Assessing the Temporal and Regional Differences in the Relationships between Infant and Child Mortality and Urban Slum Prevalence in Less Developed Countries,1990

Abstract

Drawing from various bodies of social scientific literature and research, the authors assess the extent to which infant and child mortality rates in less developed countries are impacted by the percentage of domestic populations living in urban slum conditions. Results of two-way fixed effects panel model estimates of 80 less developed countries from 1990 to 2005 indicate that growth in the percentage of populations living in urban slum conditions positively affects both forms of mortality rate. The effects, moreover, are much more pronounced for African countries than for less developed countries in Latin America and Asia and moderately larger for the Asian nations than those in Latin America. Additional findings suggest that the magnitude of the effect of urban slum prevalence on infant and child mortality increased through time for the African countries, but not for the Latin American and Asian countries in the study.

Introduction

The growth of urban slums in the less developed countries is a structural trend producing concentrated disadvantage recognisable in the overcrowding and sub-standard living conditions enveloping nearly one billion people world-wide (Vlahov et al., 2007). Their historically unprecedented rise threatens to undercut the presumed public health advantages of urban life in ways that many researchers have yet fully to consider. Indeed, world-wide, the aggregate urban slum population grew 39 per cent over the period 1990–2005 (UN-HABITAT, 2008).

The remarkable growth of urban slums in the developing countries in recent decades is a reflection of the increasing urbanisation of poverty; such reorganisation gives form to evolving “risk spaces” (Fitzpatrick and LaGory, 2000, p. 12) or areas in which a sub-population is disproportionately subject to a myriad of hazards relative to other segments of society. In stark contrast to the generally beneficial health consequences of urban social organisation, urban slum conditions are increasingly characterised by an observable ‘urban penalty’ (for example, Davis, 2006). Indeed, urban slum areas often exhibit poorer health outcomes, lower life expectancy rates, lower levels of education and diminished economic opportunities relative to non-slum urban populations (UN-HABITAT, 2006). The assertion that urban infants and young children are healthier than young children and infants in rural areas is increasingly in doubt, moreover, given the deteriorating living conditions of many urban settlements (Bartlett, 2003). Child and infant malnutrition within urban slums is often comparable with that observed in rural areas and the incidence of waterborne diseases and respiratory illnesses is generally higher (UN-HABITAT, 2006).

In this study, we build on recent cross-national research that considers the public health impacts of urban slum prevalence in less developed countries (for example, Jorgenson and Rice, 2010; Jorgenson et al., 2010; Rice and Steinkopf-Rice, 2009). In particular, we employ rigorous longitudinal model estimation techniques to assess: the extent to which the effect of urban slum prevalence on child and infant mortality rates varies by macro region; whether the magnitude of the effect of slum prevalence on both mortality outcomes increases or decreases through time; and, whether potential temporal changes in such associations vary by macro region. Like recent studies, we examine these relationships for a broad sample of less developed countries for the 1990 to 2005 period, and we also include relevant statistical controls, including fertility rates, economic development and world-economic integration. Here, macro region refers to continent-level regions and the nations in this study are located in the macro regions of Asia, Latin America (i.e. Central and South America) and Africa. Looking ahead, we expect to observe positive statistical associations between both mortality outcomes and urban slum prevalence, and we expect to observe differences by macro region with African nations exhibiting relatively stronger relationships between the mortality outcomes and urban slum prevalence, and increasingly so relative to less developed countries in Latin America and Asia. The consideration of temporal and regional differences in the relationships between infant and under-five mortality and urban slum prevalence, coupled with the employment of more rigorous model estimation techniques, are two key contributions of this work, which increase our collective understanding of the ways in which urban conditions in slum contexts impact the health and well-being of the most vulnerable segments of populations.

Literature Review

Drawing from social epidemiology as an overarching meta theoretical framework,¹ we argue that the magnitude and depth of urban slum conditions in the less developed countries are now so pronounced as to constitute a key structural characteristic shaping inequalities in health and illness viewed from a macro comparative, cross-national scale. The built urban environment is not simply a reflection of or container for social organisational patterns but a key dimension shaping variant risk spaces and the social production of morbidity and mortality—particularly as it concerns the well-being of the most vulnerable members of a population: infants and young children. Infant and children under-five mortality rates in the developing countries, in turn, are not simply the consequence of individual-level risk factors, as suggested by the dominant biomedical model, but are also socially produced within the contextual, sub-standard living conditions characteristic of urban slums. Disparities in health and illness often follow from contextual social determinants shaping variance in risk encountered by differing segments of a population. Such patterns, in turn, constitute “biological reflections of social fault lines” whereby disease distribution is forged through relative power, privilege and inequality (Farmer, 1999, p. 5; Krieger, 2001a; Krieger and Zierler, 1996).

The growth of urban slums, of note, is a trend that has gone largely unnoticed by social scientists employing quantitative, macro comparative research approaches examining health disparities in the less developed countries (for exceptions, see Jorgenson and Rice, 2010; Rice and Steinkopf-Rice, 2009);² this is problematic as managing urban growth in a manner that broadly capitalises on the advantages of urbanisation, while minimising the liabilities, is an increasingly important development objective (Cohen, 2006). Evidence that urban slum prevalence has an impact on population-level measures of mortality forces consideration of the need for preventative in addition to simply curative health care efforts in the developing countries. A preventative focus entails the evaluation of the underlying social, built and biophysical environmental factors shaping morbidity and mortality, whereas a curative approach emphasises specific medical interventions employed to combat morbidity as it arises (Ehiri and Prowse, 1999; Hill and Pebley, 1989). A long-standing debate concerns the appropriate policies to pursue within the developing countries, given the scarce resources available at any given point in time. As Ehiri and Prowse (1999) note, the curative approach has been successful in improving public health in the developing countries, but insufficient consideration of broader contextual dynamics makes it is difficult to sustain progress over the long term.

