RESEARCH

 

The global distribution of risk factors by poverty level

 

Répartition mondiale des facteurs de risque par niveau de pauvreté

 

Distribución mundial de los factores de riesgo por nivel de pobreza

 

 

Tony BlakelyI,1; Simon HalesII; Charlotte KieftIII; Nick WilsonIV; Alistair WoodwardV

IAssociate Professor, Department of Public Health, Wellington School of Medicine and Health Sciences, University of Otago, PO Box 7343, Wellington, New Zealand (email: tblakely@wnmeds.ac.nz)
IISenior Research Fellow, Department of Public Health, Wellington School of Medicine, Wellington, New Zealand
IIIData Manager and Analyst, Department of Public Health, Wellington School of Medicine, Wellington, New Zealand
IVPublic Health Medicine Specialist, Department of Public Health, Wellington School of Medicine, Wellington, New Zealand
VProfessor, School of Population Health, University of Auckland, Auckland New Zealand

 

 


ABSTRACT

OBJECTIVE: To estimate the individual-level association of income poverty with being underweight, using tobacco, drinking alcohol, having access only to unsafe water and sanitation, being exposed to indoor air pollution and being obese.
METHODS: Using survey data for as many countries as possible, we estimated the relative risk association between income or assets and risk factors at the individual level within 11 medium- and low-income subregions of WHO. WHO and The World Bank data on the prevalence of risk factors and income poverty (defined as living on < US$ 1.00 per day, US$ 1—2.00 per day and > US$ 2.00 per day) were analysed to impute the association between poverty and risk factors for each subregion. The possible effect of poverty reduction on the prevalence of risk factors was estimated using population-attributable risk percentages.
FINDINGS: There were strong associations between poverty and malnutrition among children, having access only to unsafe water and sanitation, and being exposed to indoor air pollution within each subregion (relative risks were twofold to threefold greater for those living on < US$ 1.00 per day compared with those living on > US$ 2.00 per day). Associations between poverty and obesity, tobacco use and alcohol use varied across subregions. If everyone living on < US$ 2.00 per day had the risk factor profile of those living on > US$ 2.00 per day, 51% of exposures to unimproved water and sanitation could be avoided as could 37% of malnutrition among children and 38% of exposure to indoor air pollution. The more realistic, but still challenging, Millennium Development Goal of halving the number of people living on < US$ 1.00 per day would achieve much smaller reductions.
CONCLUSION: To achieve large gains in global health requires both poverty eradication and public health action. The methods used in this study may be useful for monitoring pro-equity progress towards Millennium Development Goals.

Keywords: Poverty; Health status; Socioeconomic factors; Child nutrition disorders/epidemiology/economics; Water supply/economics; Sanitation/economics; Air pollution, Indoor/economics; Smoking/epidemiology/economics; Alcohol drinking/epidemiology/economics; Obesity/epidemiology/economics; World health; Risk factors (source: MeSH, NLM).


