12.4 Reducing health inequalities in developing countries
Oxford Textbook of Public Health
Reducing health inequalities in developing countries
Davidson R. Gwatkin
The health of the poor
Inequality in health
The practical implications of the poverty–inequality–equity distinction
Current intracountry inequalities in health status by economic class
Intracountry inequalities in use of government health care services by economic class
Trends in intercountry health status inequalities
Enhanced place of general social and economic factors
Diverse sets of ideas within the health service sector
This is a review of current professional thinking about health inequalities in developing countries and how to reduce them. It is in four parts. The first provides a brief history of recent trends in concern about health inequalities and related issues. The second is a discussion of the concept of health inequalities, and of the similarities and differences between other distributional measures in current use. The third summarizes what is known about the dimensions and magnitude of health inequalities. The fourth presents a comparable summary of current thought about how best to reduce inequalities. The review closes with a brief conclusion.
A concern about health inequalities and other distributional aspects of health status and service use has enjoyed varying degrees of attention over the years. During the 1970s and early 1980s, distributional concerns (i.e. a concern about the health status of different socio-economic groups within society as distinct from the overall societal average) were dominant in thought about international health. These concerns then receded for about a decade, from about the mid-1980s to the mid-1990s, as attention turned from equity to efficiency. Now, the pendulum has begun to swing back, and distributional concerns are on the rise.
Beginning in the early 1970s, in the field of general economic development, the traditional focus on overall per capita income growth was vigorously challenged by advocates of ‘trickle-up’ development with an emphasis on basic human needs. In the health field, a similar trend gave rise to what became known as the Health for All movement. Codified in the Alma Ata Declaration, named after the venue of a prominent conference organized by the World Health Organization (WHO) and UNICEF in 1978, the movement featured a strong emphasis on improving the health of the global poor, so that they might enjoy the health benefits already available to the better off. Given the epidemiological patterns then prevailing among the poor, inexpensive services provided by village-based paramedical personnel appeared particularly relevant for the achievement of this goal; these and other similar services came to play a central part in what became known as ‘primary health care’ (WHO 1978). Soon after, UNICEF added its strong advocacy of the ‘child survival revolution’ based on specific primary care measures (Grant 1982). In each case, the emphasis was on free services provided through government-supported health care services that were to be expanded to cover ever-increasing numbers of people.
By the mid-1980s, the situation had changed. To begin with, the overall development picture was clouded by the severe economic difficulties experienced by many poor countries, which made it clear that the cherished goal of free government health services for all was not going to be realized, at least not soon. In addition, the momentous changes in economic philosophy in the socialist countries of Eastern Europe and in China eroded the previous confidence in state-led approaches to development. These changes filtered into the health care field and began raising doubts about the appropriateness of a government’s central role in health service provision. Also, reality began to replace the euphoria of the early days of Health for All, and a closer examination of the primary health care record, rightly or wrongly, led many to question its ability to produce the dramatic benefits initially expected of it.
Thus interest began to shift from Health for All and towards what became known as ‘health sector reform’. The point can be overstated, as a concern for the distributional aspects of health status and service use continued to figure importantly in the prominent international health publications of the time, such as the World Bank’s 1993 World Development Report on health (World Bank 1993), WHO’s first World Health Report (WHO 1995) and the 1995 Annual Report of the Director of the Pan American Health Organization (Pan American Health Organization 1995). But increasingly, especially following the appearance of the World Bank’s influential Financing Health Services in Developing Countries (World Bank 1987), the health of disadvantaged groups no longer monopolized the attention of those concerned with developing country health problems. Rather, the focus moved towards sustainability, as reflected in the intensive activity on health financing that took place, and towards efficiency, as seen in the push towards greater cost-effectiveness. In epidemiological terms, the attention moved from the disease burden of the poor to that of the world as a whole and settled on the demographic–epidemiological transition, which was producing new middle and upper classes in the poor countries and whose disease characteristics were more like those of the West than those of the global poor.
Then, beginning in the mid to late 1990s, came another shift, back towards a concern for the distributional dimensions of health status and service use. An early development was the emergence of over a dozen intercountry research projects on health, poverty, and equity, supported by a wide range of donors and covering over a hundred countries (Carr et al. 1999). Another indicator is the importance given to health of the poor in recent international agency statements and policy papers.
For example, in introducing the WHO’s 1999 World Health Report, Director-General Gro Brundtland opened her review of the challenges to be addressed in order to improve the world’s health by indicating that ‘first and foremost, there is a need to reduce greatly the burden of excess mortality and morbidity suffered by the poor’ (WHO 1999a). In January 2000, she followed this up with a presentation to the WHO Executive Board outlining the strategy that WHO plans to follow in this regard (WHO 1999b) (see Chapter 12.11). More recently, a high-level WHO-appointed Commission on Macro-Economics and Health has been giving prominent attention to the needs of poor population groups in its work.
World Bank health policy statements have also been moving towards a greater emphasis on poor groups. A notable beginning was in 1997, when the Bank adopted its current strategy for work on health, nutrition, and population. According to this strategy, the Bank’s first health, nutrition, and population priority is ‘to work with countries to improve the health, nutrition and population outcomes of the world’s poor’ (World Bank 1997). This emphasis was subsequently reinforced by the World Bank’s updated overall mission statement, which began by saying that the World Bank’s principal objective is to ‘fight poverty with passion and professionalism with lasting results’ (www.worldbank.org/html/extdr/about/mission.htim).
Further movement in this direction came through the World Bank’s 2000–2001 World Development Report (World Bank 2000), which included a notable shift in the definition of poverty that such reports have employed in the past. Earlier, the reports had followed the traditional World Bank practice of defining poverty almost exclusively in financial terms. But the 2000–2001 report broke with this tradition by viewing poverty in multidimensional terms, presenting a definition under which poor health (along with inadequate education, poor nutrition, and other social dimensions of development) is placed alongside inadequate income or financial assets as a core indicator of poverty. Under this formulation, poor health becomes an integral part of poverty, rather than simply a contributor to it.
The configuration of World Bank financial support for developing county health activities has begun to show signs of shifting to conform to the new orientation just described. The principal development so far has been provision of support for health initiatives through a program of debt relief for highly indebted poor countries, undertaken in co-operation with the International Monetary Fund (IMF). Under this program, the World Bank and the IMF invited 41 particularly poor and indebted countries to prepare poverty reduction strategy papers (PRSPs) designed to demonstrate, among other things, that any additional resources made available through debt forgiveness would be spent in a manner that would benefit the poor. As of March 2001, 35 of the country governments had prepared full or interim PRSPs, and the Bank and the IMF had agreed to some US$20 billion (in net present value terms) of debt relief for 22 of them. Approximately a quarter of these funds—some US$5 billion—had gone to health programmes. Plans are currently under development for extending the PRSP process to all 78 countries qualifying for subsidized lending through the World Bank’s International Development Association (www.worldbank.org/hipc/hipc-review/hipc-review.html).
Other agencies have also been stressing the health of the poor. For example, in November 1997,the United Kingdom Department for International Development (DfID) issued a White Paper on international development committing itself to help achieve the internationally agreed target to halve the proportion of people living in extreme poverty by 2015, together with associated targets including basic health care provision and universal access to primary education by the same date (Secretary of State for International Development 1997). It followed this in November 2000 with a strategy paper outlining how it would reorient its health support in order to achieve this objective (Department for International Development 2000). Similarly, in 1999 the Rockefeller Foundation, whose Global Health Equity Initiative had already made a significant contribution, decided to make equity the principal theme for its future health work (Rockefeller Foundation 1999). In taking such steps, DfID and the Rockefeller Foundation joined organizations like the governments of The Netherlands and the Nordic countries and the many non-governmental agencies which had continued to give highest priority to reaching poor population groups throughout the earlier period when other agencies had turned more towards efficiency.
Thus, in brief, a concern for health inequalities and the health and the poor has once again come to occupy a central place in thinking about international health policies. The challenge now is for those concerned with distributional issues to take advantage of the opportunity that the current climate presents.
While the principal focus of this paper is on health inequalities, it is important to recognize that such inequalities constitute only one of several indicators of interest to those dealing with the distributional aspects of health status and service use. Two others are health equity and the health of the poor.
These three indicators or concepts are similar in some ways, different in others. However, they all share a recognition that in health, as in many other fields, societal averages typically disguise as much as they reveal. Thus their interest is not in the health conditions that prevail in society as a whole, but in the condition of different socio-economic groups within society—especially the lowest or most disadvantaged groups.
But within this shared concern lie a number of distinctions. Those interested with the health of the poor are typically concerned primarily with improving the health of that group alone, rather than with reducing differences between poor and rich. For those oriented towards equality, the principal objective is the reduction of poor–rich health differences. Those concerned with health inequities are concerned with righting the injustice represented by inequalities or poor health conditions among the disadvantaged.
These similarities and differences can most easily be understood by considering each of the three indicators and concepts in turn, and then reviewing the practical implications of thinking in terms of one or the other.
The health of the poor
A concern for poor population groups has occupied a central role in established thinking about overall socio-economic development for over two decades. It emerged in the late 1960s and early 1970s in reaction to the then dominant emphasis on overall per capita income growth rates. At the time, a concern for distribution was thought likely to detract from the overall economic growth that was considered a necessary condition for the long-term alleviation of poverty. ‘Concentrate first on overall growth’, was the prevailing view. The result might be a rise in inequality over the short term, but eventually the benefits would trickle down to the poor and, in the long run, they would end up better off than under a development strategy oriented towards their immediate needs.
The ‘trickle-up’ and ‘basic human needs’ schools of thought, which emerged to counter the view just presented, advocated dealing directly with the poor as the best means of producing sustainable growth. The many discussions about how best to define the poor population groups of concern produced two approaches.
The first, based on what is often called ‘absolute poverty’, takes a universal perspective and defines poverty in terms of a given level of income or consumption, which is equally relevant for people wherever they may be. This is usually done by defining a ‘poverty line’ as the lowest amount of money sufficient to purchase the amount of food necessary for a minimally adequate diet (and still have enough left over to buy other essentials). A well-known practitioner of this approach is the World Bank, which has devoted a great deal of time and effort to defining a suitable international poverty line and estimating the number of people living below it. The current international poverty line stands at an average per capita consumption of just over US$1.00 per day (in 1993 dollars), adjusted for purchasing power differences between countries. The consumption level of about 1200 million of the world’s population lies below this line. Almost all these people—who constitute just under a quarter of the world’s total population—live in South Asia, sub-Saharan Africa, and China (World Bank 2000).