Despite the established medical science literature substantiating the risks to infant health of urban slum living conditions, very little macro comparative research focuses on the extent to which inadequate built urban environments impact public health in the developing countries. We argue that this is potentially problematic as conceptualising the influence of urban slum conditions on infant and under-five mortality rates can contribute to the identification of the upstream, population-level challenges to public health improvement in the developing countries. As McKinlay (1981) notes, a majority of resources and attention devoted to public health concerns are applied downstream or in reference to problem-solving interventions designed to address various, and shifting, issues. Often the real and more enduring problems exist upstream in terms of access to and adequacy of health care provisioning and the social and environmental context in which different segments of a population reside, thus shaping differential exposure to risk (McKinlay 1981). Arguably, health and illness are shaped, in part, through the suitability of the built urban environment—a dynamic not reducible to individual-level biological and behavioural risk factors nor adequately addressed through downstream medical interventions alone.

Urban Slums as Built Environments and the Social Production of Mortality

The built environment consists of the “tangible settings which people create for repeated use” (Dunlap et al., 2001, p. 1) and “that part of the environment constructed by human intention and effort” (Kilmartin, 2001, p. 167). The inadequate built urban environment arguably has a direct, although not deterministic, influence on health disparities that is not synonymous with or reducible to invocations of ‘urbanisation’ or ‘poverty’. It is an expression of prevailing social and economic organisation and, in turn, the social production of uneven health and illness. Socioeconomic processes contribute to the formation of urban slum conditions, but it is the dilapidated, semi-permanent built urban environment in which inequities in health and illness are increasingly enacted. This proposition, in turn, rests upon the meta theoretical assumption that the biophysical and built environments have cognitive, behavioural and physiological impacts discernable at the individual and population levels (Dunlap and Catton, 1979).

Poverty, overcrowding, malnutrition, insufficient garbage disposal, lack of adequate water drainage, and unsafe drinking water and sanitation coalesce around the social organisation of marginalised populations in urban slums. The inadequate built urban environment within many developing countries therefore constitutes a key barrier to progressive social well-being and even a catalyst of retrogression; this may be particularly the case for children (Bartlett, 2003; Satterthwaite, 1993). The five illnesses at the root of a majority of infant and child deaths in the developing countries are pneumonia, diarrhea, malaria, measles and HIV/AIDS (UN-HABITAT, 2007). Each is prevalent in many urban slums due to substandard living conditions and overcrowding (UN-HABITAT, 2007). Inadequate access to clean water and sanitation, in particular, are a direct cause of a substantial proportion of deaths of infants annually (UNDP, 2006). Poor water quality and quantity and inadequate sanitation are linked to a number of waterborne and water-washed diseases (UNDP, 2006).

Greater morbidity and mortality among urban slum residents, especially infants and children, is not simply the consequence of household-level deficiencies but also includes health issues arising within the context of the broader slum settlement (Agarwal and Taneja, 2005; Awasthi and Agarwal, 2003; Bartlett, 2003). Inadequate water drainage and waste removal often create areas of contamination extending throughout the surrounding community (Bartlett, 2003); many slums lack safe places for children to play outdoors (Bartlett, 2003; Satterthwaite, 1993); and a litany of indoor and outdoor chemical pollutants are frequently encountered in low-income urban areas that compromise the health of children (Satterthwaite, 1993). Although slums often border and even roughly encircle urban areas, they are in general socially, politically and economically isolated from the broader urban setting and their residents lack access to many formal institutions in society (UN-HABITAT, 2003b). There are frequently considerable barriers to accessing quality health care and emergency services for slum residents (Sclar et al., 2005).

Micro- and meso-level medical science research indeed illustrates that infants residing in urban slums are subject to a litany of diseases. Neonatal mortality, or death within the first 28 days, is commonplace in many urban slums and is generally preceded by sepsis, perinatal asphyxia and prematurity (Fernandez et al., 2003; Vaid et al., 2007). Beyond the neonatal period, infants residing in urban slums frequently die of diarrheal disease and respiratory infections (Vaid et al., 2007); high rates of diarrheal disease, in particular, are a stark reflection of the lack of clean drinking water and adequate sanitation facilities (Fotso et al., 2007; Vaid et al., 2007).

Home births and births not accompanied by a trained medical professional are also commonplace in many urban slums (Fernandez et al., 2003; Gulati and Jaswal, 1998; Hoque and Selwyn, 1996; Rahi et al., 2006); this situation contributes to late recognition of neonatal illness, inadequate antenatal care and delays in seeking appropriate medical services (Fernandez et al., 2003). Further, many slum children are malnourished, increasing their susceptibility to illness (Bartlett, 2003; Ghosh and Shah, 2004; Wagstaff et al., 2004). Research illustrates that children living in urban slums in India are more malnourished than non-slum urban and even rural children (Ghosh and Shah, 2004). Malnourishment among mothers, in addition, contributes to low-birth-weight neonates and, therefore, is a key risk factor for increased infant death (Fernandez et al., 2003).

Macro-regional and Temporal Variation: Mortality and Urban Slum Prevalence in Africa

Over the period 1990 to 2005, the aggregate urban slum population on the African continent increased from 122 million to over 220 million (UN-HABITAT, 2008). This change corresponds to a 79 per cent increase in 15 years, a relatively short time-frame. Overall urban slum growth outside the African continent was a more modest 31 per cent over the same time-period (UN-HABITAT, 2008). Nations in Africa exhibit some of the highest infant and under-five mortality rates of any region in the world.³ However, since the 1990s, reductions in under-five and infant mortality within many African countries have stalled and even reversed course (Mogford, 2004).

We anticipate that the impact of urban slum prevalence on infant and child mortality is more pronounced within African countries relative to less developed countries within other macro regions (see also Jorgenson and Rice, 2010). This disproportionate effect is arguably derived from both the rapidity of urban slum growth in Africa between 1990 and 2005 and the magnitude of deprivation that often characterises urban slums in and around African cities. We also anticipate that the magnitude of the effect of urban slum prevalence on both mortality outcomes could have increased through time for nations in this macro region, relative to nations in Asia and Latin America. According to UN-HABITAT (2007), African slums, particularly sub-Saharan, are generally characterised by the greatest deprivation in terms of access to basic services such as sanitation, drinking water, living space and durable construction materials. We now turn to the cross-national panel analyses, where we assess the effects of urban slum growth on infant and child mortality in less developed countries as well as the anticipated regional and temporal variations in such relationships.