RÉSUMÉ

OBJECTIF: Estimer l'association au niveau individuel entre pauvreté des revenus d'une part et maigreur, tabagisme, alcoolisme, absence d'accès à une eau saine et à des équipements sanitaires convenables, exposition à la pollution de l'air intérieur et obésité, d'autre part.
MÉTHODES: A l'aide de données d'enquête relatives au plus grand nombre possible de pays, les auteurs ont estimé l'association correspondant au risque relatif entre le revenu ou les ressources et les facteurs de risque au niveau individuel dans 11 sous-régions de l'OMS à revenus moyens et faibles. Ils ont analysé les données de l'OMS et de la Banque mondiale sur la prévalence des facteurs de risque et du niveau de pauvreté (défini comme le fait de disposer pour vivre de moins de US $ 1,00 par jour, de US $ 1 à 2,00 par jour ou de plus de US $ 2,00 par jour) pour évaluer l'association entre pauvreté et facteurs de risque pour chaque sous-région. Ils ont estimé l'effet éventuel d'une réduction de la pauvreté sur la prévalence des facteurs de risque à l'aide des pourcentages de risque attribuable des populations.
RÉSULTATS: Il existait de fortes associations entre pauvreté et malnutrition chez les enfants n'ayant accès qu'à de l'eau et à des équipements sanitaires insalubres et exposés à la pollution de l'air intérieur dans chaque sous-région (les risques relatifs étaient deux à trois fois plus élevés pour ceux vivant avec moins de US $ 1 par jour que pour ceux vivant avec plus de US $ 2 par jour). Les associations entre pauvreté d'une part et obésité, tabagisme ou alcoolisme d'autre part étaient variables d'une sous-région à l'autre. Si toutes les personnes disposant de moins de US $ 2,00 par jour présentaient le profil de facteurs de risque de celles vivant avec plus de US $ 2,00 par jour, 51 % des expositions à de l'eau et à un réseau sanitaire non traités, 37 % des cas de malnutrition infantile et 38 % des expositions à la pollution de l'air intérieur pourraient être évités. L'objectif de développement du Millénaire, certes plus réaliste, mais encore difficile à atteindre, consistant à réduire d'un facteur deux le nombre de personnes vivant avec moins de US $ 1,00 par jour, permettrait d'obtenir des diminutions plus faibles de ces nombres de cas.
CONCLUSION: L'obtention d'améliorations conséquentes de la santé dans le monde exige à la fois l'éradication de la pauvreté et des mesures de santé publique. Les méthodes employées dans cette étude peuvent être utiles à la surveillance des progrès en faveur de l'équité dans la réalisation des objectifs de développement du Millénaire.

Mots clés: Pauvreté; Etat sanitaire; Facteur socioéconomique; Troubles nutrition enfant/économie; Alimentation eau/économie; Assainissement/économie; Pollution air ambiant/économie; Tabagisme/économie; Consommation alcool/économie; Obésité/économie; Santé mondiale; Facteur risque (source: MeSH, INSERM).


RESUMEN

OBJETIVO: Estimar la relación individual existente entre la pobreza de ingresos y la insuficiencia ponderal, el consumo de tabaco, el consumo de alcohol, el hecho de no disponer más que de agua y saneamiento insalubres, la exposición a aire contaminado en interiores y la obesidad.
MÉTODOS: Usando datos encuestales para el máximo número de países posible, estimamos el riesgo relativo de asociación de los ingresos o el patrimonio a factores de riesgo particulares en 11 subregiones de la OMS de ingresos bajos y medios. Se analizaron datos de la OMS y del Banco Mundial sobre la prevalencia de los factores de riesgo y la pobreza de ingresos (definida distinguiendo la subsistencia con menos de US$ 1,00 diarios, con US$ 1,00-2,00 diarios y con más de US$ 2,00 diarios) a fin de determinar la relación entre pobreza y factores de riesgo para cada subregión. El posible efecto de la reducción de la pobreza en la prevalencia de los factores de riesgo se estimó a partir de los porcentajes de riesgo atribuible poblacionales.
RESULTADOS: Se detectó una estrecha relación entre la pobreza y la malnutrición infantil, el hecho de no disponer más que de agua y saneamiento insalubres, y la exposición a aire contaminado en interiores dentro de cada subregión (los riesgos relativos fueron entre dos y tres veces mayores entre quienes vivían con menos de US$ 1,00 al día que en quienes subsistían con más de US$ 2,00 al día). El grado de asociación de la pobreza a la obesidad, el consumo de tabaco y el consumo de alcohol difería de una subregión a otra. Si todas las personas que viven con menos de US$ 2,00 al día tuvieran el mismo perfil de factores de riesgo que las que viven con más de US$ 2,00 diarios, se podrían evitar el 51% de los casos de exposición a sistemas de abastecimiento de agua y saneamiento no mejorados, así como el 37% de la malnutrición infantil y el 38% de la exposición a aire contaminado en locales cerrados. El más realista, pero con todo difícil, de los Objetivos de Desarrollo del Milenio de reducir a la mitad el número de personas con menos de US$ 1,00 al día se traduciría en disminuciones mucho menores.
CONCLUSIÓN: Para conseguir grandes avances en el terreno de la salud mundial se requieren medidas tanto de erradicación de la pobreza como de salud pública. Los métodos empleados en este estudio podrían ayudar a vigilar los progresos en equidad hacia los Objetivos de Desarrollo del Milenio.