The second approach, which is more country-specific, deals with what is frequently referred to as ‘relative poverty’. The practice here is to define the poverty line in terms of relevance for a specific society. This is typically done in one of two ways. One way, analogous to the international approach just described, is to determine how much income one needs to live decently according to some locally established definition of decency. Poverty lines of this sort are used in the developed as well as the developing world. In the United States, for example, the Census Bureau estimates that a family of four requires US$16 000 annually to purchase a minimally adequate diet and meet other basic needs, and that 12.7 per cent of the population falls below this level (Uchitelle 1999). The second approach is simply to define the national poverty line as some proportion—often arbitrarily determined—of a society’s average per capita income or expenditure. In the United Kingdom, a statistic frequently cited to document the prevalence of poverty refers to the proportion of the population (currently just under a quarter) with less than half the country’s average per capita income (Anonymous 1999).
This distinction between absolute and relative poverty carries over into the field of health. For instance, a careful reading of the World Bank policy statement on health, nutrition, and population cited above reveals an absolute-poverty orientation through its reference to a concern for ‘the world’s poor’, which is in line with the overall World Bank interest in people below the global poverty line as just described. However, others feel that relative poverty and deprivation are just as important, if not more so.
Inequality in health
While a concern for improving the health of the poor is widespread, it is by no means universally preferred. Many focus more on reducing inequalities, both in general and with respect to health in particular.
Such a focus has long occupied a particularly important place in thinking about international health issues. To say that the focus has been exclusively on inequality would be to overstate the case, for it is possible to cite expressions of concern for poverty in prominent international health documents from at least the time of the Alma-Ata Declaration (1978) onwards. However, it is rare for a prominent international health statement not to give at least equal, if not more, weight to inequality reduction. For example, at the same time as the Alma-Ata Declaration professed its concern for the unacceptable health conditions found among the hundreds of millions among the world’s poor, it also advocated primary health care because of its potential ‘to close the gap between the “haves” and the “have-nots” ‘, i.e. to lessen health inequalities (WHO 1978). The previously cited World Health Report 1995 (WHO 1995), which had a great deal to say about the health of the poor, was subtitled Bridging the Gaps, referring to the inequalities between poor and rich. A recent major WHO publication in this area emphasizes the importance of being concerned with poor–rich health inequalities, rather than simply focusing on the health of the poor alone (WHO 1996).
Similarly, health inequalities have played a much more central part than the health of the poor alone in a long European tradition of concern. For instance, the well-known 1980 Black Report in the United Kingdom was entitled Inequalities in Health (Department of Health and Social Security 1980), as was the exercise that produced its successor, the 1998 Acheson Report (Department of Health 1998). In the same vein, the 1984 targets of the WHO Regional Office for Europe (EURO) were expressed in terms of reducing poor–rich disparities. ‘By the year 2000’, said the WHO document in which these targets were presented, ‘the actual differences in health status between countries and between groups within countries should be reduced by at least 25 per cent, by improving the health of disadvantaged nations and groups’ (Whitehead 1990).
However, just as there are different approaches to poverty alleviation, so too are there various views about the most appropriate strategies for the reduction of inequalities. Illustrative of the issues that arise in discussing the reduction of health inequalities are the following questions.
What dimensions of inequality matter most? The most traditional approach has been to think of differences in health status according to an individual’s income or economic standing. However, the economic dimension is by no means the only one that matters, and some would consider other dimensions even more important. Gender inequalities in health status have received a great deal of attention in recent years. Ethnic inequalities in health have been of particular concern in many areas, such as South Africa and the United States. Education and occupation have also been widely used as a basis for dividing populations in assessing intergroup health differentials, although often more as a proxy for economic status than as indicators of interest in their own right. Yet another approach might be called ‘pure’ health inequality, i.e. the ordering of people on the basis of their health status, from most to least healthy regardless of income or any other attribute, for the purpose of measuring health diversity in a society (Gakidou et al. 2000). In so doing, people applying this approach are drawing on a long tradition of studies with respect to income distribution.
How is inequality to be measured? There are almost as many statistical definitions of inequality as there are statisticians, and the various definitions can produce very different interpretations of the same situation or trend. Until recently, one particular measure—the Gini coefficient—has been dominant, at least in economic thinking, supplemented by comparisons between the poorest and richest population quintiles (or between people above and below the poverty line) when the data available were insufficient for the calculation of the Gini coefficient. While the Gini coefficient probably remains the most frequently used indicator even now, its position is slipping. In health, as in economics, several other disparity indicators are under active consideration (Anand et al. 2001), with no clear consensus about a preferred alternative.
What aspects of inequality are most important? There are many different views. Some would argue for looking at inequalities in health status as the outcome that counts; others favour focusing on health services as the determinant of health status that health professionals can most easily influence. Within each of these two streams of thought are further distinctions. Health status, for example, can be determined either through a physical examination or through self-assessment. (The two approaches can produce quite different results, in that people found to be relatively unhealthy through a physical examination do not always consider themselves to be less healthy than people whose health was determined by examination to be considerably better.) With respect to health services, there are distinctions between (a) use and financing, (b) among public, private non-profit, and private for-profit services, and (c) between preventive and curative services. Health services that come out ahead in one of these respects may lag from another perspective.
Should the focus be local or global? A great deal of attention is currently being paid to inequalities within countries. However, there is also strong interest in some quarters in differences among countries and world regions.
Poverty and inequality, as described above, are both primarily empirical concepts. Equity, by contrast, is a normative concept (a question of values) and closely associated with the concept of social justice. When applied to health, equity has traditionally been most often linked to the reduction of inequalities. Thus, one of the most widely cited definitions of health inequity is that it ‘refers to differences in health which…are considered unfair and unjust’. In a similar vein, the WHO/EURO document on health equity cited above indicated that ‘equity requires reducing unfair disparities…’ and that ‘pursuing equity in health and health care development means trying to reduce unfair and unnecessary social gaps in health and health care…’ (Whitehead 1990).
However, equity need not be exclusively a matter of reducing inequalities. It can also be associated with poverty, as one could argue that it is unjust to allow people to continue living in poverty when adequate resources are available within the society at large to lift them out of it. Such a link figures prominently in general thinking about social justice, and it also appears in writings on health equity.
A particularly well-known example of poverty-oriented general thought about equity is the ‘maximin’ principle of distributive justice posited by John Rawls. That principle and others like it call for resources to be distributed in such a way that the worst-off people in society (i.e. those occupying the ‘minimum’ position) have the maximum possible amount of gain. What happens to the better-off through such a pattern of resource distribution is extraneous to the maximin principle (Rawls 1971). A variation on this theme, as applied to health, would consider any health gains among the rich in the course of implementing efforts to improve the health of the poor as welcome side-benefits, rather than regrettable because of the dilution in inequality reduction that they represent (Marchand et al. 1998).
While not many equality-oriented advocates of health equity seem prepared to go this far, almost all incorporate at least traces of such a poverty-oriented equity definition in their statements. The traces are to be seen most clearly in the tendency of equality-oriented discussions to disavow interest in one of the arguably more effective potential ways of reducing poor–rich health inequalities—assassination of the rich. Rather, the focus of all known inequality-oriented health equity proposals is on lessening poor–rich differences through special efforts to improve the health of the poor—a focus that makes the proposals sound suspiciously similar to what one might wish to do under a poverty-oriented health equity approach.
Thus, for instance, the previously cited inequality-oriented definition of health equity referring to the inequalities of health that are unjust and unfair was developed in conjunction with the WHO/EURO health equity objective, which called for a reduction in health disparities by improving the health of the disadvantaged (Whitehead 1990). The WHO’s 1996 health equity document, while giving primacy to poor–rich health differences, also called for ensuring an adequate standard for the entire population, noting that, ‘for some,“equity” means that all social groups should have a basic minimum level of well-being and services’ (WHO 1996).
However, regardless of whether one considers health equity to be related more to equality or poverty, the introduction of normative or social justice considerations also raises questions.
When is an inequality unfair? Not always, certainly. It is quite possible to imagine a situation marked by health inequalities that are not necessarily inequitable. One example is an inequality that is irremediable (Whitehead 1990). Another might be two population groups with similar incomes but marked differences in life expectancy attributable to different lifestyles. If the less healthy group adopts its lifestyle in full awareness of the risks involved, the resulting differences in life expectancy might be said to be simply a reflection of differences in the social preferences of the two groups rather than any fundamental inequity. Or, to illustrate the same point with a more general example, if two individuals are in fact unequal in capacity, equal treatment would be unfair to the more capable of the two. In such a case, equity might well call for unequal treatment. In other words, equity and equality are by no means synonymous and need to be carefully distinguished from one another.
On what basis can one decide when the resources in a society are adequate to alleviate poverty? ‘Adequacy’ is not a binary concept, such that there is one level of resource availability above which availability is totally adequate, and below which it is completely inadequate. Rather, there is a spectrum running from a total lack to infinite availability of resources, often with no obvious cut-off point along the way. Also, perceptions can differ—resources that seem adequate to one person may not be so to another.
The practical implications of the poverty–inequality–equity distinction
What has been said so far provides a basis for suspecting that, in general, there are only limited practical implications in adopting one or other of the three approaches covered here. As has been noted, even those who seem furthest apart (those giving highest priority to reductions in poor–rich health inequalities in the name of equity, and those concerned with improving the health of the poor) end up sounding rather similar, once one realizes that the approach preferred by advocates of inequality reduction looks primarily to improvements in the health of the disadvantaged.
A more careful look reinforces this view that the poverty–equality–equity distinction is often largely academic. The most obvious situation is in a low-income country where the most cost-effective measures available for the improvement of health in the society as a whole are also those that are especially beneficial to the poor. As pointed out in the 1993 World Development Report (World Bank 1993), the contents of minimum service packages that feature such measures—management of the sick child, prenatal and delivery care, family planning, etc.—are especially relevant for low-income groups. Where this is the case, adoption of the approach that is most sensible for the poor is also more beneficial for the poor than for the rich, and can thus be expected to produce a reduction in poor–rich differences.
However, the record would not be complete without noting that there are at least some circumstances where an interest in improving the health of the poor can imply a different approach from that resulting from a concern for inequality reduction. Two examples can illustrate the point. One concerns inter-regional resource allocations by international agencies, and the other deals with disease priorities, whether at the global or the national level.
The inter-regional resource allocation example involves looking at the implications of different approaches as to where an international assistance agency might logically apply its health resources. Consider three alternatives.
An absolute-poverty approach According to the World Bank figures cited earlier, some 90 per cent of the world’s 1200 million people living below the poverty line live in Asia and Africa (World Bank 2000). This being the case, an international agency guided by an absolute-poverty objective would wish to put virtually all of its health resources into those regions. There would be much less justification for working in Latin America, and practically none at all for health activity in the Middle East or Eastern Europe, where hardly anyone is so poor as to lie below the international poverty line.
A relative-poverty approach Relative poverty exists in every country. From this perspective, there could be as strong a justification for supporting pro-poor health activities in one region of the world as in any other.
