Are urban children really healthier? Evidence from 47 developing countries
Introduction
On average, child health outcomes are better in urban than in rural areas of developing countries. Understanding the nature and the causes of this rural–urban disparity is essential in contemplating the health consequences of the rapid urbanization taking place throughout the developing world and in targeting resources appropriately to raise population health. Comparison of mean levels of health is not sufficient for these purposes. It ignores variation in health with population characteristics, such as income, that are not necessarily invariant to urbanization and can potentially be used to target resources more effectively than is possible with a simple rural–urban distinction. One objective of this paper is not only to document the magnitude of rural–urban disparities in child nutritional status and mortality across 47 developing countries but also to determine the extent to which these disparities are explained by differences in population characteristics across urban and rural settings. Even if population characteristics were to explain all of the rural–urban difference in child health, targeting health resources on the basis of rural–urban location would still be efficient if there were homogeneity in these characteristics within rural and urban sectors. But the greater is within sector population heterogeneity, the stronger is the argument for allocating resources in relation to characteristics besides rural–urban location. Living standards, for example, obviously do vary within urban settings. In fact, income inequality is typically greater in urban areas than it is in rural areas (Deaton & Drèze, 2002; Kuznets, 1965). Health programs that target the rural population overlook the urban poor who may enjoy little or no health advantage over their rural counterparts. The second objective of the paper is to compare health outcomes for poor urban and rural children and to examine rural–urban differences in the degree of socioeconomic inequality in these outcomes. This will contribute to appraisal of the case for paying greater attention to poor urban populations in the prioritization of health programs.
There is a considerable body of literature documenting the rural–urban disparity in child health outcomes in the developing world. Most of the literature focuses on discrepancies in measures of child nutritional status. This clearly demonstrates that, on average, urban children are better nourished; they are less likely to suffer chronic malnourishment (stunting) and to be severely underweight (Fotso (2006), Fotso (2007); Menon, Ruel, & Morris, 2000; Ruel et al., 1998; Sahn & Stifel, 2003; Smith, Ruel, & Ndiaye, 2005; von Braun, McComb, Fred-Mensah, & Pandya-Lorch, 1993). In the recent literature, less attention has been given to rural–urban differences in child mortality but that which exists shows that urban children face a lower risk of dying before their first, or fifth, birthday (Brockerhoff, 1995; Cai & Chongsuvivatwong, 2006; Cleland, Bicego, & Fegan, 1992; Gould, 1998; Sastry, 1997; Wang, 2003). Table 1 provides a summary of the recent cross-country evidence on the rural–urban gap in child health outcomes in the developing world.
Rural–urban differences in mean outcomes do not reveal the considerable variation in health experiences of children within rural and urban settings. Sahn and Stifel (2003) find that the contribution of the rural–urban gap to total variation in child nutritional status is quite small in 14 Sub-Saharan African countries. Total inequality in children's height-for-age in Sub-Saharan Africa (SSA) appears to be mainly a matter of inequality within urban and rural areas. So, although the rural–urban disparity is large, it is not the primary source of variation in child health. Clearly, populations are not homogenous within rural–urban sectors and one has to take care not just to compare their means.
A difference in mean outcomes certainly does not imply that an urban child can expect to enjoy better health than her otherwise identical counterpart in a rural setting. The disparity may largely derive from differences in population characteristics, such as levels of income and education. The literature suggests that population and community characteristics are important in explaining the rural–urban disparity in child health outcomes (Fotso, 2007; Sastry, 1997). Smith et al. (2005) report significant rural–urban differences in the levels of household proximal and socioeconomic determinants of child nutritional status using Demographic and Household Survey (DHS) data from 36 developing countries. They find very few significant differences across urban and rural settings in the effects of determinants on child nutrition. From this it is concluded that the urban advantage is due to the superior conditions, including behavioral factors such as nurturing practices, rather than differences in the effects of conditions on nutrition. But the authors do not quantify the share of the rural–urban disparity that is explained by differences in conditions.
Despite better average health outcomes in urban areas, there is some evidence of little or no differences in health between rural and urban poor children (Cameron, Kgamphe, Leschner, & Farrant, 1992; WHO, 1993; WRI, UNEP, UNDP, & WB, 1996). The higher mean in urban areas may be simply due to a lower proportion of poor children but it might also be that there is a higher socioeconomic gradient in child health in urban areas (Bitran, Giedion, Valenzuela, & Monkkonen, 2005). Menon et al. (2000) have shown that the socioeconomic gradient in childhood stunting is indeed higher in urban areas of 10 developing countries and Fotso (2006) finds the same for Sub-Saharan African countries. Ruel, Haddad, and Garrett (1999) present a similar finding regarding prevalence of diarrhea in Latin-America. The last column of Table 1 provides a summary of evidence comparing socioeconomic inequality in child health indicators across urban and rural areas.
From the existing evidence it is clear that there is a rural–urban gap in mean child nutritional outcomes but a few studies suggest that this is at least partly explained by differences in levels of proximal and socioeconomic determinants of nutrition. There is also some evidence that while mean child nutritional status is higher in urban areas, socioeconomic inequality is also higher.