Palabras clave: Pobreza; Estado de salud; Factores socioeconómicos; Trastornos de la nutrición del niño/economía; Abastecimiento de agua/economía; Saneamiento/economía; Contaminación del aire interior/economía; Tabaquismo/economía; Consumo de bebidas alcohólicas/economía; Obesidad/economía; Salud mundial; Factores de riesgo (fuente: DeCS, BIREME).



 

 

Introduction

There is a large body of research conducted in richer countries on the socioeconomic determinants of health that contrasts health and disease status among individuals of varying socioeconomic positions (1—7). The same is not true of poorer regions of the world, although research is starting to map the distribution of health by socioeconomic status at the individual level within poorer countries (see, for example, http://www.worldbank.org/poverty/health/data) (8—11). It is important that this gap continues to be filled since ecological data may give a misleading picture of what is happening at the level of individuals (12, 13). Individual-level data are also required to set targets and monitor progress towards reducing health inequalities (14), and these targets have been identified as a priority by WHO in relation to monitoring and ensuring pro-equity progress towards achieving the Millennium Development Goals (15).

The association between socioeconomic position and health risk factors varies over time and between regions of the world (8, 16). Relationships observed in high-income countries may not hold in the middle- and low-income countries that account for about 80% of the world's population.

The aim of this paper is to describe the association between poverty and the prevalence of major risk factors for ill-health at the individual level among the 5 billion people living in low- and middle-income regions. The study was conducted as part of the WHO Comparative Risk Assessment project (17, 18). Poverty was defined in absolute terms. We sought survey data for individuals that included both risk factors and a measure of socioeconomic position from as many countries and regions as possible. Data were obtained for six major risk factors that have also been included in the WHO Comparative Risk Assessment project. They are: being underweight; using tobacco; drinking alcohol; having access only to unsafe water, sanitation and hygiene; being exposed to indoor air pollution from solid fuels; and being overweight or obese (which were combined).

 

Methods

Estimates of the association of income poverty with risk factors were conducted separately for up to 11 of the 14 WHO subregions. WHO divides the world into six general regions: Africa, the Americas, the Eastern Mediterranean, Europe, South-East Asia and the Western Pacific. Countries within each of these regions are then divided into subregions based on levels of child and adult mortality (18). In those countries in stratum A there is very low child mortality and very low adult mortality; in stratum B there is low child mortality and very low adult mortality; in stratum C there is low child mortality and high adult mortality; in stratum D there is high child mortality and high adult mortality; and in stratum E there is high child mortality and very high adult mortality.

The three richest regions have low child mortality and low adult mortality and are classified as the Americas, stratum A; Europe, stratum A; and Western Pacific, stratum A. These subregions are excluded from our analyses because they have negligible levels of absolute poverty (Fig. 1, web version only, available at: http://www.who.int/bulletin).

To arrive at our aim of estimating the prevalence of risk factors by income poverty level for WHO subregions, we used four steps. First, we determined the association between socioeconomic factors (i.e. asset score or income) and risk factors within each WHO subregion. To unify the results, our second step generalized the results found in step 1 to relative risks by level of income poverty. Our third step estimated the prevalence of each risk factor within the levels of income poverty by WHO subregion. The fourth and final step estimated population-attributable risk percentages for various counterfactual changes in income poverty. Box 1 summarizes these steps, and they are described further below. A more detailed description is available elsewhere (19).

Step 1: associating asset score or income with risk factor

We determined the association between socioeconomic status and risk factors using Demographic and Health Survey data (or DHS, available from http://www.measuredhs.com/) for malnutrition among children, access only to unsafe water and sanitation, and risk of maternal obesity. We used data from the Living Standards Measurement Study (or LSMS, available from http://www.worldbank.org/lsms) for indoor air pollution (the use of smoke-producing fuels in cooking, such as wood, coal, and charcoal) and alcohol and tobacco use.