An equality approach Assuming that most of the existing health inequalities observed in the developing world are also inequitable and that inequality reduction interventions are equally effective, an equity approach would imply a particularly high priority for countries where health inequalities are greatest. Recent research points to the existence of large country-to-country differences in the degree of health inequality, which in turn suggests that some countries deserve much more attention than others from an equity perspective. According to one recent study (Wagstaff 2000), Brazil, Nicaragua, South Africa, and Nepal have large health-status inequalities and would thus be of high priority, while health-status inequalities are quite low in Ghana, Pakistan, and Vietnam which would accordingly merit a low priority.
With respect to the second set of examples, dealing with disease priorities, the available information is unfortunately inadequate to permit citation of ‘real-world’ experiences. However, the basic point can be demonstrated through two schematic illustrations, one from a global and one from a national perspective.
Global disease priorities A global institution focusing on absolute poverty would logically devote primary attention and resources to communicable diseases, as they are the dominant causes of deaths and disability among the global poor (Gwatkin and Guillot 1999; Gwatkin et al. 1999). In an institution concerned with relative poverty, there would be a case for a much broader concern. Such an institution would be involved not only with the poor in Africa and Asia, but also with the disadvantaged populations in Eastern European countries, among whom non-communicable diseases may well be the dominant problem.
Disease priorities within advanced developing and transition countries While communicable diseases are dominant among the global poor, chronic diseases in advanced developing and transition countries are, as just noted, likely to be responsible for the majority of deaths and disability among the poor—but, in all likelihood, for a smaller percentage of the poor than of the rich. The implications of such a situation can be illustrated by reference to a country where non-communicable diseases cause 60 per cent of deaths among the poor and 90 per cent of deaths among the rich. From a burden-of-disease perspective, such figures point to non-communicable diseases as a natural focus for a programme concerned with poverty alleviation, as such diseases cause the majority of deaths. But such a focus, if introduced on a society-wide basis, could well lead to an increase in inequality. This is because non-communicable diseases are even more important for the rich than for the poor, so that the benefit to the rich of any general evenly distributed decline in non-communicable diseases would be correspondingly greater. Thus, in a situation like this, burden-of-disease considerations would argue for the highest priority to be given to one type of disease (i.e. non-communicable diseases) from the perspective of improving the health of the poor, and to a different type of disease (i.e. communicable diseases) from an inequality-reduction perspective.
The second example is obviously oversimplified, ignoring cost-effectiveness and targeting considerations that may well be more important than disease burden factors in the establishment of health service priorities. However, while lessening the example’s relevance for ‘real-world’ decision-making, the oversimplification is of value in facilitating understanding of the basic point that remains valid despite it: there is not an inevitable congruity between national-level policy prescriptions that are optimal for improving the health of the poor and those that are best for reducing health inequalities.
While the contents of the preceding section make it clear that a focus on inequalities is by no means the only one of relevance for approaching health differentials in developing countries, health inequalities remain of considerable interest and will be the topic of what follows. As space limitations prevent adequate coverage of the full range of health inequalities that might be considered, the discussion will focus on inequalities by socio-economic status. It will deal with the three following inequalities as illustrations of the different types that exist: (a) current intracountry inequalities in health status by economic class, (b) current intracountry inequalities in the use of government health care services, and (c) trends in intercountry health status inequalities.
Current intracountry inequalities in health status by economic class
Until recently, data limitations have prevented the direct examination of intracountry health inequalities by any of the three indicators typically employed by economists: income, consumption, or assets (wealth). Instead, as noted earlier, indicators such as father’s occupation or mother’s education have served as proxies for economic status, in addition to pointing to attributes of interest in themselves. This is changing, however, and a number of efforts are under way to provide more direct measures of economic status as a basis for assessing intracountry health differentials. Of particular relevance for the present discussion are findings from three multicountry comparative study programmes: country health and poverty reports, Living Standards Measurement Survey estimates, and WHO estimates.
Perhaps the most extensive of the programmes is the country health and poverty report project organized by the World Bank (Gwatkin et al. 2000). This project features preparation of tabulations, by asset or wealth quintile, for approximately 30 health, nutrition, population status, and service use indicators. The tabulations, designed primarily to provide basic distributional information for the use of policy-makers, are currently available for 44 countries in Africa, Asia, Latin America, and the Near East.
The information presented in the reports is drawn from household data collected through the Demographic and Health Survey programme sponsored by the United States Agency for International Development. This well-known programme of comparative country studies, typically covering 5000 to 10 000 households in each country investigated, is oriented especially to the collection of information about vital events and maternal/child health. It is considerably less strong with respect to information about economic status, as it contains no questions about income or consumption. However, its standard individual and household survey instruments include a number of questions about household assets—availability of electricity, possession of consumer goods (e.g. bicycle, radio), flooring material, source of drinking water, etc. Using principal components analysis, these can be combined into a single index of household assets or wealth which is of interest in its own right and approximates reasonably well the consumption measures that economists tend to prefer (Filmer and Pritchett 2001).
In preparing a report, a country’s population is divided into quintiles on the basis of the asset index, and the value of each health, nutrition, or population indicator is tabulated for each population quintile. The status indicators include infant and under-5 mortality rates, total and adolescent fertility rates, and such commonly used indices of malnutrition as stunting and low weight-for-age. Typical of the service indicators are immunization rates, medical treatment for diarrhoea and acute respiratory infections among children, use of antenatal and professional delivery care, and contraceptive prevalence.
A summary of the available information about infant and under-5 mortality, to which the current discussion will be limited, appears in Table 1. The figures are expressed in terms of two indicators. The first is a poor–rich ratio, i.e. the ratio of infant or under-5 mortality in the poorest population quintile to that in the richest quintile. The second is a concentration index, which is a measure similar to the well-known Gini coefficient that is commonly used in measuring income inequalities. As in the case of the Gini coefficient, the value of the concentration index can range from –1.0 (if all infant or under-5 deaths occur in the poorest population quintile to +1.0 (if all deaths are in the richest quintile) (Wagstaff et al. 1991).
Table 1 Intracountry disparities in infant and under-5 mortality
The unweighted poor–rich ratio for all countries together is about 1.9 for infant mortality and about 2.1 for under-5 mortality. This suggests that, on average, the newly born child in the poorest population quintile of a developing country is roughly twice as likely to die in infancy as is a child born into the richest quintile. The unweighted mean concentration indices are of the order of –0.11 or –0.12. This is modest compared with the range of about –0.2 to –0.5 typically found with respect to consumption as measured by the Gini coefficient. However, such modesty is not unexpected, given that mortality rates are much more tightly bounded variables than are consumption measures.
The data also point to considerable country-to-country variation. For instance, the low–high ratio for infant mortality ranges from 1.11 in Namibia to 4.17 in Bolivia. Differences in the concentration index are also large. The range is from –0.003 in Namibia to –0.251 in Brazil.
Regardless of which index is used, the data appear to support two generalizations about the variation that exists. First, intracountry inequalities seem larger for child (1–5) than for infant (0–1) mortality. This can be inferred from the modestly higher level of under-5 (i.e. 0–5) mortality relative to infant (0–1) mortality found for both inequality indices in each region (and also in 33 of the 40 individual countries covered). Infant mortality is nested within under-5 mortality and typically contains well over half the deaths on which under-5 mortality rates are based. Removal of less unequally distributed infant deaths in order to produce a child mortality rate would thus be likely to increase further the already larger poor–rich differences seen in the 0 to 5 rates. This would be consistent with current understanding about the pathogens to which poor children are exposed during the immediate post-weaning period. Second, intracountry socio-economic inequalities in infant and child mortality appear smaller in sub-Saharan Africa than in other parts of the world; in Latin America, they seem somewhat larger. This is readily visible from Table 1, where the average values for both the poor–rich ratio and the concentration index for sub-Saharan Africa (1.7 to 1.8; –0.08 to –0.10) are lower than the global mean (1.9 to 2.1; –0.11 to –0.12), and lower still than the Latin American average (2.7 to 3.0; –0.14 to –0.17). This can be explained rather easily for Latin America, given the high levels of inequality found in that region (Deininger and Squire 1996) and initial findings from other work that confirm the existence of a direct relationship between income and health inequality (Wagstaff 1999). The low degree of health inequality in sub-Saharan Africa is more difficult to understand, in the light of recent findings that income is almost as unequally distributed there as in Latin America ( Deininger and Squire 1996).
However, these or any other conclusions that might be drawn from the infant or under-5 mortality data in the country health and poverty reports must be conditioned by at least two considerations. The first is the existence of rather large standard errors in many cases, which means that the concentration indices are statistically different from zero (as measured by a 5 per cent confidence interval) in only about half the countries covered. This occurred even though the tabulations were based on births during the 10 years prior to the survey, in order to increase the sample size. The second is the presence of regular or monotonic declines in mortality from the poorest to the richest quintiles in only approximately half the countries. The anomalies are usually quite small, often consisting of a high point in the next poorest quintile that is suggestive of under-reporting of mortality among the poorest population groups. However, in a few African countries there is a pronounced ‘spike’ in the middle quintiles that remains to be explained.
The results of the Living Standards Measurement Survey estimates are roughly in line with those discussed above. The study in question is by Adam Wagstaff of the World Bank and the University of Sussex (Wagstaff 2000). It is based on datasets for nine developing countries. Eight of the datasets are from the World Bank Living Standards Measurement Survey programme, which covers approximately 25 developing countries. The survey instruments are oriented towards the measurement of household consumption, so that Wagstaff was able to use consumption as his economic indicator, rather than the asset index featured in the work based on the Demographic and Health Survey described above. On the other hand, because of the less reliable mortality data available through the Living Standards Measurement Survey, Wagstaff was forced to employ estimation methods that were not necessary with the Demographic and Health Survey data.
Six of the nine countries that Wagstaff covered were among the 40 in the work based on the Demographic and Health Survey, making it possible to compare the concentration indices for these countries produced by the two studies. This comparison is presented in Table 2.
Table 2 Concentration indices for infant and under-5 mortality within six developing countries as measured by different studies
On average, the Wagstaff concentration indices are somewhat smaller than those resulting from the country health and poverty report exercise, although this is not the case for each country. Perhaps more significantly, there is an overlap between the 5 per cent confidence intervals for the two sets of estimates in 11 of the 12 cases presented (i.e. in all six countries for infant mortality, and in all countries except Vietnam for under-5 mortality). By this standard, the two sets of estimates can be considered mutually consistent, although, given the rather large size of the confidence intervals of the two studies, the standard can accommodate considerable variation among figures qualifying for mutual consistency status under it.