The Analyses

The Dataset

We analyse a cross-national panel dataset consisting of countries for which data are available for the two dependent variables and all independent variables included in the analyses. These countries are considered as less developed, meaning that they all fall below the upper quartile of the World Bank’s (2007) income classification of nations. At the time of this study, the key independent variable is only available for less developed countries for the years 1990 and 2005, thereby restricting the analyses to such countries and respective yearly observations. In particular, the dataset consists of two observations each for 80 countries (i.e. observations for 1990 and 2005), the latter of which we list in Table 1. We arrange the listing of nations by the macro-regional categories of Latin America (22 countries), Asia (20 countries) and Africa (38 countries). Given our limited sample size and degrees of freedom, we limit the number of predictors in any reported model to no more than 12.

Table 1.

Countries included in the analyses

Africa	Asia	LatinAmerica
Algeria	Bangladesh	Argentina
Angola	Bhutan	Belize
Benin	China	Bolivia
Botswana	India	Brazil
Burkina Faso	Indonesia	Chile
Burundi	Iran	Colombia
Cameroon	Jordan	Costa Rica
Chad	Laos	Dominican Republic
Congo	Lebanon	Ecuador
Congo, Democratic Republic	Malaysia	El Salvador
Egypt	Mongolia	Guatemala
Ethiopia	Nepal	Guyana
Gabon	Pakistan	Honduras
Gambia	Philippines	Jamaica
Ghana	Sri Lanka	Mexico
Guinea	Syrian Arab Republic	Nicaragua
Guinea-Bissau	Thailand	Panama
Kenya	Turkey	Paraguay
Lesotho	Viet Nam	Peru
Madagascar	Yemen	Suriname
Malawi		Uruguay
Mali		Venezuela
Mauritania
Morocco
Mozambique
Namibia
Niger
Nigeria
Rwanda
Senegal
Sierra Leone
South Africa
Tanzania
Togo
Tunisia
Uganda
Zambia
Zimbabwe

Model Estimation Technique

We employ the ‘extreg’ suite of commands in Stata (version 9) to estimate linear fixed effects panel regression models. This approach is one of the most commonly used longitudinal methods in the comparative social sciences because it addresses the problem of heterogeneity bias (see Halaby, 2004). Heterogeneity bias in this context refers to the confounding effect of unmeasured time-invariant variables that are omitted from the regression models. To correct for heterogeneity bias, fixed effects models control for omitted variables that are time invariant but that do vary across cases, where cases in this study are the nations in the analysed panel dataset. The fixed effects approach provides a stringent assessment of the relationships between our predictors and both mortality outcomes given that the associations between them are estimated net of unmeasured between-country effects. We also include a dummy variable for 2005, which controls for potential unobserved heterogeneity that is cross-sectionally invariant within periods (i.e. period-specific intercept). The inclusion of period-specific intercepts is equivalent to modelling temporal fixed effects, and including both period-specific intercepts and unit-specific fixed effects is analogous to estimating a two-way fixed effects model (Baum, 2006).⁴ Two-way fixed effects models are ideal for hypothesis testing, providing more consistent and reliable results than other estimation techniques, such as random effects models (for example, Allison, 2009).

The general linear two-way fixed effects model is as follows

y_{it} = β x_{it} + u_{i} + w_{t} + e_{it}

Subscript i represents each unit of analysis (i.e. country); subscript t represents the time-period; and $y_{it}$ is the dependent variable for each country at each time-period. $β x_{it}$ represents the vector of coefficients for predictor variables that vary over time; $u_{i}$ is the country-specific disturbance term (estimated with the within estimator); $w_{t}$ is the period-specific disturbance term that is constant across all countries (estimated with the dummy variable for 2005); and $e_{it}$ is the disturbance term unique to each country at each point in time.

With the xtreg suite of commands in Stata the case-specific fixed effects are estimated with the within estimator, which involves a mean deviation algorithm for the dependent variable and each time-varying independent variable. For each case and for each time-varying variable (both outcome and predictor variables), Stata computes the means over time for that case

{\bar{y}}_{t} = \frac{1}{n_{i}} \sum_{t} y_{it} and {\bar{x}}_{t} = \frac{1}{n_{i}} \sum_{t} x_{it}

where, $n_{i}$ is the number of measurements for case i. Next, the case-specific means are subtracted from the observed values of each variable

y_{it}^{*} = y_{it} - {\bar{y}}_{i} and x_{it}^{*} = x_{it} - {\bar{x}}_{i}

Finally, Stata regresses $y^{*}$ on $x^{*}$ , plus variables to represent the effect of time. This is also sometimes called the ‘conditional’ method because it conditions out the coefficients for the fixed effects dummy variables (Allison, 2009).

Dependent Variables

The first dependent variable is infant mortality rate, which refers to the probability of a child dying between birth and the age of one, expressed per 1000 live births. We obtain these data from the World Resources Institute’s on-line Earthtrends database, who obtain them from The United Nations Children’s Fund (UNICEF) on-line.⁵ The second dependent variable, which we obtain from the same source as the first dependent variable, is child mortality rate. This measure refers to the probability of a child dying between birth and the age of five, expressed per 1000 live births. The World Resources Institute obtains the child mortality rate data from the same source as the infant mortality rate data. For additional details on either dependent variable, we refer readers to the Earthtrends and UNICEF on-line databases. Both dependent variables as well as all continuous independent variables are logged to normalise their positively skewed distributions.