The DHS covers 53 countries with an average sample size of about 5000 (Table 1, web version only, available at: http://www.who.int/bulletin). The most recent survey in each country was used if the country had been surveyed more than once during the period 1986—2000. A child was defined as malnourished if his or her weight-for-age Z-score was less than -2 using the National Centre for Health Statistics or Harvard reference populations. The availability of water and sanitation was defined as in Prüss et al. (20). Data on body weight were available only for mothers of children aged 0—4 years. Being overweight or obese was defined as having a body mass index > 25 kg/m².

The DHS does not include data on poverty or income. We therefore constructed an asset score using approximately 500 000 DHS observations for all countries combined, following the general method developed by The World Bank (21), and using the first factor from a factor analysis of four variables that were most consistently available across countries. These were electricity supply, educational status, housing construction material and urban—rural status. (If we had used more than four asset variables per country, many countries would have been excluded from the analysis.) Given that only four variables (each with relatively few values) were available for the factor analysis, only 96 discrete asset score values were generated. We calculated the prevalence of malnutrition, unsafe water and sanitation and mothers being overweight by WHO subregion for each discrete value of the asset score.

We then fitted non-parametric linear weighted regressions separately by WHO subregion using the data aggregated to unique values of the asset score to allow for non-linear associations, using the Proc Loess procedure in SAS. (Proc Loess conducts a series of automated linear regressions at each x value, where the data considered include a bandwidth of data on either side. Each observation within this bandwidth is assigned a local weight that decreases the further away it is from the central x value.)

The proportion of people with the given risk factor at each unique value of the asset score variable was the response variable, and the asset score rank was the predictor variable. (The datasets were too large to run the models on unit-level DHS data. However, given that the data were aggregated by discrete value of the asset score, the results would be similar to the regression on the unit-level data.) The asset score rank (range = 0—1) was calculated separately for each subregion using DHS survey weights and data on population counts within each subregion to ensure representativeness of all people in all countries for whom we had data in each subregion. (See reference 19 for more details.) The actual weighting in the regression was by the number of DHS observations represented by each datapoint. Fig. 2 shows an example of a fitted curve for child malnutrition on asset scores in the subregion Africa, stratum D (a region with high child mortality and high adult mortality). Each circle plots the proportion of malnourished children by each unique value of the ranked asset score, and the size of each circle is proportional to the number of DHS observations. The fitted regression line is also shown.

We were able to access data from the LSMS for 11 of about 25 countries with these data (Table 1, web version only, available at: http://www.who.int/bulletin) from surveys conducted between 1991 and 1999. Data on alcohol and tobacco use in Bulgaria, Ghana, South Africa and Tajikistan were available only in the form of household expenditure data (not individual consumption); for Azerbaijan there was a composite variable of combined alcohol and tobacco consumption. The analyses of the LSMS data varied from the DHS analyses as follows. First, these data included an income variable that we equivalized for household economies of scale by dividing by the square root of the number of people in the household and then ranking households from 0 to 1. Second, we undertook the regression analyses by country to avoid the problem of varying purchasing power parity between countries. Third, because these datasets were smaller we undertook a locally weighted linear regression using Stata software for the unit-level data, modelling the dependent variables as the logit.

Neither DHS data nor LSMS data were available for China, the country that dominates the Western Pacific Region, stratum B (an area of low child and low adult mortality). Instead we used data from the 1993 China Health and Nutrition Survey (available at http://www.cpc.unc.edu/china), and analysed these data in a manner comparable to the LSMS analyses.

Step 2: relative risk of risk factor by poverty

We used data from The World Bank on income poverty in 76 countries to estimate the distribution of poverty within countries and WHO subregions (Table 1, web version only, available at: http://www.who.int/bulletin); these data were the World Development Indicators and data from Chen & Ravallion (22) and Ravallion et al. (23). The World Bank income poverty estimates used consumption data where possible, adjusted all dollar estimates to one point in time (1993), and adjusted for purchasing power parity. We used the trichotomous poverty variable: living on < US$ 1.00 per day, US$ 1—2.00 per day and > US$ 2.00 per day. Altogether, 89% of the population in the Eastern Mediterranean stratum B (low child and low adult mortality) resided in countries that did not have country-level estimates of absolute poverty, therefore all estimates in this paper for this region must be treated with particular caution. However, at least 70% of the population in each of the remaining 10 subregions resided in countries with poverty estimates.