It remains to be seen whether such consistency exists between the country health and poverty reports and the Wagstaff estimates on the one hand, and the WHO estimates on the other. This third set of estimates, under preparation by the WHO, employs a very different method from those of the other two. Rather than relying on household data, as in the two previous studies, the WHO exercise is based on a cross-country dataset. The datasets used consist of two figures for each country included: the percentage of the population residing below the poverty line, and the country average for a particular health indicator of interest. The application of a statistical technique known as ecological inference makes it possible to derive from such data an estimate for the level of the health indicators among people above and below the poverty line in each country. For example, if one knows the percentage of the population above and below the poverty line and the average infant mortality for each country in a dataset, it is possible to estimate the infant mortality rate among people above and below the poverty line in each of the countries (Jamison 1997; WHO 1999a).
The first data from this exercise have been published, in an annex to the 1999 edition of the World Health Report (WHO 1999a); they provide estimates of the under-5 and 15–59 mortality for males and females and of tuberculosis prevalence for 49 developing and transition countries.
The figures most nearly comparable with those of the country health and poverty report and Wagstaff exercises discussed earlier are the under-5 mortality estimates. Here, the unweighted average poor–non-poor ratio for all 49 countries (male and female combined) is about 6.1:1. (Weighting by population size, in lieu of the number of births for which information is not provided, gives a ratio of 5.4:1.) About 30 per cent of the people in the 49 countries live below the absolute poverty line (whether calculated on a weighted or unweighted basis). Thus under-5 mortality among the poorest 30 per cent or so of the population is of the order of five to six times greater than that among the highest 70 per cent. This appears to be well above the poorest 30 per cent–richest 70 per cent ratio implied by the previously cited country health and poverty report finding that, on average, under-5 mortality among the poorest 20 per cent of a developing country’s population is about twice as high as among the highest 20 per cent. This in turn suggests that the use of country-level household data, as in the country health and poverty report and Wagstaff exercises, is likely to produce intracountry differentials that are considerably smaller than indirect inferences based on country-level comparisons.
For the moment, however, this can be considered no more than a suggestion, to be confirmed or disproved on the basis of further more careful examination of the methodology and findings of all three study programmes. Also to be determined is the relevance for this particular issue of the common view that microlevel (household-level) data are more reliable—or, perhaps more accurately, less unreliable—than macrolevel (country-level) information for policy-oriented analyses.
Intracountry inequalities in use of government health care services by economic class
Concomitant with the rapid increase in research on intracountry socio-economic differentials in health status, as described in the preceding section, has come a growth in concern about differentials in access to health services. As access is difficult to define and measure, most work in this area has focused on the related issue of health service use. (Use is associated with access but differs from it in that one can have access to services but not use them for any of several reasons—for example, a lack of need for medical care because of continuing good health.)
Of particular interest has been the extent to which government health services have been able to reach disadvantaged population groups. This is being increasingly studied through the application of a technique called ‘benefit–incidence analysis’, borrowed from the field of public finance. Benefit–incidence analysis can be seen as an equity-oriented analogue of cost-effectiveness analysis used to study efficiency; while cost-effectiveness analysis is oriented towards determining how much output a health system produces per unit of input, benefit–incidence analysis seeks to assess how a health system’s outputs are distributed across socio-economic classes.
Specifically, the benefit–incidence approach measures the financial subsidies accruing to different socio-economic groups through the use of government health services. It combines two types of information. The first consists of data from household surveys concerning the number and socio-economic status of people using different kinds of government services when ill. From these, it is possible to determine the number of per capita visits to a given type of government service, during a specified time period, for each socio-economic group of interest (such as income quintile of the population). The second is information from government financial reports about the total cost of the services in question (net of any income received from user fees), and from service statistics concerning the number of people using those services. These data permit an estimate of the unit cost of the different services, i.e. the average cost of providing the services in question to a single individual. The number of per capita visits to a given type of service is then multiplied by the unit cost of that service, and the results are aggregated to produce an estimate of the total financial benefit accruing to each socio-economic group through government expenditures on the range of services covered.
Information is currently available about benefit–incidence surveys on health that have been undertaken in 23 countries. This information is summarized in Table 3. The most reliable data come from sub-Saharan Africa, the site of a comparative study of seven countries organized by the World Bank (Castro-Leal et al. 2000). As each of the country exercises drew on household datasets collected through similar surveys undertaken by the Bank’s Living Standards Measurement Survey programme, problems of intercountry comparability appear significantly lower than is the case with respect to the other studies for which findings are available.
Table 3 Percentage of financial subsidy from government health services accruing to poorest and richest quintiles of the population: regional averages
The findings from these studies make it clear that the rich benefit financially more than the poor from government health services in sub-Saharan Africa. This was the case in all seven of the countries covered by the principal systematic research effort thus far undertaken. The difference was particularly notable with respect to hospital services, but even primary care normally benefited the rich somewhat more than the poor. On average, the richest 20 per cent of the population received well over twice as much financial benefit as the poorest 20 per cent from overall government health service expenditures. In all but two of the seven countries, the richest 20 per cent also gained more than the poorest 20 per cent of the population from primary care expenditures.
In Asia, the situation appears mixed. On average, overall government health care expenditures in the five countries with available data appear to favour the rich slightly more than the poor. However, this is an average of very dissimilar situations: three (Indonesia, Mongolia, Vietnam) in which the rich gain far more than the poor, and two others (Malaysia, the Philippines) in which the poor receive larger financial benefits than the rich. It should be noted, however, that these findings may be less secure than that just cited for Africa. In part, this arises from the uncertain comparability of the studies cited, but more from the fact that the benefit–incidence tradition is less well established in Asia than in Africa or Latin America. Also, no fully published findings are available for the two largest countries, China and India.
To judge from the two countries in central Europe with available study data (Bulgaria and Romania), the situation appears similar to Africa. In each country, the rich gain more than the poor from primary as well as from hospital care. Overall, the financial benefit that government health services convey to the rich is nearly twice as large as that gained by the poor.
The situation in Latin America appears different. Information available for seven countries suggests that the poorest quintile gains more than the richest quintile in all but one (Brazil). On average, the poor in these countries receive twice as much benefit as the rich. However, these figures need to be viewed with caution, especially in comparison with those for Africa, cited above, for two reasons. One is that government-delivered health services represent a much smaller percentage of total government health expenditures than elsewhere. The health benefits that flow through social security systems, on which Latin American governments tend to spend almost as much as on health services they provide directly, are also important. (For example, according to one recent review about 17 per cent of government health expenditures were through social security systems, compared with 16 per cent for services provided directly by central governments and 9 per cent for local government services (R. Suarez-Berenguela, Pan American Health Organization, unpublished work, 1998).) As such programmes focus on formal sector employees, they tend to be oriented towards the middle and upper classes; when their benefit incidence is taken into account, the overall impact of government health care expenditures could well be regressive. A second consideration is technical. Many of the Latin American studies appear to be based on the benefits accruing to households rather than to individuals. As poorer families tend to be larger than rich ones, use of the household as the basis of analysis provides an impression of greater progressivity than do findings that refer to individuals.
In addition to the specific considerations indicated with respect to Asia and Latin America, the findings just described are affected by several general characteristics of the benefit–incidence approach that need to be taken into account. Five are particularly worthy of note.
In accordance with the tradition of the benefit–incidence literature, the findings are presented in terms of absolute rather than relative benefit (e.g. pesos per capita rather than percentage of per capita income). In relative terms, the poor are likely to benefit more than the rich because the incomes of the richest 20 per cent are normally many times greater than those of the poorest 20 per cent.
The conclusions are derived from estimates that cover only expenditures. These could differ quite significantly from conclusions that look at the revenue side as well, and measure only net benefits and incidence, i.e. the amount a given income group gains from government health expenditures relative to the amount of taxes that the group pays for those services. The potential difference results from the fact that, while the poor may gain less from government health services, they may pay substantially lower taxes as they live outside the organized economy.
The distribution of financial benefits covered by the figures is not necessarily the same as the distribution of therapeutic benefit, which is arguably more relevant. The two would not correspond when, for example, the services that the poor receive consist principally of primary interventions that, although inexpensive, are quite effective in treating illness, while the expensive tertiary care received by the rich is of limited therapeutic value.
The data provided cover only government curative health care services (hospitals, health posts, etc.). They do not include expenditures on government health education or preventative health programmes such as immunization campaigns. It remains unclear how much of a bias is introduced by the omission of such activities, as they typically represent a rather small portion of total government health expenditures and do not necessarily reach the poor all that more effectively than curative programmes. However, their omission at the very least significantly increases the possibility of inaccuracy in the benefit–incidence information available.
Figures for the financial benefit accruing to any specific economic group provide no guidance on how well that benefit corresponds to need. For example, to say that the financial benefit accruing to the poor is twice that accruing to the rich sounds progressive, but it may not be if the poor need, say, four times as large a financial benefit as required by the rich in order to compensate for the greater degree of illness that they experience.
Trends in intercountry health status inequalities
The data currently available can be used to assess trends in some types of country health inequalities, for example by gender or geographic area. However, they are not sufficient to support any assessment of trends in intracountry inequalities in health status or service use by socio-economic status, which is the focus of the current discussion.
However, it is possible to investigate trends with respect to inequalities by socio-economic status across countries. This can be done by looking at time series data for the average levels of infant mortality and life expectancy that are regularly published by such agencies as the United Nations Population Division, the World Bank, the WHO, and others.
What follows are two such investigations. One compares time trends in groups of countries. The second looks at changes over time in the distribution of infant mortality and life expectancy across individual countries.
Trends among country groups.
Table 4 presents the summary results of the first investigation, which is a simple exercise based on World Bank data for 1970, 1980, and 1990. The purpose is to provide an initial sense of changes over this period in the size of the infant mortality and life expectancy gaps between the rich and poor parts of the world. To this end, two types of comparison are presented: (a) between the rich industrialized countries and all others, and (b) between the same group of rich industrialized countries and the world’s poorest countries (i.e. omitting the more advanced developing countries). In each case, the definitions of country groups are those of the World Bank. Intergroup boundaries are expressed in terms of per capita incomes, and countries were assigned to the appropriate group on the basis of the situation prevailing in 1990.
Table 4 Trends in health inequality between country groups, 1970–1990
The results of the exercise vary according to the statistical measure and the indicator of health status used. In both types of comparison, infant mortality differences decline over time in absolute terms but rise in relative terms. For example, the comparison of the richest and poorest country groups shows that the absolute difference in infant mortality falls from about 125 in 1970 to about 95 in 1990. However, the infant mortality gap expressed in relative terms rises. While infant mortality in 1970 was about 6.5 times higher in the poorest countries than in the richest countries, it was over 11 times higher by 1990. The same is true for the comparison between the richest and all other countries. For life expectancy, both types of country group comparison show a decrease in inequality in both absolute and relative terms.