Key Independent Variables

Percentage of the total population living in urban slum conditions is a relatively new measurement available from the UN-HABITAT UrbanInfo database.⁶ Since they are newly available and thus employed in limited numbers of prior comparative studies (for example, Jorgenson and Rice, 2010; Jorgenson et al., 2010; Rice and Steinkopf-Rice, 2009), we provide a relatively more detailed description of these data. All other measures used in the current study are common in prior research. For these estimates, an urban household is defined as a slum dwelling if it lacks one or more of the following: access to an improved water supply, access to improved sanitation, sufficient living area and durability of construction. More specifically, an improved water supply is one that provides a sufficient quantity of water for family use (at least 20 litres/person/day), at an affordable price (less than 10 per cent of total household income), without requiring extreme effort to obtain (less than one hour a day for the minimum sufficient quantity). In addition, an improved water supply consists of the following delivery systems: piped connection to house or plot, public standpipe serving no more than five households, bore hole, protected dug well, protected spring or rain water collection. Improved sanitation consists of a private or public toilet shared between a reasonable number of people. Improved sanitation consists of the following services: direct connection to public sewer, direct connection to a septic tank, pour flush latrine or a ventilated pit latrine. A living area is considered sufficient if there are no more than three people per habitable room (minimum of 4 square metres of space). A dwelling is defined as durable if it is built in a non-hazardous location and exhibits structural qualities adequate to protect its inhabitants from the extremes of climatic conditions, including rain, heat, cold and humidity. As we mentioned in the dataset description earlier, point estimates for the urban slums data are only available for less developed countries for 1990 and 2005 at the time of this study, which restricts the national representation and temporality of the reported analyses.

To assess if the effect of the percentage of the total population living in urban slum conditions on both outcomes varies by macro region, we calculate and employ interactions between the urban slums measure and the three macro regions (measured as dummy variables) listed in Table 1. Since we estimate fixed effects models, the dummy-coded macro region measures are unnecessary and excluded from the model estimates since they are perfectly collinear with the unit-specific fixed effects (Allison, 2009). The slope for Latin America is the reference group in the reported models that include these interactions. The test of statistical significance for the coefficients determines whether the slope for the particular interaction and the reference category—in this case Latin America—differ significantly.⁷

To assess if the effect of the percentage of the total population living in urban slum conditions on either mortality outcome increases or decreases in magnitude through time, we calculate and employ interactions between the key predictor and a dummy variable for 2005. In relevant models, the effect in 1990 is the reference group. Such an approach is becoming increasingly common in longitudinal research on various public health (for example, Brady et al., 2007) and environmental outcomes (for example, Jorgenson and Clark, 2009, 2010). However, this relatively simple two-way interaction does not allow for considering if the temporal dynamics of the effect between the dependent variables and the study’s key predictor vary by macro region. Further, while the two-way interactions between macro region and the percentage of the total population living in urban slum conditions focus on regional differences, the estimated effects for the latter are assumed to be time invariant, even though data for multiple time points are analysed. Thus, we employ a third form of interaction, which are technically three-way interactions between macro region, time and the percentage of the total population living in urban slum conditions. For this final set of interactions, Latin America in 1990 is the reference group.⁸ We elaborate on the interpretation of these and all other interaction coefficients in the results section.

Additional Independent Variables⁹

Gross domestic product (GDP) per capita is included as a control for level of economic development. These data, which we gather from the World Bank (2007), are measured in 2000 US dollars. Prior research consistently shows a negative association between infant and child mortality rates and level of economic development in less developed countries (for example, Chung and Muntaner, 2006; Frey and Field, 2000; Hertz et al., 1994; Jorgenson and Rice, 2010; Moore et al., 2006; Ram, 2006; Shandra et al., 2003; Shen and Williamson, 1999, 2001).

We control for fertility rate, which is known to be a key contributor to infant and child mortality (for example, Brady et al., 2007; Heuveline, 2001; Jorgenson, 2009; Rice, 2008; Shandra et al., 2011). Generally speaking, as fertility rates increase so do child mortality and infant mortality rates, since more fertility means more chances for mortality, all else being equal. The measures of fertility rates, which we obtain from the World Bank (2007), represent the number of children that would be born to a woman if she were to live to the end of her childbearing years and bear children in accordance with current age-specific fertility rates.

Exports as percentage of total GDP is included to control for a country’s level of integration in the world economy. These data are obtained from the World Bank (2007). While not the focus of the present study, neo-liberal perspectives (for example, Gilpin, 2001) would posit that greater world economy integration of this form will stimulate economic development and thus increase human well-being, which would involve lowering infant and child mortality rates. Conversely, critical globalisation perspectives (for example, Robinson, 2004) might posit that higher levels of exports as a percentage of total GDP for less developed countries is a structural mechanism that partially allows for more developed countries to maintain favourable terms of trade, thereby suppressing domestic economic development and well-being within the former, which could lead to increases in infant and child mortality rates (see also Shandra et al., 2010).

As noted in the model estimation technique section, we also include a dummy variable for the year 2005. Besides its incorporation for purposes of estimating two-way fixed effects models, this dummy variable is a necessary statistical control since it is used in combination with other time-variant and time-invariant predictors for calculating most of the employed two-way and three-way interactions. We also add the necessary two-way interactions between region and the 2005 dummy variable (i.e. Asia X 2005 and Africa X 2005) to the two models that include the three-way interactions between the urban slums measure, region and time.

Descriptive statistics and bivariate correlations for all included dependent and independent variables are provided in the Appendix.

Results

Tables 2 and 3 report the findings for the panel analyses.¹⁰ The analyses in Table 2 focus on the main effect of urban slum prevalence as well as its potential macro-regional variation. The analyses in Table 3 focus on the general temporal dynamics of its effect as well as the extent to which the temporal dynamics vary across the three macro regions. We provide unstandardised coefficients and robust standard errors for all predictors, the constant and its standard error for each model and the r-squared overall value for each model.

Table 2.