Next, we overlaid these poverty estimates onto the results of the non-parametric regression. For example, in Africa, stratum D (high child and high adult mortality) where 55.5% of the population live on < US$ 1.00 per day, we assumed that the prevalence of malnutrition among children in this socioeconomic position was equivalent to the average of that predicted by the non-parametric regression (from step 1) for rank score values of asset scores ranging from 0 to 0.555. This is shown diagrammatically in Fig. 2; for rank scores up to 0.555 the average proportion of malnutrition using the extrapolated regression line of best fit is 0.405 (or a prevalence of 40.5%). That is, we assumed that a group's ranking by income poverty and asset score (at the level of three categories) was equivalent. (Elsewhere, we have demonstrated a strong, but not perfect, association between asset scores and income in Pakistan, a country where we could calculate both an asset score and household income (19)).

Having calculated the prevalence of each risk factor by band of income poverty, we calculated the relative risks by comparing those living on < US$ 1.00 per day and those living on US$ 1—2.00 per day. For example, in the countries in Africa, stratum D the relative risk for those living on < US$ 1.00 per day was 2.26 (40.5%/17.9%) and for those living on US$ 1—2.00 per day it was 1.44 (25.7%/17.9%) (Fig. 2). We were also able to estimate the relative risks for people living on exactly US$ 2.00 per day relative to those living on > US$ 2.00 per day using the predicted prevalence at US$ 2.00 per day.

Step 3: prevalence of risk factors by income poverty

Step 2 produced our best estimates of the associations of relative risks in each subregion. However, the best estimate of the overall prevalence of a risk factor for each subregion (i.e. not stratified by income poverty) was provided by other comparative risk assessment teams working on the wider WHO-sponsored project. It may be that the survey data we used to estimate the associations of relative risks were not the best that could be used to estimate overall prevalence. We therefore incorporated the external best estimates of overall prevalence and algebraically back-calculated the prevalence of each risk factor by income poverty using these estimates of overall risk factors, our relative risks derived from step 2, and The World Bank-based estimates of income poverty for each subregion.

Step 4: population-attributable risk percentages

We calculated population-attributable risk percentages for three counterfactual scenarios:

  • everyone in each subregion living on < US$ 2.00 per day adopts the prevalence of risk factors of those living on > US$ 2.00 per day;
  • everyone in each subregion living on < US$ 2.00 per day adopts the prevalence of risk factors of those living on exactly US$ 2.00 per day; and
  • half of the people in each subregion who are living on < US$ 1.00 per day adopt the prevalence of risk factors of those living on > US$ 1.00 per day.

The third scenario is based on the Millennium Development Goal of eradicating extreme poverty and hunger and its accompanying target to "halve, between 1990 and 2015, the proportion of people whose income is less than $1 a day" (see http://www.unmillenniumproject.org for more information). In this scenario we calculated only the risks of malnutrition among children, unsafe water and sanitation, and indoor air pollution.

 

Results

Fig. 3 shows the estimated prevalence of each risk factor by level of income poverty within each of the WHO subregions. The comparable relative risks of each risk factor by income poverty (using >US$ 2.00 per day as the reference group) are shown in Table 2 (web version only, available at: http://www.who.int/bulletin).

Several patterns are evident. First, there are strong associations across all WHO subregions between absolute income poverty and increasing malnutrition among children, access only to unsafe water and sanitation, and exposure to indoor air pollution. The prevalence of malnutrition among children for a given level of income poverty varies across subregions (Fig. 3) but the relative risks are remarkably similar (Table 2) except for the Western Pacific Region in stratum B (low child mortality and low adult mortality) which is strongly influenced by China and the China Health and Nutrition Survey. Regarding access to improved water and sanitation, there are marked differences by level of income poverty in the regions of Africa, the Americas and South-East Asia. Again, there is relatively little variation in access to improved water and sanitation in the Western Pacific Region for those in stratum B. Our results suggest that in the African Region in stratum D and the Western Pacific Region, in stratum B a high percentage of people are exposed to indoor air pollution regardless of their level of income poverty; however, strong patterning by poverty is present in other subregions.