Trends among individual countries
Table 5 summarizes results of the second individual country exercise which is based on a World Bank dataset containing average infant mortality rates and life expectancies in all major countries of the world for 1970, 1980, and 1990. It features the measurement of changes in the degree of inequality among the countries in the dataset over the period from 1970 to 1990 through the application of a set of standard disparity indices: the standard deviation, the slope index of inequality, the relative index of inequality, the index of dissimilarity, and the concentration index. Two variants of the exercise were undertaken: the first covering all 130 countries, and the second dealing with only the developing and transition countries (approximately 100).
Table 5 Trends in health inequality among countries, 1970–1990
As with the earlier exercise, the results are mixed. This is particularly the case with respect to infant mortality, where in each variant the poor–rich country gap widens according to some indicators and narrows with respect to others. The trend in life expectancy is somewhat less ambiguous, narrowing in all but one of the 10 cases presented, although in some cases the figures show that the degree of narrowing was marginal.
All in all, the findings reported above are far from conclusive. The one possible exception is the moderately clear difference between infant mortality and life expectancy trends—the former are mixed, and the latter generally point to a narrowing gap. This hints at a narrowing in poor–rich country mortality differences at older ages that is faster than, and possibly in a direction different from, the trend at younger ages. But it is no more than a hint, as one cannot exclude the possibility that the difference is a statistical artefact, attributable to the subtleties of the relationship among mortality levels at different ages included in the life expectancy measure.
In brief, about all one can say with confidence is that any change in intercountry mortality inequalities between 1970 and 1990 period has not been dramatic enough to show up clearly through the application of standard disparity measurement approaches to the data available. Any more definitive statement will have to await more careful and sophisticated data analyses than have been possible within the limited scope of the current review.
Thought about how to reduce the inequalities discussed in the preceding section has thus far lagged well behind the growing desire to reduce them that was described at the outset. The nature of this thought has also changed from that which prevailed in the late 1970s and early 1980s, when a concern for health inequalities and the health of the poor was last prominent in international health circles. As noted earlier, the emphasis at that time was on a single health-service-based approach—a set of relatively simple and inexpensive community-based interventions collectively known as primary health care. Current thought differs from this in two ways. First, at the conceptual or general strategic level, a much larger role is being accorded to general social and economic factors lying outside the health service sector. Second, at the operational level within the health sector, there exists a much more diverse set of ideas for helping the poor— many of them adapted from other sectors.
Enhanced place of general social and economic factors
The first of these changes—the more central place currently accorded general social and economic factors—can be seen in the documents of many agencies. Two of the more prominent documents, one each from the WHO and the World Bank, can serve as illustrations.
The WHO document is the four-part health and poverty strategy recently proposed by the Director-General (WHO 1999b). The first part is ‘acting on the determinants of health by influencing development policy.’ The second is ‘reducing risks through a broader approach to public health’, i.e. including not just basic public health services, safe and adequate food, clean water, and sanitation, but also the reduction of violence and environmental hazards, conflicts, and natural disasters. The more traditional health sector approaches come only later, in the third and fourth parts, which are ‘focusing on the health problems of the poor,’ and ‘ensuring that health systems serve the poor more effectively’.
The World Bank document is the chapter on health, nutrition, and population in its Poverty Reduction Strategy Sourcebook (www.worldbank.org/poverty/strategies/sourctoc.htm), a massive document prepared in 1999–2000 as a reference work for developing country officials preparing the poverty reduction strategy papers described earlier. The chapter on health, nutrition, and population in this multifaceted volume is organized according to a framework that calls attention to three different sources of influence on health outcomes.
Households and communities At the household and community level, three sets of factors are described:
household actions and risk factors, such as dietary, sanitary, and sexual practices
household assets, human (such as education) as well as physical and financial
community factors, such as the physical surrounding and degree to which the different members of the household feel responsible for one another’s well being.
Health system and related factors Here, too, three sets of factors are laid out:
health service provision, especially considerations like accessibility, quality, and price
health financing, such as insurance or public subsidies for essential services
supply in related sectors, covering issues such as the availability and quality of food, water, and energy.
Government policies and actions Two sets of factors are presented:
health sector policies/actions, such as policies on essential drugs or towards private sector services
policies/actions in other sectors, such as agriculture and education, which can have indirect but powerful side-effects on health.
In this World Bank framework, as in the WHO statement, the potential contribution of health services is duly noted, but such services constitute only one of the many factors that determine health status. For instance, they occupy only eight of the boxes that appear in the diagrammatic presentation of the framework, and only around 20 to 25 per cent of the text explaining the framework is devoted to them. Much more prominence is given to the other broader influences on health.
The extent to which such broader thinking in the headquarters of international agencies will affect what happens at the operational level remains unclear. At present, the impact appears limited. For example, to judge from reading the health chapters of country PRSPs, current plans do not look very different from the health strategies of two decades ago. There are occasional references to things like basic health insurance and private sector involvement that one would not have seen earlier, but typically the PRSP chapters feature measures like prenatal care, vaccinations, and essential drug provision that would have been at the centre of a strategy paper written at the height of the primary care movement of the early 1980s (Gwatkin 2001b). However, it is far too early to pass any final judgement, since the situation could well change over the years ahead if international thinking continues to be as influential in the formulation of developing country health policies as it has in the past.
Diverse sets of ideas within the health service sector
The second difference between earlier and current thought, concerning the diverse sets of ideas at the operational level within the health sector, can be illustrated with reference to four approaches that are currently in use or under active discussion. None of the four is exclusive to health. Rather, each represents the application to health of approaches being used in efforts to reduce inequalities in overall socio-economic status.
Two of the four—targeting and participatory approaches—are well established. The other two—protection against the financial risks of illness and the statement of health objectives in distributional terms—are much newer. Each appears sufficiently valuable or promising to merit careful consideration. However, none is a panacea and, on the basis on present knowledge, it is far from certain that even the four together would be strong enough to offset fully the current tendency of health programmes to reach the better-off more effectively than the disadvantaged, as described above in the section on inequalities in health service use.
Targeted health services
The expression ‘targeting’ refers to a set of techniques used to increase the percentage of benefits from a particular intervention that flows to the poor. Health is just one of the many sectors where these techniques are employed. Other common ones include the identification of individuals or groups qualified to receive subsidized food supplies, to obtain employment through government-supported public works programmes, or in some cases to qualify for subsidized loans through microcredit programmes.
Targeting has many supporters, but it also has detractors who are of two types. One type argues that, on the basis of findings like those cited below, targeting techniques do not work very well in distinguishing poor from rich and reaching the former. The second type is concerned with the ‘top-down’ nature of targeted measures, which deliver free or subsidized services or commodities. As such, they are thought to represent hand-outs that invite dependency rather than foster the initiative among beneficiaries seen as central for long-term poverty alleviation or health improvement (Sen 1994).
There are many different targeting techniques available, and many ways of categorizing them. One of the more common categorization approaches features a distinction between ‘individual’, ‘direct’, or ‘narrow’ targeting on the one hand, and ‘indicator/characteristic’, ‘indirect’, or ‘broad’ targeting on the other.
The former type refers to efforts to identify poor individuals and see that as much of the service concerned reaches as many of them as possible. The objective is to come as close as possible to the goal of 100 per cent coverage with zero leakage, i.e. seeing that all the poor are served and that all of those served are poor. The latter type of targeting deals with attributes rather than individuals. Rather than trying to identify individuals who are poor, for instance, it might feature the provision of services in slum areas in anticipation that the great majority of recipients will be poor. In doing so, it recognizes that it will not be able to reach all of the poor (some of whom live outside slums), and that at least some of those receiving services will not be poor (as not everyone living in a poor area is poor). However, it accepts these limitations as prices worth paying in order to attain two important advantages. One is administrative practicality or efficiency, through avoidance of the considerable effort typically required to distinguish between poor and non-poor individuals with even a modest degree of precision. The second is political—the belief that poverty-oriented service programmes are much more likely to gain the political support needed for survival if members of the middle and upper classes gain enough from them to have an incentive to defend their continuation.
These different targeting methods are not mutually exclusive and are often used in combination. There is no known instance of their achieving or even approaching perfection, but there is considerable evidence that, when employed with care and determination, they can increase the percentage of service benefits that accrue to the poor. The best-known instance of this comes from a set of studies covering nearly 50 government and private service programmes (including but not limited to health) in Latin America. These studies found that in the 18 carefully targeted programmes, some 70 to 75 per cent of benefits accrued to the poorest 40 per cent of households, compared with 55 to 60 per cent of the benefits from government primary health care and education projects and 30 to 35 per cent of the benefits from untargeted food distribution programmes. All three of the principal targeting methods used worked equally well, and the administrative costs of targeting were typically under 10 per cent of total programme expenses (Grosh 1994).
There are many specific targeting techniques available, each with unique features, strengths, and weaknesses. Three of them—individual targeting, geographic targeting, and disease targeting— can serve to illustrate the potential and limitations of the different options..
Certain forms of individual targeting have long been practised in the health field. A common example is the use of simple diagnostic measures in screening programmes to identify individuals at high risk of serious illness and thus in need of priority attention. More recently, a great deal of attention has been given to the identification of poor individuals in order to exempt them from users’ fees introduced in government health facilities of developing countries during health sector reforms.
The experience of efforts to identify individuals for this purpose has varied widely. There have been more reported failures than reported successes. But there have been successes as well, and rather clear differences in the design of the more and less successful efforts that can serve as guidelines for the design of future initiatives.
Towards the effective end of the effective–ineffective spectrum lie some of the Latin American projects noted above. Also instructive is the case of Thailand, which has been offering free medical care to low-income groups, through an initiative known as the Low Income Support Program, since 1975. The programme has been modified numerous times as the government has gained experience. In its present form, it is open to Thai families with monthly incomes of under 2000 baht, which constitutes about 25 per cent of the country’s population. Local officials of the Ministry of Home Affairs determine who qualifies under the income criterion just noted and issue identity cards, valid for 3 years, to those families found eligible. The cardholders are exempted from fees at government health facilities. The cost is covered by a special allocation to the service-providing facilities from the Ministry of Public Health, which allocates about 8 per cent of its total budget to this end. The programme now covers some 11 million people, or about 20 per cent of the country’s total population—equivalent to about 80 per cent of the eligible poor population. Independent surveys indicate that about 20 per cent of those covered are non-poor who are improperly included. Adjusting for this suggests that, in total, the programme covers about 65 per cent of Thailand’s poor, with a leakage rate of 20 per cent (Khoman 1997).
At the other end of the spectrum are a series of casual efforts, especially in Africa. For example, according to a recent questionnaire study of 26 developing countries (most but not all in Africa), almost three-quarters of the countries responding reported at least some kind of official policy to exempt the poor from user fees. But, said the study organizers, ‘there were numerous informational, administrative, economic, and political constraints to effective implementation of these exemptions’ (Russell and Gilson 1997). For instance, in most countries policy guidelines on whom to exempt were quite vague, local exemption administrators found it very difficult to assess household incomes, and/or the potentially eligible were often unaware that exemptions were available. The findings of other reviews are similar: in general, found one such review, individual targeting efforts in sub-Saharan Africa ‘have been compromised by a variety of difficulties, including excessive leakage, overly subjective exemption criteria, informal identification and verification procedures, and excessive costs’ (H. Waters, unpublished work, 1995).