Unstandardised coefficients for the regression of infant mortality and child mortality rates on selected independent variables: two-way fixed effects model estimates for 80 less developed countries, 1990–2005

	IMR model A	IMR model B	CMR model A	CMR model B
Percentage of total population living in	0.208**	−0.035	0.239**	−0.013
urban slum conditions	(0.052)	(0.097)	(0.056)	(0.077)
GDP per capita	−0.268**	−0.200**	−0.279**	−0.209*
	(0.096)	(0.071)	(0.096)	(0.083)
Fertility rate	0.401**	0.276	0.499**	0.370*
	(0.143)	(0.148)	(0.162)	(0.168)
Exports as a percentage of total GDP	−0.062	−0.092	−0.081	−0.113#
	(0.059)	(0.059)	(0.066)	(0.066)
Percentage of total population living in		0.525**		0.545**
urban slum conditions X Africa		(0.157)		(0.163)
Percentage of total population living		0.242*		0.253**
inurban slum conditions X Asia		(0.102)		(0.088)
2005 dummy	−0.199**	−0.270**	−0.218**	−0.291**
	(0.043)	(0.053)	(0.049)	(0.059)
Constant	4.918**	4.567**	5.174**	4.810**
	(0.747)	(0.610)	(0.807)	(0.704)
R² overall	0.779	0.717	0.790	0.765

Notes: Unstandardised coefficients flagged for statistical significance; #p < 0.10 * p < 0.05 **p < 0.01 (two-tailed); standard errors in parentheses; Latin America is reference category in model B for both outcomes; IMR denotes infant mortality rate; CMR denotes child mortality rate.

Table 3.

	IMR model A	IMR model B	CMR model A	CMR model B
Percentage of total population living in urban slum conditions	0.220**	0.001	0.258**	0.026
	(0.079)	(0.096)	(0.081)	(0.078)
Percentage of total population living in urban slum conditions X Asia		0.251**		0.271**
		(0.095)		(0.086)
Percentage of total population living in urban slum conditions X Africa		0.303*		0.316*
		(0.146)		(0.153)
GDP per capita	−0.270**	−0.141	−0.283**	−0.144
	(0.088)	(0.074)	(0.096)	(0.086)
Fertility rate	0.405**	0.186	0.506**	0.269
	(0.142)	(0.148)	(0.159)	(0.170)
Exports as a percentage of total GDP	−0.063	−0.083	−0.082	−0.100
	(0.060)	(0.057)	(0.066)	(0.066)
2005 dummy	−0.174	−0.266**	−0.176	−0.275*
	(0.104)	(0.094)	(0.117)	(0.110)
Percentage of total population living in urban slum conditions X 2005	−0.009	−0.051	−0.014	−0.057
	(0.035)	(0.031)	(0.039)	(0.037)
Percentage of total population living in urban slum conditions X 2005 X Asia		0.012		0.006
		(0.028)		(0.032)
Percentage of total population living in urban slum conditions X 2005 X Africa		0.078**		0.080**
		(0.022)		(0.026)
Asia X 2005		−0.218		−0.300
		(0.177)		(0.178)
Africa X 2005		0.107		0.109
		(0.185)		(0.201)
Constant	4.899**	4.485**	5.143**	4.686**
	(0.759)	(0.559)	(0.804)	(0.643)
R² overall	0.778	0.821	0.788	0.753

Notes: Unstandardised coefficients flagged for statistical significance; * p < 0.05 ** p < 0.01 (two-tailed); standard errors in parentheses; Latin America is reference category in model B for both outcomes; IMR denotes infant mortality rate; CMR denotes child mortality rate.

Table 2 presents the estimates for two models of each outcome labelled as ‘model A’ and ‘model B’ for either IMR (infant mortality rate) or CMR (child mortality rate). Model A is a baseline, and includes the percentage of the total population living in urban slum conditions, GDP per capita, fertility rate, exports as a percentage of total GDP and the period-specific intercept for the year 2005. Model B includes all predictors in model A as well as the interactions between the percentage of the total population living in urban slum conditions and macro region, where Latin America is the reference group.

Table 3 also presents the estimates for two models of each outcome with the same labels as the models in Table 2. Model A includes all the baseline predictors as well as the interaction between the percentage of the total population living in urban slum conditions and the dummy variable for 2005. Model B includes all two-way and three-way interactions, which allow for assessing the extent to which the effect of the percentage of the total population living in urban slum conditions on either outcome changes through time for the nations in each of the three macro regions.

Turning first to the baseline model for both outcomes in Table 2, the effect of total population living in urban slum conditions is positive and statistically significant, which corresponds with prior research (Jorgenson and Rice, 2010). Further, and as expected, the effect of GDP per capita on both infant and child mortality rates is negative and statistically significant, while the effect of fertility rate on both outcomes is positive and statistically significant and the time dummy for 2005 is negative and statistically significant. However, the estimated coefficient for exports as a percentage of total GDP is non-significant in the baseline model for both mortality outcomes. We note that these baseline two-way fixed effects models explain close to 80 per cent of variation in both infant mortality and child mortality rates for the nations included in the analyses.

The results for model B in Table 2 indicate that the effect of the total population living in urban slum conditions on both infant mortality and child mortality rates does indeed vary across the three included macro regions. As expected, the effect is largest for the nations in Africa, followed by the nations in Asia, and then for the nations in Latin America. The effects of the control variables are similar to their effects in model A except for fertility rate, which becomes statistically non-significant for the IMR model and exports as a percentage of total GDP exhibits a negative, marginally statistically significant effect in the CMR model. Of particular relevance for the purposes of the study, the estimated coefficients for the two-way interactions, which suggest non-trivial macro-regional differences in the association between urban slum size and both mortality outcomes, do not account for possible changes in magnitude through time. We now turn to the second series of analyses that explore such possible temporal and regional differences.

Model A for both outcomes in Table 3 suggests that the effect of total population living in urban slum conditions is time invariant. More specifically, the interaction between this key predictor and the time dummy variable for 2005 is non-significant for infant mortality and child mortality rates. The statistically significant effects of development and fertility rate are consistent with the baseline model estimates in Table 2. The results of model B, however, indicate that the suggested time-invariant nature of urban slum prevalence on both mortality outcomes is somewhat invalid. While the model estimates suggest that the effect is non-significant for the Latin American nations in both 1990 and 2005, they also suggest that its effect is positive but time invariant for the Asian nations and positive for the African nations at both time points yet increasing in size from 1990 to 2005. Thus, and overall, in the context of the effect of urban slum prevalence on both infant and child mortality rates, we find notable regional and temporal variation, net of important statistical controls.