A mixed pattern was evident for using tobacco and alcohol and being overweight (Fig. 3 and Table 2). No data were available for countries in the South-East Asia Region in stratum B or for countries in the Eastern Mediterranean Region in stratum B; only data on the prevalence of being overweight were available for countries in South-East Asia Region, stratum D. There was no apparent association between tobacco use and income poverty in countries in the African Region in stratum D, countries in the Americas in stratum B, countries in the European Region in strata B and C, and in the Western Pacific Region in stratum B, but consumption was more common among non-impoverished individuals in the African Region in stratum E and in the Americas in stratum D. Only in the Eastern Mediterranean Region in stratum D was tobacco consumption more common among those living on < US$ 1.00 or US$ 2.00 per day.

Alcohol consumption was lower among those living on < US$ 1.00 or US$ 2.00 per day in all subregions. A similar pattern was observed for being overweight, except for those countries in Western Pacific Region, stratum B. However, there was a strong association between being overweight and having a higher income only in the three poorest subregions in which being overweight or obese were least common (in the African Region in mortality strata D and E, and in the South-East Asia Region in mortality strata D, Table 2).

Fig. 4 shows the population-attributable risk percentage of poverty summed across all WHO subregions for which we had data. For example, if impoverished children had the same prevalence of malnutrition as children living on > US$ 2.00 per day, the overall prevalence of malnutrition would be 37% lower. The attributable risk estimates for indoor air pollution were of a similar magnitude and were greater for unimproved water and sanitation. Due to their much weaker and more variable associations with poverty, the attributable risks for tobacco use were smaller and fluctuated across regions.

 

 

The final counterfactual scenario illustrates what might happen for those risk factors of particular relevance to the Millennium Development Goals if the proportion of people who had an income of < US$ 1.00 a day was halved. Table 3 shows the estimates of the population-attributable risk percentage for malnutrition, unimproved water and sanitation, and indoor air pollution under this scenario assuming that those moving out of poverty gain the risk factor prevalences of those living on > US$ 1.00 a day and that those remaining on < US$ 1.00 per day retain the same risk factor prevalence. Overall, in the 11 subregions the prevalence of malnutrition among children is estimated to decrease by 6%, the prevalence of unimproved water and sanitation would decrease by 11% and the prevalence of exposure to indoor air pollution would drop by 5%. However, there was considerable heterogeneity in these results by risk factor and subregion.

 

Discussion

These results provide an approximate measure of the global burden of risk factors by absolute poverty. Our attempt to be as comprehensive as possible has some unavoidable limitations. First, the availability and quality of data for each risk factor and region were limited: a concern that is well recognized in this field (24, 25). We searched for, but did not find, adequate data on blood pressure and cholesterol by socioeconomic status for low- and middle-income regions. Tobacco and alcohol use were estimated from data on household economic consumption. Improved data should become available through the World Health Survey. (For additional information see http://www3.who.int/whs/.) We support WHO's recommendation for "improved surveillance systems and better access to global information" (26). Second, the assumption that the ranking of asset scores provides a good proxy for income rank is reasonable but deserves further study. Third, the relationship between poverty and health status is almost certainly bidirectional (10, 27): being healthy contributes to one's capability to escape poverty (28) and health service fees may tip people into poverty (16). Fourth, in deriving attributable risk estimates by poverty we have not attempted to control for confounding. Therefore, our results are better seen as an attempt at globally mapping risk factors by absolute poverty than as quantitative estimates of causal associations.

With these caveats in mind, our findings are consistent with patterns evident in other studies that have found poverty to be associated with multiple risks to health (26, 27, 29, 30). Importantly, the association of individual-level income poverty with a given risk factor often varies by subregion and would not necessarily be inferred correctly by an ecological analysis of regional poverty and the prevalence risk factors. For example, in some subregions exposure to indoor air pollution appears to be high for everyone whereas elsewhere there is more individual-level variation by income poverty.