A plausible explanation of why some individual targeting programmes have been more successful than others emerges from a study of 29 such efforts in health. The study suggested two factors that differentiated between the nine efforts considered successful and the 20 that were not. One was the existence of clear, formal, and explicit criteria for eligibility. The second was the determination of eligibility by someone other than the health service provider, such as a village council (Willis 1993). A third factor might be hypothesized from the Thailand case cited above—the provision of a mechanism and of resources to reimburse health service providers for income foregone in providing free services.
The idea behind geographic targeting is straightforward—the poorer the area to which resources are allocated, the greater the likelihood that the individuals who benefit from those resources will be poor. Like individual targeting, geographic targeting is a generic approach, in the sense of being equally applicable to activities in almost any sector rather than being specific to health. Geographic targeting can be applied with widely varying degrees of precision. The amount of improved accuracy resulting from increases in precision will depend upon the spatial pattern of the distribution of poverty within the society concerned.
Perhaps the simplest and least precise form of geographic targeting is the emphasis often placed on rural areas, where the available information suggests that poverty is in general considerably more prevalent than it is in the cities. For example, World Bank data for the world’s low-income countries indicated that in the mid-1990s the agricultural workers who constituted 66 per cent of the labour force produced only 27 per cent of the economic output. The same appeared to be the case in the middle-income countries, where 32 per cent of the workers were in the agricultural sector which accounted for 11 per cent of the national output (World Bank 1998).
Other more precise forms of targeting involve a focus on poor states or provinces, or subdivisions within these. Typically, these are identified on the basis of data for per capita income or output produced by government statistical offices. Several countries, particularly in Latin America, have sought to be even more precise by identifying villages or other small communities that are particularly poor. This has typically been done through the construction of a basic needs or similar index based on questions contained in national censuses such as literacy rates, education levels, and quality of housing.
Recently, there has been experimentation with techniques for identifying small areas on the basis of measures more obviously and directly related to consumption, traditionally the indicator preferred by economists concerned with poverty. The techniques concerned involve combining data from in-depth sample surveys, which ask many questions from a relatively small number of households, and from national censuses, which ask a few questions about all the households in a country. The basic idea is to identify those questions on the household survey instruments that are also included in the national census, and best predict the consumption levels of the households covered. Average values for the questions thus identified can then be calculated for individual villages covered by the census data in order to predict the average consumption levels prevailing in those villages, and the poorest villages can be selected on this basis. The use of such techniques is still in infancy, but initial experience with them has been promising in such widely varying settings as Burkina Faso, Ecuador, and South Africa (H. Alderman et al., World Bank, unpublished work, 1999; Bigman et al. 2000; Hentschel et al. 2000).
Finding the areas with the greatest number of poor people is only part of the story, however. Equally important is the development of services that can reach the poor in those areas. This can be challenging, as poor areas frequently lack the capital and physical infrastructure necessary for effective programme initiation.
Targeting by disease
As with geographic targeting, two steps are involved in disease targeting. The first is to identify the diseases that are most important for the poor, and the second is to develop delivery mechanisms that reach the poor in order to lessen the impact of those diseases.
With respect to the first of these steps, a recent assessment has established that, at the global level, the diseases of greatest importance to the poor tend to be communicable. For example, among the 20 per cent of the global population living in the poorest countries, communicable and related conditions were responsible for about 59 per cent of all deaths in 1990, compared with 32 per cent caused by non-communicable diseases and about 9 per cent attributable to accidents and injuries. In that year, communicable and related diseases were responsible for over three-quarters of the global poor–rich mortality gap. Any acceleration in the rate of communicable disease decline, evenly distributed across all global social groups, would benefit the poorest 20 per cent some 10 times more than the richest 20 per cent. This stands in sharp contrast to the impact of a comparable acceleration in the fall of non-communicable diseases which would benefit the richest 20 per cent about three to four times as much as the poorest 20 per cent (Gwatkin and Guillot 1999).
Thus, at the global level, there is strong justification for emphasizing improved approaches for dealing with communicable diseases. However, the available evidence also indicates that there is a great deal of country-to-country variation relevant for the establishment of national policies. This is particularly the case in demographically advanced developing and transitional countries. In such settings, where overall mortality levels are generally low, it is quite possible for non-communicable conditions to be more important than communicable diseases even among the poor. At the same time, however, communicable diseases are still likely to be more important for the poor than for the rich, in the sense of being responsible for a larger minority of death and disability in the former group than in the latter.
The point can be illustrated by a study performed in Mexico in 1994 which provided estimates for rural and urban regions, using a combined mortality–disability measure called disability-adjusted life years (DALYs). Even in the rural population, presumably consisting primarily of poor people, more DALYs were lost as a result of non-communicable conditions than because of communicable diseases (44 per cent versus 37 per cent). However, communicable diseases, even though the minority cause of death and disability in rural areas, were still a considerably larger minority in the rural group than in the urban group (37 per cent versus 25 per cent of total DALYs lost) (Frenk et al. 1998).
In such a situation, which appears typical of that prevailing in much of Latin America and Eastern Europe, any country-wide emphasis on non-communicable diseases would be highly relevant for the poor. However, it would be more likely to increase than to decrease poor–rich disparities as non-communicable diseases are likely to be still more important for the rich. If the poor are to benefit more than the rich, it would be necessary to supplement disease targeting with some other targeting approach (such as individual or geographic targeting) to increase the proportion of total benefits reaching the needy. (One would also wish to incorporate cost-effectiveness considerations into any programme design in order to ensure the production of adequate health benefits overall.)
The situation is similar with respect to the second of the two issues posed above, concerning the effectiveness of intervention delivery systems in serving the poor. It is not safe to assume that a focus on diseases relevant for the poor will in itself lead to improvements in that group of a country’s population, as those diseases also affect upper- and middle-income groups at least to some extent. It is quite possible to imagine a situation in which intervention programmes against diseases relevant for the poor fail to develop the outreach capacity required to reach the neediest, so that the programme benefits are limited primarily to better-off groups. The available information suggests that this is happening in many countries, although by no means all, and that, where this is happening, disease targeting will have to be supplemented by geographic, individual, or some other type of targeting if the poor are to be served.
The information containing such suggestions comes from data for the several intervention measures covered in the country health and poverty reports referred to above with respect to infant and under-5 mortality. Particularly relevant are the quintile-specific data for interventions directed against communicable diseases among children—diseases that, as suggested above, are especially important for the global poor. The interventions include immunizations against measles, diphtheria, whooping cough, and tetanus, and curative treatment of diarrhoea and acute respiratory diseases. Quintile-specific data are also available for antenatal care and deliveries attended by trained medical personnel, which deal with a condition which is not communicable, at least not in the traditional sense, but which turns out to be concentrated particularly among disadvantaged groups.
These data show that the situation varies greatly from country to country. For example, consider the data for the percentage of children immunized against all the diseases noted above, which are summarized in Table 6. Looking first at the percentage of children immunized among the poorest 20 per cent of the population, there is only a modest variation among the different regions—from 35 per cent in Africa to about 43 per cent in Asia, the Near East, and North Africa. However, the country-to-country variations within each region are dramatic, particularly in Sub-Saharan Africa where the range is from under 5 per cent (Niger) to over 70 per cent (Malawi, Zambia, Zimbabwe). A review of the poor–rich ratios and concentration indices, which permit a comparison of immunization rates among the poorest 20 per cent with those in higher socio-economic groups, reveals notably smaller disparities in Latin America than in other parts of the developing world. However, there are also very large country-to-country variations within each region. The reasons for the variations remain to be determined.
Table 6 Levels and intracountry inequalities in immunization rates
Participatory approaches to health service planning and delivery
As noted earlier, targeting is not universally popular among development professionals. To some, it is seen as an outgrowth of the tendency of many public health professionals to determine the most important health issues on the basis of the epidemiological considerations that are important to them; this is then done without pausing to determine whether the priorities thus established correspond with the priorities of the intended beneficiaries. An alternative approach, preferred by people holding this view, is to involve the prospective beneficiaries from the beginning. This means determining their priorities as the basis for project development and developing modes of service delivery that they are likely to find congenial. Such a ‘consumer-oriented’ or ‘client-oriented’ approach is often supported on pragmatic as well as conceptual grounds, through reference to the likelihood of increased utilization resulting from the provision of services that people want rather than those that outsiders think they need.
This approach has been present in development thought since at least the 1950s when it went by the name of development on the basis of prospective beneficiaries’ ‘felt needs.’ Strong traces of it have also existed in international public health circles, among community medicine specialists who share many of the same concerns about technocratic epidemiology just expressed. The concern of such specialists can be seen in the strong emphasis on community-based approaches appearing in the Alma Ata Declaration of 1978.
At present, the approach is gaining strength. Work is going on in many venues, but recent trends are most clearly illustrated with reference to developments within the World Bank. While that institution remains better known for its technocratic orientation than for its participatory instincts, an incipient interest in participation can be seen through three streams of work. Two of these are analytical—identifying the poor through participatory poverty assessments, and determining what the poor want in the way of health services and what they think of the services they currently receive. The third is operational—increasing reliance on non-governmental or community organizations rather than government agencies to deliver health and other services.
Participatory poverty assessments
The basic premise underlying the participatory poverty assessment (PPA) approach is that villagers are much better able to define poverty and identify the poor among them than are outside government authorities. In a sense, PPAs can thus be seen from a technocratic perspective as a way of identifying people who deserve the highest priority under an individual targeting programme, although the information gathered goes well beyond this in helping determine how communities perceive poverty.
PPAs usually employ relatively non-quantitative methods based on guided discussions with village members. The discussions typically proceed in two stages. The first stage consists of reaching consensus about the appropriate indicators of wealth and poverty. The second consists of identifying individual families or people as rich or poor on the basis of the agreed-upon indicators.
By encouraging people first to establish criteria for poverty before identifying individuals who were poor, the PPA seeks to lessen the problem commonly attributed to village-based beneficiary approaches—the alleged proclivity of village leaders to identify their relatives, friends, and political supporters as those most deserving programme benefits. Just how well it succeeds in this regard remains unclear.