Conclusion

The growth of urban slums in the developing countries is a notable trend of the past several decades and promises to remain so, far into the 21st century. Such growth is a material expression of the urbanisation of poverty amidst insufficient, and often declining, provisioning of public services. Rather than a numerical minority, urban slum residents are becoming the predominant expression of urbanisation and represent a ‘new paradigm’ or fundamental realignment of human settlement patterns forcing a reconsideration of the rural–urban dichotomy as an overarching conceptualisation of socio-organisational form (UN-HABITAT, 2003a, p. 6). Attainment of the Millennium Development Goals (MDGs), benchmarks established by the United Nations member-states in 2000, is indeed predicated in large part upon improving the living conditions of urban slum residents in the developing countries. This goal is recognised in particular in MDG 7, Target 11, which endeavours: “By 2020, to have achieved a significant improvement in the lives of at least 100 million slum-dwellers”.¹¹ This is a notable goal, but if current trends continue there will be over 500 million more urban slum residents world-wide in 2020 than when the MDGs were originally established in 2000.

Drawing from the theoretical and empirical research in social epidemiology and urban political economy, the present study examined the impact of the proportion of the total population living in urban slums on infant and under-five mortality among 80 less developed countries. Results of two-way fixed effects panel regression analysis illustrate that growth in the percentage of the total population living in urban slum conditions over the period 1990–2005 increased both infant and child mortality rates. These results hold, net of various economic and demographic factors. The analysis thus illustrates the urban built environment as a key factor shaping the social production of infant and child mortality in less developed countries.

The impact of urban slums conditions on infant and under-five mortality between 1990 and 2005 is particularly pronounced for African countries relative to less developed countries in Latin America and Asia. The analysis further indicates that the magnitude of this effect increased over the period within African nations but did not change in magnitude through time in the Latin American or Asian countries under examination—at least at the national level. The urban transition, of note, began later in Africa than elsewhere; a clear majority of nations in this macro region, particularly sub-Saharan African countries, are now characterised by an unprecedented pace of urban slum formation and expansion. Urbanisation and urban slum formation and expansion are now virtually synonymous.

The results of the present study highlight the relevance of the built urban environment in terms of the social production of infant and child mortality. Indeed, the growth of urban slums threatens to undercut the public health advantages of urbanisation in ways many researchers have yet fully to consider. The “spatial landscape of poverty” is shifting to the burgeoning urban areas of the developing countries (UN-HABITAT, 2003a, p. 54)—visible in the enclaves of prosperity and commerce tied to the world economy surrounded by vast expanses of social, political and economic exclusion.

Epidemiological studies have greatly contributed to the identification of the individual-level risk factors shaping the incidence of morbidity and mortality, but a renewed emphasis on the identification of “basic social conditions” is crucial to delineating the contextual determinants that further constitute fundamental causes of health or illness (Link and Phelan, 1995, p. 80; 1996). Few studies from a macro comparative perspective explicitly focus on the factors promoting and constraining urban slum prevalence growth in less developed countries. In turn, future research efforts that examine in detail such concerns would constitute an important contribution to existing knowledge regarding the factors that shape the social production of infant and child mortality rates in less developed countries and would also advance our collective understanding of such unhealthy built environments from a broader urban political economy perspective.

Footnotes

Appendix

Table A1.

Descriptive statistics and bivariate correlations

	Mean	S.D.	1	2	3	4	5	6	7	8	9	10	11
1 Infant Mortality Rate	3.95	0.69
2 Child Mortality Rate	4.27	0.83	0.99
3 Percentage of total population living in urban slum conditions	2.73	0.93	0.33	0.32
4 GDP per capita	6.71	1.12	−0.79	−0.80	−0.09
5 Fertility Rate	1.40	0.41	0.84	0.86	0.30	−0.72
6 Exports as Percent Total GDP	3.30	0.61	−0.45	−0.43	−0.13	0.43	−0.36
7 Percentage of total population living in urban slum conditions X Africa	1.38	1.55	0.65	0.68	0.38	−0.49	0.64	−0.06
8 Percentage of total population living in urban slum conditions X Asia	0.59	1.16	−0.08	−0.11	0.08	−0.03	−0.21	−0.12	−0.45
9 2005 Dummy	0.50	0.50	−0.27	−0.26	0.05	0.11	−0.34	0.28	0.04	−0.01
10 Percentage of total population living in urban slum conditions X 2005	1.39	1.57	−0.10	−0.09	0.40	0.03	−0.17	0.19	0.17	0.03	0.89
11 Percentage of total population living in urban slum conditions X 2005 X Asia	0.29	0.89	−0.17	−0.19	0.10	0.04	−0.28	0.07	−0.29	0.68	0.33	0.34
12 Percentage of total population living in urban slum conditions X 2005 X Africa	0.72	1.35	0.33	0.35	0.30	−0.28	0.27	0.09	0.64	−0.27	0.53	0.64	−0.17

Notes: See methods section for variable descriptions; two observations on 80 countries; N = 160.

Notes

Funding Statement

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

References

Agarwal

Taneja

(2005) All slums are not equal: child health conditions among the urban poor, Indian Pediatrics, 42, pp. 233–244.

Allison

(2009) Fixed Effects Regression Models. Thousand Oaks, CA: Sage.

Awasthi

Agarwal

(2003) Determinants of childhood mortality and morbidity in urban slums in India, Indian Pediatrics, 40, pp. 1145–1161.

Bartlett

(2003) Water, sanitation, and urban children: the need to go beyond ‘improved’ provision, Environment and Urbanization, 15(2), pp. 57–70.

Baum

(2006) An Introduction to Modern Econometrics Using Stata. College Station, TX: Stata Press.

Berkman

L. F.

Kawachi

(2000) A historical framework for social epidemiology, in: Berkman

L. F.

Kawachi

(Eds) Social Epidemiology, pp. 3–12. New York: Oxford University Press.