If poverty is defined as living on < US$ 2.00 per day, and the associations reported here are regarded as mostly causal, then the percentages of malnutrition among children, of having access to only unsafe water and sanitation and of exposure to indoor air pollution that are attributable to poverty are substantial (Fig. 2). Halving the proportion of the world's population living on < US$ 1.00 per day (as in the target for the year 2015 specified in the Millennium Development Goal on poverty eradication) might reduce the prevalence of these risks by up to one-third in the African and American subregions (Table 3). This falls far short of the 50% reduction in prevalence required to achieve the Millennium Development Goal targets. The implication is that on its own economic development (at least of a magnitude to be achieved by 2015), is unlikely to be sufficient to reach the Millennium Development Goals on reducing malnutrition and improving unsafe water and sanitation. Rather we need public health programmes to be implemented in parallel with poverty reduction strategies and indeed these would support each other (15, 27, 28).

The patterns of tobacco use and obesity in our results are consistent with historical trends in the industrialized world. People in higher socioeconomic strata have tended to adopt new behaviours (e.g. cigarette smoking) early and discard them relatively quickly on learning of the health consequences, and people in lower socioeconomic strata tended to take up these behaviours later. These transitions have occurred at different times in low- and middle-income countries, and this may explain the lack of a consistent pattern in Table 2. However, the tobacco results must be interpreted cautiously as the data used in our analyses were sparse and based on household consumption data.

The term "double burden of disease" has been used to describe populations afflicted by both old-world communicable diseases and nutritional diseases and new-world chronic diseases (9, 25, 31). This double burden arises due to "a protracted epidemiological transition" (32) whereby chronic noncommunicable diseases increase in incidence (usually among the rich within the population) and communicable and nutritional diseases persist (usually among the poor within the population). During the 21st century many adverse risk factors, such as tobacco use, excessive alcohol use, and obesity, may become most prevalent among poor individuals within poor regions. There is a risk that the scourge of communicable and nutritional diseases may also persist, resulting in a double burden of disease that is concentrated not only among poor regions but also among poor individuals in poor regions.

 

Acknowledgements

Helpful comments on this paper were received from Ruth Bonita and Emmanuella Gakidou.

Funding: This study was funded by the World Health Organization (Epidemiology and Burden of Disease Unit, Global Programme on Evidence for Health Policy; Majid Ezzati, Alan Lopez and Anthony Rodgers).

Conflicts of interest: none declared.

 

References

1. Black D, Morris J, Smith C, Townsend P. Inequalities in health: report of a Research Working Group. London: Department of Health and Social Security; 1980.        

2. Feinstein J. The relationship between socioeconomic status and health: a review of the literature. Milbank Quarterly 1993;71:279-322.        

3. Kaplan J, Keil J. Socioeconomic factors and cardiovascular disease: a review of the literature. Circulation 1993;88:1973-98.        

4. Marmot M, Wilkinson R, editors. Social determinants of health. Oxford: Oxford University Press; 1999.        

5. Berkman L, Kawachi I, editors. Social epidemiology. New York: Oxford University Press; 2000.        

6. Howden-Chapman P, Tobias M, editors. Social inequalities in health: New Zealand 1999. Wellington: Ministry of Health; 2000.        

7. Blakely T, Woodward A, Pearce N, Salmond C, Kiro C, Davis P. Socio-economic factors and mortality among 25-64 year olds followed from 1991 to 1994: the New Zealand Census-Mortality Study. New Zealand Medical Journal 2002;115:93-7.        

8. Wagstaff A. Socioeconomic inequalities in child mortality: comparisons across nine developing countries. Bulletin of the World Health Organization 2000;78:19-29.        

9. Bradshaw D, Steyn K. Poverty and chronic diseases in South Africa. Tygerberg, South Africa: Burden of Diseases Research Unit; 2001. p. 123.        

10. Leon D. Common threads: underlying components of inequalities in mortality between and within countries. In: Walt G, editor. Poverty, inequality and health. Oxford: Oxford University Press; 2001. p. 58-87.        