This approach has so far been applied in nearly 50 countries, especially in Africa, by the World Bank (Robb 1999), and no doubt in countless other locations by other agencies. Among the best-known is Tanzania, where discussions like those described above took place in 87 villages in different parts of the country and involved over 6000 village residents (Narayan 1997). In each village, the residents produced maps locating the dwellings of those they considered to belong to the poorest of five wealth categories. In the Dodoma region, for example, the poor were defined as (a) mostly old or disabled, (b) people who lived by begging, (c) childless women, and/or (d) people who were mentally unfit. In the Kilimanjaro region, participants defined the poor as people who were (a) landless, (b) did not plant crops, (c) depended upon relatives, and (d) lived in substandard housing. The percentage identified as poor by applying these criteria in the study villages closely approximated the percentage defined as living below the poverty line through more formal consumption-based methods, although there may have been differences in the particular individuals who qualified as poor.
Another multicountry PPA has been undertaken in connection with preparation of the World Bank’s 2000–2001 World Development Report, which is focusing on poverty. The exercise is known as Consultations with the Poor (Narayan et al. 2000a,b). It is in two parts: the first is a summary of some 80 earlier PPAs from around the world involving interviews with over 60 000 people; the second consists of 23 country studies, covering about 20 000 poor individuals. These new studies contain a great deal of information about how the poor perceive health and health services. They also cover a wide range of topics, including many which, strictly speaking, are more closely associated with the beneficiary assessments described in the following section than with PPAs.
While PPAs are oriented primarily (although not exclusively) towards determining just who is poor, beneficiary assessments focus more on what the poor want in the way of services and/or what they think of the services offered to them. Given this focus, they might be considered as a type of market research, designed to produce information of value to programme managers which can help them design programmes corresponding more closely to consumers’ preferences.
Like PPAs, beneficiary assessments also tend to rely primarily on qualitative methods. Commonly used approaches include informal conversational interviews, participant observation, and focus group discussions. Upon occasion—as with the Voices of the Poor exercise described earlier—a particular qualitative study will include elements of both a beneficiary assessment and PPA, an indication of the fluidity of the distinction between them.
Beneficiary assessments appear to have been undertaken rather frequently in connection with health programmes, at least in connection with those receiving World Bank support. For example, a review of World Bank experience in 1995 listed some 20 beneficiary assessments made in the course of developing or evaluating health, population, and nutrition projects, more than twice the number prepared in any other sector of World Bank activity (Salmen 1995). An informal reading of subsequent World Bank project documents suggests that there have been many more since then. A systematic review of work done by individuals and agencies not associated with the World Bank would no doubt uncover many more. Also, in addition to studies considered as strictly defined beneficiary assessments, there exists a vast array of more general village-based anthropological and sociological literature about health which contains information about the health priorities of the poor that is of obvious relevance for programme development.
Virtually all of the beneficiary assessments known to have been performed to date have been undertaken for project preparation or evaluation rather than for formal publication. Such studies tend to disappear after having served their intended purpose, and this results in a problem of physical unavailability that makes it impossible to provide an overview of findings to see if there are common themes across countries. However, one can give a flavour of what the studies show through a few illustrations from recent studies that remain extant. For example, a focus group discussion in Bangladesh revealed that the major deterrent to the use of government facilities was the unpredictable volume of under-the-table payments requested by facility employees, thus making it difficult for the prospective clients to know in advance whether they could afford the services. A study conducted by participant observers in Lesotho found that the village health worker programme was having little impact because the workers were emphasizing preventive services which, while in line with professional thought, were of little interest to villagers. The villagers actually wanted curative services that the workers were not qualified to provide (Salmen 1995). In Ethiopia, policy-makers learned from a beneficiary assessment that the antenatal services they offered were being poorly used because of a cultural belief that pregnant women considered it improper to admit to any pain or discomfort during pregnancy (Salmen 1995). Anecdotal information based on such surveys in a number of countries suggests that the availability of drugs (and injections) is frequently a factor to which poor patients attract considerable importance.
Alternative delivery mechanisms
Frustration over the bureaucratic and political constraints that prevent government ministries from offering services that reach the poor effectively is giving rise to an interest in identifying alternative mechanisms. Typically, this means looking to agencies that have a larger degree of participation by, or at least are much closer to, the poor clients who are the intended beneficiaries. Most attention has thus far gone to two types of agency—non-government organizations and social funds.
With respect the former, there is a widespread belief that non-governmental organizations are much better than government health programmes at identifying and serving the poor. This is not inevitably the case: Tanzania and Zambia represent instances where the economic level of people served through the non-governmental health sector appears higher than that of patients in government facilities (F. Diop, unpublished work, WHO, 1997; Munishi 1997). However, where non-governmental organizations do reach the poor effectively, government grants to or contracts with non-governmental organizations for the delivery of services represent an alternative to efforts to target poor individuals through the government service system. This is not always easy for governments to do, given the resistance that can be expected from their employees who stand to lose if this practice becomes widespread. External agencies find it easier to move in this direction and appear increasingly interested in doing so. In Bangladesh, to cite perhaps the most prominent example, a massive volume of donor support has permitted the Bangladesh Rural Advancement Committee, which now employs some 25 000 to 30 000 people, to become one of the country’s principal providers of health and other services in poor rural areas.
Another alternative to direct government health (and other) service delivery is a social fund, a relatively new form of quasi-governmental financial intermediary that channels funds directly to small-scale projects for poor and vulnerable groups operated by local governments, non-governmental organizations, or community groups. Social funds have been created in well over 30 countries, especially in Africa and Latin America, as a way around rigidities in traditional government ministries which prevent them from reaching the poor effectively. Thus far, the Inter-American Development Bank and the World Bank have made loans of over US$3.5 billion to more than 50 countries to support social funds (S. Jorgensen, World Bank, unpublished work, 1999). Roughly one-third of social fund support has traditionally gone to health, nutrition, and population (S. Carvalho, World Bank, unpublished work, 1995).
Data limitations prevent any clear assessment of just how well social funds have in fact reached the poor. But such information as exists suggests an overall picture quite similar to that presented above with respect to individual targeting in more traditional government programmes, i.e. social funds which adopt explicit carefully designed and implemented approaches to beneficiary identification work much better than those which do not (S. Carvalho, World Bank, unpublished work, 1995).
Protecting the poor from the financial consequences of illness
A third and more recent approach to dealing with health inequalities differs from the two just described in taking what might be called a financial orientation, i.e. the reason for being concerned with the health of the disadvantaged is not an interest in health per se, but rather the financial implications of poor health for the people concerned. Lessened health inequality through assistance to the disadvantage is seen less as an end in itself than as a means towards the end of reduced financial inequality. The rationale for such an approach lies in the belief, increasingly supported by empirical evidence, that the path out of poverty is not smooth. Families do not simply rise steadily out of poverty and, once out of it, follow a steadily upward trajectory. Rather, they often fall back, sometimes temporarily and sometimes permanently, as a result of external shocks over which they have little control.
Often, these external shocks are economic in nature, as in the case of the financial crisis that affected much of Asia in the mid-1990s. Sometimes, the shock is caused by civil strife, as in parts of Africa and the former Yugoslavia. But poor health also appears to play a significant role, partly because of the high costs of medical care, and partly because of the income lost in those cases where the sick person is a major breadwinner for the family.
The two most complete studies of this issue undertaken to date are from Bangladesh. One, carried out in an urban slum, found that over 40 per cent of people in the poorest quartile of the population had missed work during the preceding month, and that this had cost them an average of nearly 75 per cent of their family income in that period (Pryer 1989). The other, a survey of more than 60 villages, found that poor health was the principal cause of 15 to 20 per cent of the cases in which previously non-poor people had slipped into poverty. Many others avoided falling below the poverty line, measured in consumption terms, by drawing down their family assets (Sen 1997).
The three other known studies focused on medical costs, and dealt only indirectly with foregone income. Two cover poor rural areas in China. Each of these studies found that medical costs borne by poor families in the era following the collapse of rural medical co-operatives is extremely high. One study reported that medical costs represent over 15 per cent of the total expenditures among poor families. One-third of the families had to dip into assets in order to cover the costs; 70 to 90 per cent of the sick who did not seek care gave the high cost as the reason (W. Fu et al., unpublished work, 1995). The other Chinese study found that the average cost of a one-time hospitalization equalled well over the annual income of a poor peasant, (b) costs of other care received by the poor cost them about 10 per cent of their incomes, and (c) 80 per cent of poor people needing hospitalization did not obtain it because of the expense. The study concluded that ‘illness is a major reason why peasants become impoverished or (why) those who have made their way out of poverty revert to being poor’ (Expert Committee for the Study of Policy and Administration 1996). The third study is from Mexico where over 4 per cent of the poorest 30 per cent of the population experienced medical expenses equal to over half of their incomes during the first 6 months of 1992—a percentage far higher than experienced at higher income levels (Frenk et al. 1998).
Beyond this are indications that, regardless of how important or unimportant the financial consequences of ill health might be when measured empirically, the poor worry a lot about them. In the interviews conducted with 20 000 poor people throughout the world in connection with the Consultations with the Poor study, the fear of ill health because of its potential financial implications arose again and again. In these interviews, ill health was the most frequently cited reason why families became impoverished (Narayan 2000).
Findings like these argue for a concern with protection against severe or catastrophic illness, whose costs, in the form of medical expenses and/or lost income, can easily drag families down into poverty or prevent their rising out of it. Several mechanisms for providing such protection are currently under consideration. They include the development of risk-pooling or insurance plans and subsidized or free secondary medical care for income-earning adults.
Establishing health objectives in distributional terms
A final and even more recent approach is oriented towards health planners and policy-makers at the national and international levels. Its focus is on the way that health objectives are established in macrolevel economic and health development plans. Typically, such health objectives are established in terms of society-wide averages—reduction of a country’s overall infant mortality rate by x per cent, an increase in a society’s average life expectancy by y years, etc. While averages like these are informative about conditions prevailing in a society as a whole, they say nothing about the inequalities that prevail in it. As a result, they are poorly suited for the development of strategies or the assessment of progress towards inequality reduction, or the improvement of conditions among the poor.
A recent exercise, based on a set of international development goals established in connection with the Global Social Summit conference held in Copenhagen in 1995, suggested that this is considerably more than an academic quibble. The health goals appearing in the set included a call for a two-thirds reduction in the average infant mortality rate in all countries by the year 2015. When examined from an intracountry inequality perspective, using the data on interquintile differences presented earlier, there proved to be several combinations of decline in different quintiles that could result in both a two-thirds reduction in overall infant mortality and a significant widening in interquintile differences. Under some circumstances, the two-thirds reduction in infant mortality could be attained without any significant decline in the rate prevailing in the poorest quintile (Gwatkin 2000).
This has led to an interest in finding an alternative way of stating health objectives in terms of, say, an x per cent reduction of the difference in infant mortality between the poorest and richest quintiles, or a y-year improvement in life expectancy among people below a society’s poverty line. A statement of health objectives in such terms would be consistent with current thinking about economic development objectives, which are increasingly being expressed in poverty or distributional terms rather than with regard to societal averages. (The economic goals of the Social Summit, for instance, refer not to increasing average per capita incomes, as was the earlier convention, but rather to decreasing the numbers of people living in poverty.)
