2[8], [9], [10]3[11]) (2) cross-region data for a given country [12], [13]; and (3) household-level surveys including DHS and Fertility surveys [2], [6], [14], [15], [16]. In comparison to earlier cross-country studies, the main contributions of this study are: (1) use of the improved data source on health from DHS; (2) investigation of health determinants both at the national level and disaggregated by urban and rural location; and (3) application of the regression estimates to an effectiveness analysis which ranks alternative policy interventions based on their health impacts.
This paper is organized as follows. Section 2 provides an overview of health outcomes in low-income countries. In Section 3, we discuss data issues and summarize the data sources used in this study. Section 4 focuses on the issues relating to estimation methods. Section 5 presents the main findings. Section 6 illustrates the application of the results to effectiveness analysis. Section 7 concludes.
Section snippets
Patterns of health outcomes in LDCs
To capture the general patterns of health outcomes using DHS data, we focus on two measures: level of healthiness and inequality in health. Child mortality rates are generally regarded as the principal measures of country-level health status,4 although more comprehensive measures would include indicators measuring morbidity. The latter measures tend to be less reliable and in most cases, are
Why use DHS?
Data comparability is particularly important in conducting cross-county analysis of health outcomes, such as child mortality rates, which are sensitive to data sources and estimation methods.7
Estimation
To estimate the health determination model, we begin with a model which specifies health outcomes as a function of key groups of variables. Our choice of explanatory variables are based both on economic theory and empirical evidence from earlier work in this area. These include: (1) income; (2) social and environmental indicators, such as female education, access to sanitation, and access to safe water; (3) policy variables such as the share of public health expenditure in GDP; and (4)
Results
We first investigate the simple bivariate association between mortality rates and all potential explanatory variables. The correlation matrices are summarized in Table A2a–c. At the national level, variables which are highly correlated with mortality rates, ranked in descending order, include access to electricity, asset index, GDP per capita, access to piped water, access to sanitation, and female secondary education. However, the ranking is different at the regional level. In the urban data,
Effectiveness analysis
Effectiveness analysis is useful for ranking policy options. In principle we can apply the estimated coefficients from the model to an effectiveness analysis. The estimated coefficients provide measures of net impact of each intervention (which corresponds to each explanatory variable), keeping all other impacts constant. In turn, the inverse of the estimates, which we label as the effectiveness coefficients, are useful for comparing alternative interventions. They provide a measure of
Conclusion
The findings of this study on child mortality using cross-country DHS data consolidate results from earlier studies and add new evidence. We find evidence indicating that health interventions implemented in the past decade may not have been as effective as intended in reaching the poor at the national level. This is based on two observations from the global patterns of health outcomes cross 60 low-income countries: (1) there exists a negative association between the level and inequality in
Acknowledgements
The author gratefully acknowledges the financial support from Swedish International Development Cooperation Agency (SIDA), TF024884. I also thank Julia Bucknall, Jan Bojö, Katherine Bolt, David Coady, Deon Filmer, Marianne Fay, Kirk Hamilton, Janet Hohnen, Jenny Lanjouw, Peter Lanjouw, Rama Chandra Reddy, Stefano Pagiola, Priya Shyamsundar, Jonathan Wadsworth, Adam Wagstaff and the referees for their useful discussions and comments on this paper. The findings, interpretations, and conclusions
References (46)
Data base for development analysis: an overview
Journal of Development Economics
(1994)- et al.
The impact of public spending on health: does money matter?
Social Science and Medicine
(1999) Health inequalities and the health of the poor: what do we know? What can we do?
Bulletin of the World Health Organization
(2000)- Wagstaff A. Inequalities in Health in Developing Countries: Swimming Against the Tide? Policy Research working paper,...
- et al.
Multilevel analysis of neighborhood socioeconomic context and health outcomes: a critical review
Journal of Epidemiology Community Health
(2001) - Wang L, Hughes G, Fan X. Determinants of child mortality and morbidity: Findings from the China National Health Survey,...
- Alderman H, Hentschel J, Dabates R. With the help of one's neighbors. World Bank policy working paper 2627,...
- et al.
Wealthier is healthier
Journal of Human Resources
(1996) - Filmer D, King EM, Pritchett L. Gender Disparity in South Asia: Comparisons Between and Within Countries. Policy...
Factors associated with trends in infant and child mortality in developing countries during the 1990s
Bulletin of the World health Organization
(2000)
Mortality, fertility and gender bias in India
Population and Development Review
Does piped water improve child health in poor families? Propensity score matching estimates for rural India
Journal of Econometrics
Reducing child mortality in India in the new Millennium
Bulletin of the World Health Organization
Combining census and survey data to trace the spatial dimensions of poverty: a case study of Ecuador
The World Bank Economic Review
Estimating wealth effects without expenditure data-or tears: an application to educational enrollments in states of India
Demography
Cited by (159)
Energy poverty and health expenditure: Evidence from Ghana
2021, Energy EconomicsCitation Excerpt :For detail results on the headcount ratio and the average intensity of multidimensional energy perversity refer to Appendix II, Table B Multiple indicators are adopted to measure health outcomes, consistent with extant literature (Awaworyi Churchill et al., 2017; Benefo and Schultz, 1996; Kim and Yang, 2011; Wang, 2003). Studies have shown that poorer households tend to spend higher proportions of their income on health-related expenses relative to their counterparts in higher income groups.
The long-run effects of health aid in low-income countries
2023, Journal of Public Health in AfricaON THE EFFECTS OF ACCESS TO ELECTRICITY ON HEALTH CAPITAL ACCUMULATION IN SUB-SAHARAN AFRICA
2023, Journal of Energy and Development