11. Gwatkin DR. Poverty and inequalities in health within developing countries: filling the information gap. In: Walt G, editor. Poverty, inequality and health. Oxford: Oxford University Press; 2001. p. 217-46.        

12. Robinson W. Ecological correlations and the behavior of individuals. American Sociological Review 1950;15:351-7.        

13. Diez-Roux A. Bringing context back into epidemiology: variables and fallacies in multilevel analysis. American Journal of Public Health 1998;88:216-22.        

14. Gwatkin D. Health inequalities and the health of the poor: What do we know? What can we do? Bulletin of the World Health Organization 2000;78:3-18.        

15. World Health Organization. The world health report 2003 — Shaping the future. Geneva: WHO; 2003. (Also available from http://www.who.int/whr.)        

16. Wagstaff A. Poverty and health sector inequalities. Bulletin of the World Health Organization 2002;80:97-105.        

17. Ezzati M, Lopez A, Rodgers A, Murray C, editors. Comparative quantification of health risks: global and regional burden of disease attributable to selected major risk factors. Geneva: World Health Organization; In press.        

18. Ezzati M, Lopez A, Rodgers A, Vander Hoorn S, Murray C, and the Comparative Risk Assessment Collaborating Group. Selected major risk factors and global and regional burden of disease. Lancet 2002;360:1347-60.        

19. Blakely T, Hales S, Kieft C, Wilson N, Woodward A. The global distribution of risk factors by poverty: a complementary CRA. In: Ezzati M, Lopez A, Todgers A, Murray C, editors. Comparative quantification of health risks: global and regional burden of disease attributable to selected major risk factors. Geneva: World Health Organization; In press.        

20. Prüss A, Kay D, Fewtrell L, Bartam J. Estimating the burden of disease due to water, sanitation, and hygiene at the global level. Geneva: World Health Organization, Centre for Research into Environment and Health; 2001. p. 1-26.        

21. Filmer D, Pritchett L. Estimating wealth effects without expenditure, data — or tears. Washington, DC: Development Economics Research Group; 1988.        

22. Chen S, Ravallion M. How did the world's poorest fare in the 1990s? Washington, DC: The World Bank; 2000.        

23. Ravallion M, Datt G, van de Walle D. Quantifying absolute poverty in the developing world. Review of Income and Wealth 1991;37:345-61.        

24. Leon D, Walt G, editors. Poverty, inequality and health. Oxford: Oxford University Press; 2001.        

25. Sen K, Bonita R. Global health status: two steps forward, one step back. Lancet 2000;356:577-82.        

26. World Health Organization. The world health report 2002 — Reducing risk, promoting healthy life. Geneva: WHO; 2002. p. 239. (Also available from http://www.who.int/whr.)        

27. Commission on Macroeconomics and Health. Investing in health for economic development. Geneva: World Health Organization; 2001. p. 200.        

28. Sen A. Development as freedom. New York: Alfred A. Knopf; 1999.        

29. Gwatkin D, Guillot M, Heuveline P. The burden of disease among the global poor. Lancet 1999;354:586-9.        

30. Beaglehole R, Magnus P. The search for new risk factors for coronary heart disease: occupational therapy for epidemiologists? International Journal of Epidemiology 2002;31:1117-22.        

31. World Health Organization. The world health report 1999 — Making a difference. Geneva: WHO; 1999. p. 136. (Also available at http://www.who.int/whr.)        

32. Frenk J, Bobadilla J, Veda J, Cervantes M. Health transition in middle-income countries: new challenges for health care. Health Policy and Planning 1989;4:29-39.        

33. WHO Comparative Risk Assessment Working Group. Comparative risk assessment: interim guidelines. Geneva: Global Programme on Evidence for Health Policy; 2000.        

 

 

Submitted: 13 January 2004 — Final revised version received: 16 August 2004 — Accepted: 1 September 2004

 

 

1 Correspondence should be sent to this author.

World Health Organization Genebra - Genebra - Switzerland
E-mail: bulletin@who.int