Brady

Kaya

Beckfield

(2007) Reassessing the effect of economic growth on well-being in less-developed countries, 1980–2003, Studies in Comparative International Development, 42, pp. 1–35.

Cassel

(1964) Social science theory as a source of hypotheses in epidemiological research, American Journal of Public Health, 54(9), pp. 1482–1488.

Chung

Muntaner

(2006) Political and welfare state determinants of infant and child health indicators: an analysis of wealthy countries, Social Science & Medicine, 63, pp. 829–842.

10.

Cohen

(2006) Urbanization in developing countries: current trends, future projections, and key challenges for sustainability, Technology in Society, 28, pp. 63–80.

11.

Cwikel

(2006) Social Epidemiology: Strategies for Public Health Activism. New York: Columbia University Press.

12.

Davis

(2006) Planet of Slums. New York: Verso.

13.

Dunlap

R. E.

Catton

W. R.

Jr (1979) Environmental sociology, Annual Review of Sociology, 5, pp. 243–273.

14.

Dunlap

R. E.

Michelson

Stalker

(2001) Environmental sociology: an introduction, in: Dunlap

R. E.

Michelson

(Eds ) Handbook of Environmental Sociology, pp. 1–32. Westport, CT: Greenwood Press.

15.

Ehiri

J. E.

Prowse

J. M.

(1999) Child health promotion in developing countries: the case for integration of environmental and social interventions?, Health Policy and Planning, 14(1), pp. 1–10.

16.

Farmer

(1999) Infections and Inequalities: The Modern Plagues. Berkeley, CA: University of California Press.

17.

Farmer

Nizeye

Stulac

Keshavjee

(2006) Structural violence and clinical medicine, PLoS Medicine, 3(10), pp. 1686–1691.

18.

Fernandez

Mondkar

Mathai

(2003) Urban slum-specific issues in neonatal survival, Indian Pediatrics, 40, pp. 1161–1166.

19.

Fitzpatrick

LaGory

(2000) Unhealthy Places: The Ecology of Risk in the Urban Landscape. New York: Routledge.

20.

Fotso

J.-C.

Ezeh

A. C.

Madise

N. J.

Ciera

(2007) Progress towards the child mortality millennium development goal in urban sub-Saharan Africa: the dynamics of population growth, immunization, and access to clean water, BMC Public Health, 7, p. 218.

21.

Frey

R. S.

Field

(2000) The determinants of infant mortality in less developed countries: a cross national test of five theories, Social Indicators Research, 52, pp. 215–234.

22.

Ghosh

Shah

(2004) Nutritional problems in urban slum children, Indian Pediatrics, 41, pp. 682–696.

23.

Gilpin

(2001) Global Political Economy: Understanding the International Economic Order. Princeton, NJ: Princeton University Press.

24.

Gulati

J. K.

Jaswal

(1998) Maternal and child health care in slums of Ludhiana City, Indian Journal of Maternal and Child Health, 9(1), pp. 48–51.

25.

Halaby

(2004) Panel models in sociological research: theory into practice, Annual Review of Sociology, 30, pp. 507–544.

26.

Hertz

Herbert

J. R.

Landon

(1994) Social and environmental factors and life expectancy, infant mortality, and maternal mortality rates: results of a cross-national comparison, Social Science and Medicine 39, pp. 105–114.

27.

Heuveline

(2001) Demographic pressure, economic development, and social engineering: an assessment of fertility declines in the second half of the twentieth century, Population Research Policy Review, 20, pp. 365–396.

28.

Hill

Pebley

A. R.

(1989) Child mortality in the developing world, Population and Development Review, 15(4), pp. 657–687.

29.

Hoque

Selwyn

B. J.

(1996) Birth practice patterns in urban slums of Dhaka, Bangladesh, Women & Health, 24(1), pp. 41–58.

30.

Jorgenson

A. K.

(2009) Foreign direct investment and the environment, the mitigating influence of institutional and civil society factors, and relationships between industrial pollution and human health: a panel study of less-developed countries, Organization & Environment, 22, pp. 135–157.

31.

Jorgenson

A. K.

Clark

(2009) The economy, military, and ecologically unequal relationships in comparative perspective: a panel study of the ecological footprints of nations, 1975–2000, Social Problems, 56, pp. 621–646.

32.

Jorgenson

A. K.

Clark

(2010) Assessing the temporal stability of the population/environment relationship: a cross-national panel study of carbon dioxide emissions, 1960–2005, Population & Environment, 32, pp. 27–41.

33.

Jorgenson

A. K.

Rice

(2010) Urban slum growth and human health: a panel study of infant and child mortality in less-developed countries, 1990–2005, Journal of Poverty, 14, pp. 382–402.

34.

Jorgenson

A. K.

Rice

Clark

(2010) Cities, slums, and energy consumption in less-developed countries, 1990–2005, Organization & Environment, 23, pp. 189–204.

35.

Kilmartin

(2001) Designing the built environment, in: Dunlap

R. E.

Michelson

(Eds) Handbook of Environmental Sociology, pp. 167–191. Westport, CT: Greenwood Press.

36.

Krieger

(2000) Epidemiology and social sciences: towards a critical reengagement in the 21st century, Epidemiologic Reviews, 22(1), pp. 155–163.

37.

Krieger

(2001a) Theories for social epidemiology in the 21st century: an ecosocial perspective, International Journal of Epidemiology, 30(4), pp. 668–677.

38.

Krieger

(2001b) Historical roots of social epidemiology: socioeconomic gradients in health and contextual analysis, International Journal of Epidemiology, 30, pp. 899–900.

39.

Krieger

(2001c) A glossary for social epidemiology, Journal of Epidemiology and Community Health, 55(10), pp. 693–700.

40.

Krieger

Zierler

(1996) What explains the public’s health: a call for epidemiologic theory, Epidemiology, 7(1), pp. 107–109.

41.