This paper presents a comprehensive and consistent analysis of the magnitude and explanation of rural–urban disparities in child health throughout the developing world. It adds to the existing literature by using the most recent data from 47 countries to estimate the size of rural–urban relative risks for both child stunting and mortality and to determine the extent to which these disparities can be accounted for by rural–urban differences in socioeconomic and demographic factors. By also comparing the degree of socioeconomic inequality in child health across rural and urban settings, the paper develops a cohesive argument concerning the nature and the implications of rural–urban differences in the distribution of child health. More specifically, this paper extends the existing literature in five respects. First, it looks at the rural–urban gap in both childhood mortality and stunting. While there is considerable evidence that malnutrition is an informative health indicator in developing countries and a good predictor of mortality (Pelletier, Frongillo, & Habicht, 1993; Pelletier, Frongillo, Schroeder, & Habicht, 1995; Schroeder & Brown, 1994), the magnitude and the explanation of the rural–urban disparities in the two indicators may differ. Harttgen and Misselhorn (2006) show that access to health care has a greater impact on child mortality than on malnutrition. Since rural areas are usually more deprived of health care facilities, this could cause rural–urban mortality differentials to be greater than those in malnutrition. In fact, from a cross-country analysis, Fay, Leipziger, Wodon, and Yepes (2005) find that, after controlling for socioeconomic factors, stunting is negatively associated with the urbanization rate whereas the opposite is true of infant and child mortality. Besides environmental hazards and pollution, a possible explanation could be the higher prevalence of HIV/AIDS in urban, densely populated areas (Dyson, 2003). Differences between urban and rural areas in food supply, including its diversity and security, should reflect more strongly in nutritional indicators than in mortality. Further, urban areas are characterized by a greater dependence on cash income, weaker informal safety nets and greater labor force participation of women (Ruel et al., 1999), which may all impact differently on child malnutrition than mortality.
Second, this paper paints a broad picture of rural–urban disparities in child malnutrition and mortality by using data on 47 developing countries. Malnutrition is measured using the new growth standards that were released by the World Health Organization in April 2006 (WHO, 2006). The new standards adopt a fundamentally prescriptive approach designed to describe how all children should grow rather than merely describing how children grew in a single reference population at a specified time (Garza & de Onis, 2004). For example, the new reference population includes only children from study sites where at least 20% of women are willing to follow breastfeeding recommendations. Use of this new reference population could affect estimates of rural–urban disparities since some of the factors used in predicting potential growth, such as breastfeeding, differ in prevalence between urban and rural locations. This is one of the first studies presenting estimates of nutritional status based upon these new standards.
The third contribution of this paper is to quantify the extent to which the rural–urban gaps in child malnutrition and mortality are explained by differences in population characteristics. Fourth, the paper extends the evidence on socioeconomic inequality within urban and rural areas to a broader set of countries and health indicators and employs concentration indices to summarize inequality across the entire distribution rather than simply comparing extremes as in ratio measures. Finally, this paper pays attention to both relative and absolute rural–urban inequality. As recently demonstrated by Lynch, Smith, Harper, and Bainbridge (2006), relative and absolute inequality are not necessarily explained by the same factors. Whereas most economic and epidemiological research has focused on relative inequalities, policy makers may be most interested in absolute inequality.
Section snippets
Data and methods
Data are from the most recent Demographic Health Surveys (DHS) of all 47 countries for which anthropometric data are available. Table 2 lists all the countries, years of survey and sample sizes.
Nutritional status is measured by a binary indicator of chronic malnourishment, or stunting. A child is considered stunted if its height falls two standard deviations below the median height of children of the same age and gender in a ‘healthy’ reference population. The new Multicentre Growth Reference
Rural–urban disparity
The proportion of children that are stunted and that died before the age of 5 in rural and urban areas as well as the rural–urban ratios in these proportions are given in Table 3. Fig. 1 illustrates the rural–urban relative risks of stunting and under-5 mortality for all 47 developing countries grouped by region. There are significant differences in the rural–urban stunting rates in all but 4 countries (Comoros, Madagascar, Namibia and Uzbekistan). For Comoros and Uzbekistan, this could be
Conclusion and discussion
There are considerable rural–urban differences in average child health outcomes in the entire developing world. The median rural–urban relative risk is 1.4 for both stunting and child mortality but rural–urban disparities in the two indicators are not strongly correlated across countries and regions. The most striking difference between the two is in the LAC region, where the rural–urban relative risk for stunting is more than 1.5 times greater than that for mortality. The magnitude of the
Acknowledgments
The authors are grateful to the Institute for Housing and Urban Development Studies for funding the project on “Urbanization, Health and Health Inequality”, from which this paper derives. We also thank three anonymous referees for their detailed and useful comments that have helped us to improve the paper.
References (50)
Child survival in big cities: The disadvantages of migrants
Social Science & Medicine
(1995)- et al.
Achieving child health related millenium development goals: The role of infrastructure
World Development
(2005) Urban–rural differentials in child malnutrition: Trends and socioeconomic correlates in Sub-Saharan Africa
Health and Place
(2007)African mortality and the new ‘urban penalty’
Health and Place
(1998)- et al.
Are urban poverty and undernutrition growing?
World Development
(1999) - et al.
Some urban facts of life: Implications for research and policy
World Development
(1999) What explains rural–urban differentials in child mortality in Brazil?
Social Science & Medicine
(1997)- et al.
Why is child malnutrition lower in urban than in rural areas? Evidence from 36 developing countries
World Development
(2005) - et al.
Co-coverage of preventive interventions and implications for child-survival strategies: Evidence from national surveys
The Lancet
(2005) Determinants of child mortality in LDCs. Empirical findings from demographic and health surveys
Health Policy
(2003)