A restatement of goals would not in itself save lives, however. Restated goals would be of value only to the extent to which they succeed in directing the attention of policy-makers towards health inequalities and the health of the poor, thereby preparing the way for the development of effective interventions to deal with those issues. This implies a need not just for restated goals, but also for the use of these goals to shape policies and programmes.
This review began with the argument that a concern for health inequalities has recently begun to rise to prominence after a decade during which it attracted little attention. In closing, it is appropriate to ask what will be required to ensure that it remains at the centre of attention and leads to effective action.
Had this question been posed as recently asd 2 or 3 years ago, the temptation would have been strong to respond that effective action would require progress with respect to all three of the other topics covered in this review. That is, there appeared to be a clear need for closer agreement on what should be the governing concept for activities in this area—reducing health inequalities or improving the health of the poor, a better understanding of the magnitude and dimensions of health inequalities, and improved programme approaches for reaching the poor and improving their health, to supplement the approaches discussed in the section on reducing inequalities.
Now, however, the first and second of these three issues appear to constitute somewhat less formidable obstacles to progress than they did earlier. As noted in the section on concepts, the first issue, i.e. the distinction between reducing health inequalities and improving the health of the poor, is often (although admittedly not always) largely academic, with only limited policy implications. With respect to the second issue, the lack of understanding of health inequalities, the research findings reported in the section on inequalities have brought about a sharp increase in the amount of knowledge that exists.
This is not to argue that either of these issues has yet been by any means fully resolved, just that the amount of progress with respect to each of them has been considerably greater than that with respect to the third issue—the lack of proven policy and programmatic approaches. Thus, this third issue now stands out as the one most in need of attention.
Of course, overlying this issue is the much larger question of political commitment. If the importance currently being attached to poverty-oriented overall development is correct, major progress in reducing health inequalities will require not just a reorientation of the health sector, but also a broad social determination to alleviate poverty.
Anand, S., Diderichsen, F., Evans, T., Shkolnikov, V., and Wirth, M. (2001). Measuring disparities in health: methods and indicators. In Challenging inequities in health: from ethics to action (ed. T. Evans, M. Whitehead, F. Diderichsen, A. Bhuiya, and M. Wirth), pp. 49-67. Oxford University Press, New York.
Anonymous (1999). Labour’s crusade. Economist, 25 September.
Bigman, D., Dercon, S., Giullaume, D., and Lambotte, M. (2000). Community targeting for poverty reduction in Burkina Faso. World Bank Economic Review, 14, 167–93.
Carr, D., Gwatkin D., Fragueiro D., and Pande, R. (1999). A guide to country-level information about equity, poverty, and health available from multi-country research programs. World Bank, Washington, DC.
Castro-Leal, F., Dayton, J., Demery, L., and Mehra, K. (2000). Public spending on health care in Africa. Do the poor benefit? Bulletin of the World Health Organization, 78, 66–74.
Deininger, K. and Squire, L. (1996). A new data set measuring income inequality. World Bank Economic Review, 10, 565–91.
Department for International Development (2000). Better health for poor people: strategies for achieving the international development targets. Department for International Development, London.
Department of Health (1998). Independent inquiry into inequalities in health (Acheson Report). HMSO, London.
Department of Health and Social Security (1980). Inequalities in health. Department of Health and Social Security, London.
Expert Committee for the Study of Policy and Administration (of the Chinese Ministry of Health) (1996). Policy proposals on enhancing health care development and reform in poor rural areas. Health Work Bulletin, 6 November.
Filmer, D. and Pritchett, L. (2001). Estimating wealth effects without expenditure data—or tears: an application of educational enrollments in states of India. Demography, 38, 115–32.
Frenk, J., González Block, M., and Lozano, R. (1998). Seis tesis equivocadas sobre las politicas de salud en el combate a la pobreza. Este Pais, 84, 28–36.
Gakidou, E., Murray, C., and Frenk, J. (2000). Defining and measuring health inequality: an approach based on the distribution of health expectancy. Bulletin of the World Health Organization, 78, 42–54.
Grant J. (1982). The state of the world’s children, 1982–83. Oxford University Press, New York.
Grosh, M. (1994). Administering targeted social programs in Latin America: from platitudes to practice. World Bank, Washington, DC.
Gwatkin, D. (2000). Health inequalities and the health of the poor: What do we know? What can we do? Bulletin of the World Health Organization, 78, 3–18.
Gwatkin, D. (2001a). Poverty and inequalities in health in developing countries. In Poverty, inequality, and health: an international perspective (ed. D. Leon and G. Walt). Oxford University Press.
Gwatkin, D. (2001b). The need for equity-oriented health sector reforms. International Journal of Epidemiology, 30, 720–3.
Gwatkin, D. and Guillot, M. (1999). The burden of disease among the global poor:current situation, future trends, and implications for strategy. World Bank and Global Forum for Health Research, Washington, DC.
Gwatkin, D., Guillot, M., and Heuveline, P. (1999). The burden of disease among the global poor. Lancet, 354, 586–9.
Gwatkin, D.R., Rutstein, S., Johnson K., Pande, R., and Wagstaff, A. (2000). Socio-economic differences in health, nutrition, and population in Bangladesh. World Bank, Washington, DC. (See also comparable publications covering Benin, Bolivia, Brazil, Burkina Faso, Cameroun, Central African Republic, Colombia, Comores, Côte d’Ivoire, Dominican Republic, Ghana, Guatemala, Haiti, India, Indonesia, Kenya, Kyrgyz Republic, Madagascar, Malawi, Mali, Morocco, Mozambique, Namibia, Nepal, Nicaragua, Niger, Nigeria, Pakistan, Paraguay, Peru, Philippines, Senegal, Tanzania, Togo, Turkey, Uganda, Vietnam, Zambia, and Zimbabwe.)
Hentschel, J., Lanjuow, J., Lanjouw, P., and Poggi, J. (2000). Combining census and survey data to trace the spatial dimension of poverty: a case study of Ecuador. World Bank Economic Review, 14, 147–65.
Jamison, D. (1997). The health of the poor: global and country-specific estimates for some indicators. Presented at World Bank Health, Nutrition, Population, and Poverty Seminar, 16 December.
Khoman, S. (1997). Rural health care financing in Thailand. In Innovations in health care financing, (ed. G. Schieber). World Bank, Washington, DC.
Marchand, S., Wikler, D., and Landesman, B. (1998). Class, health, and justice: health and society. Milbank Memorial Fund Quarterly, 76, 449–67.
Munishi, G. (1997). Private health care in Tanzania: private health sector growth following liberalization in Tanzania. International Health Policy Program Working Paper, International Health Policy Program, Washington, DC.
Narayan, D. (1997). Voices of the poor: poverty and social capital in Tanzania. World Bank, Washington, DC.
Narayan, D. (2000). What do the poor have to say about health, nutrition, and population? Presented at World Bank Health, Nutrition, Population, and Poverty Seminar, 1 February.
Narayan, D., Patel, R., Schafft, K., Rademacher, A., and Koch-Schulte, S. (2000a). Voices of the poor: can anyone hear us? Oxford University Press, New York.
Narayan, D., Chambers, R., Shah, M., and Petesch, P. (2000b). Voices of the poor: crying out for change. Oxford University Press, New York.
Pan American Health Organization (1995). The search for equity: annual report of the director. Pan American Health Organization, Washington, DC.
Pryer, J. (1989). When breadwinners fall ill: preliminary findings from a case study in Bangladesh. IDS Bulletin, 20, 49–57.
Rawls, J. (1971). A theory of justice. Harvard University Press, Cambridge, MA.
Robb, C. (1999). Can the poor influence policy? Participatory poverty assessments in the developing world. World Bank, Washington.
Rockefeller Foundation (1999). A new course of action. Rockefeller Foundation, New York.
Russell, S. and Gilson, L. (1997). User fee policy to promote health service access for the poor: a wolf in sheep’s clothing? International Journal of Health Services, 27, 359–79
Salmen, L. (1995). Beneficiary assessment: an approach described. World Bank Environment Department Paper 23, World Bank, Washington, DC.
Secretary of State for International Development (1997). Eliminating world poverty: a challenge for the 21st century. HMSO, London.
Sen, A. (1994). The political economy of targeting. In Public spending and the poor: theory and evidence (ed. D. van de Walle and K. Nead), pp. 11–24. World Bank/Johns Hopkins University Press, Baltimore, MD.
Sen, B. (1997). Health and poverty in the context of development strategy: a case study on Bangladesh. Macroeconomics, Health, and Development Series, No. 26, WHO, Geneva.
Uchitelle, L. (1999) More cash in hand, but poorer. International Herald Tribune, 19 October.
Wagstaff, A. (1999). Inequalities in infant and child mortality in the developing world: How large are they? How can they be reduced? Presented at World Bank Health, Nutrition, Population and Poverty Seminar, 30 March.
Wagstaff, A. (2000). Socioeconomic inequalities in child mortality: comparisons across nine developing countries. Bulletin of the World Health Organization, 78, 19–29.
Wagstaff, A., Paci, P., and van Doorslaer, E. (1991). On the measurement of inequalities in health. Social Science and Medicine, 33, 545–57.
Whitehead, M. (1990). The concepts and principles of equity and health. Document EUR/ICP/RPD/414, WHO Regional Office for Europe, Copenhagen.
WHO (World Health Organization) (1978). Primary health care. Report of the International Conference on Primary Health Care, Alma-Ata, USSR, 6–12 September 1978. WHO, Geneva.
WHO (World Health Organization) (1995). World health report 1995: bridging the gap. WHO, Geneva.
WHO (World Health Organization) (1996). Equity in health and health care: a WHO/SIDA initiative. Document WHO/ARB/961, WHO, Geneva.
WHO (World Health Organization) (1999a). World health report 1999: making a difference. WHO, Geneva.
WHO (World Health Organization) (1999b). Poverty and health. Report by the Director General. Document EB105/5 presented to the 105th session of the WHO Executive Board. WHO, Geneva.
Willis, C. (1993). Means testing in cost recovery of health services in developing countries. Phase I: review of concepts and literature, and preliminary field work design. Health Financing and Sustainability Major Applied Research Paper 7, Abt Associates, Bethesda, MD.
World Bank (1987). Financing health services in developing countries: an agenda for reform. World Bank, Washington, DC.
World Bank (1993). World development report 1993: investing in health. Oxford University Press, New York.
World Bank (1997). Sector strategy: health, nutrition, and population. World Bank, Washington, DC.
World Bank (1998). 1998 world development indicators. World Bank, Washington, DC.
World Bank (2000). World development report 2000–2001: attacking poverty. Oxford University Press, New York.