Link

B. G.

Phelan

J. C.

(1995) Social conditions as fundamental causes of disease, Journal of Health and Social Behavior, Extra Issue, pp. 80–94.

42.

Link

B. G.

Phelan

J. C.

(1996) Understanding sociodemographic differences in health: the role of fundamental social causes, American Journal of Public Health, 86(4), pp. 471–473.

43.

Macintyre

Ellaway

(2000) Ecological approaches: rediscovering the role of the physical and social environment, in: Berkman

L. F.

Kawachi

(Eds) Social Epidemiology, pp. 332–348. New York: Oxford University Press.

44.

McKinlay

J. B.

(1981) A case for refocusing upstream: the political economy of illness, in: Conrad

Kern

(Eds) The Sociology of Health and Illness: Critical Perspectives, pp. 613–663. New York: St Martin’s Press.

45.

Mogford

(2004) Structural determinants of infant mortality in sub-Saharan Africa: a cross-national study of ecnomic and social influences from 1970–1977, Biodemography and Social Biology, 51, pp. 94–120.

46.

Moore

Teixeira

Shiell

(2006) The health of nations in a global context: trade, global stratification, and infant mortality rates, Social Science & Medicine, 63, pp. 165–178.

47.

Oakes

J. M.

Kaufman

J. S.

(2006) Introduction: advancing methods in social epidemiology, in: Oakes

J. M.

Kaufman

J. S.

(Eds) Methods in Social Epidemiology: Research Design and Methods, pp. 3–20. San Francisco, CA: Jossey-Bass.

48.

Rahi

Taneja

D. K.

Misra

Mathur

N. B.

. (2006) Newborn care practices in an urban slum of Delhi, Indian Journal of Medical Sciences, 60(12), pp. 506–513.

49.

Ram

(2006) Further examination of the cross-country association between income inequality and population health, Social Science & Medicine, 62, pp. 779–791.

50.

Rice

(2008) Material consumption and social well-being within the periphery of the world economy: an ecological analysis of maternal mortality, Social Science Research, 37, pp. 1292–1309.

51.

Rice

Steinkopf-Rice

(2009) The concentration of disadvantage and the rise of an urban penalty: urban slum prevalence and the social production of health inequalities in the developing countries, International Journal of Health Services, 39(4), pp. 749–770.

52.

Robinson

(2004) A Theory of Global Capitalism. Baltimore, MD: Johns Hopkins University Press.

53.

Satterthwaite

(1993) The impact on health of urban environments, Environment and Urbanization, 5(2), pp. 87–111.

54.

Sclar

Garau

Carolini

(2005) The 21st century health challenge of slums and cities, The Lancet, 365, 5 March, pp. 901–903.

55.

Shandra

London

(2011) World Bank structural adjustment, water, and sanitation: a cross-national analysis of child mortality in sub-Saharan Africa, Organization & Environment, 24, pp. 107–129.

56.

Shandra

Nobles

London

Williamson

(2003) Dependency, democracy, and infant mortality: a cross-national analysis of developing nations, Social Science and Medicine, 59, pp. 321–333.

57.

Shandra

London

(2010) Debt, structural adjustment, and non-governmental organizations: a cross-national analysis of maternal mortality, Journal of World-Systems Research, 16, pp. 217–245.

58.

Shen

Williamson

(1999) Maternal mortality, women’s status, and economic ependency in less developed countries: a cross-national analysis, Social Science and Medicine, 49, pp. 197–214.

59.

Shen

Williamson

(2001) Accounting for cross-national differences in infant mortality decline (1965–1991) among less developed countries: effects of women’s status, economic dependency, and state strength, Social Indicators Research, 53, pp. 257–288.

60.

Turshen

(1977) The political ecology of disease, Review of Radical Political Economics, 9(1), pp. 45–60.

61.

UNDP (United Nations Development Programme) (2006) Human Development Report. New York: Oxford University Press.

62.

UN-HABITAT (2003a) Slums of the World: The Face of Urban Poverty in the New Millennium? London: Earthscan (http://www.unhabitat.org).

63.

UN-HABITAT (2003b) The Challenge of Slums: Global Report on Human Settlements 2003. London: Earthscan (http://www.unhabitat.org).

64.

UN-HABITAT (2006) State of the World’s Cities Report 2006/7. London: Earthscan (http://www.unhabitat.org).

65.

UN-HABITAT (2007) The urban penalty: the poor die young (http://www.unhabitat.org; accessed May 2007).

66.

UN-HABITAT (2008) UrbanInfo database (http://www.devinfo.info/urbaninfo/; accessed July 2009).

67.

Vaid

Mammen

Primrose

Kang

(2007) Infant mortality in an urban slum, Indian Journal of Pediatrics, 74, pp. 449–453.

68.

Vlahov

Freudenberg

Proietti

Ompad

. (2007) Urban as a determinant of health, Journal of Urban Health: Bulletin of the New York Academy of Medicine, 84(1), pp. i16–i26.

69.

Wagstaff

Bustreo

Bryce

Claeson

(2004) Child health: reaching the poor, American Journal of Public Health, 94(5), pp. 726–736.

70.

Waitzkin

(1981) The social origins of illness: a neglected history, International Journal of Health Services, 11(1), pp. 77–103.

71.

World Bank (2007) World development indicators (CD-ROM version). World Bank, Washington, DC.

72.

Yankauer

Jr (1950) The relationship of fetal and infant mortality to residential segregation: an inquiry into social epidemiology, American Sociological Review, 15, pp. 644–648.

Assessing the Temporal and Regional Differences in the Relationships between Infant and Child Mortality and Urban Slum Prevalence in Less Developed Countries,1990–2005

Abstract

Introduction

Literature Review

Urban Slums as Built Environments and the Social Production of Mortality

Macro-regional and Temporal Variation: Mortality and Urban Slum Prevalence in Africa

The Analyses

The Dataset

Model Estimation Technique

Dependent Variables

Key Independent Variables

Additional Independent Variables 9

Results

Conclusion

Footnotes

Appendix

Notes

Funding Statement

References

Additional Independent Variables⁹