Distributional Implications of Alternative Strategic Responses to the Demographic-Epidemiological Transition—An Initial Inquiry
Davidson R.Gwatkin
INTRODUCTION
Over the past decade, transitions in the developing world’s demographic and epidemiological situations have become increasingly obvious. Mortality and fertility have been declining, and disease patterns have been shifting. Illness and death among the young caused by the diarrhea-malnutrition-pneumonia triad have been progressively giving way to newer configurations dominated by chronic and degenerative diseases among adults and the elderly.
These changes raise the possibility that the time has come for a corresponding shift in developing countries health strategies. No longer can one simply assume that it remains adequate to continue stressing oral rehydration, immunization, and related approaches. As the epidemiological and demographic transitions proceed, it would seem reasonable to anticipate a need to become increasingly concerned with preventing strokes and heart attacks as well.
Were societies epidemiologically and demographically homogeneous, the case for such a shift would be straightforward. But societies are not homogeneous. Different groups in them suffer from different kinds of diseases at different ages. A change in focus from one set of diseases and age groups to another could benefit some groups at the expense of others.
This question of who gains and who loses from a change in priorities is
Davidson R.Gwatkin is director of the International Health Policy Program. The author wishes to thank Jun Zhu for his effective research assistance and the many readers of earlier drafts for their valuable suggestions.
of particular interest in health because of the egalitarian tendencies that have typified professional thinking in the field during recent years.1 From an egalitarian perspective, any change is to be welcomed to the extent that its benefits accrue to the poor, and to be resisted to the degree that the gains go instead to the better-off and leave the poor at an even greater disadvantage than before.
For those adhering to this viewpoint, is a shift in priorities from communicable diseases among infants and children toward chronic diseases at older ages to be welcomed or resisted? To the extent it is to be resisted, what alternative responses to the demographic and epidemiological transitions might be considered? Such are the questions toward which this exploration into the distributional consequences of different approaches to the Third World’s changing health conditions is directed.
THE OVERALL SITUATION
Although most of this exploration focuses on the country level that is of particular interest, it can best begin with a brief look at the situation of the Third World as a whole and of the major regions within it. Such a look can provide both a sense of how central a place the demographic and epidemiological transitions deserve in designing health improvement strategies and an initial hint about the possible distributional consequences of these strategies. The inquiry is facilitated by the availability of recent figures produced at the World Bank on cause- and age-specific mortality in the Third World as a whole and in its major regions. A summary of these figures appears in Tables 1A and 1B (Bulatao and Stephens, 1989).
The most immediately obvious feature of these tables is the shift in overall age- and disease-specific mortality patterns that they were developed to document. This shift is very large. If mortality trends in the developing world proceed according to the Bulatao-Stephens projections for example, the percentage of total deaths occurring among children less than
TABLE 1A Deaths by Age (percent)
|
Year |
|||
Region and Age Group |
1970 |
1985 |
2000 |
2015 |
Less Developed Countries |
||||
0–14 |
50.0 |
42.9 |
26.7 |
18.5 |
15–64 |
27.8 |
28.9 |
33.2 |
34.3 |
65+ |
22.2 |
28.2 |
40.0 |
47.2 |
Total |
100.0 |
100.0 |
99.9 |
100.0 |
Life expectancy at birth (years) |
(57.5) |
(62.0) |
(66.0) |
(68.5) |
Latin America and the Caribbean |
||||
0–14 |
38.3 |
33.3 |
19.2 |
10.4 |
15–64 |
28.7 |
27.6 |
30.4 |
31.1 |
65+ |
33.1 |
39.0 |
50.3 |
58.6 |
Total |
100.1 |
99.9 |
99.9 |
100.1 |
Life expectancy at birth (years) |
(62.5) |
(66.5) |
(70.5) |
(72.5) |
Asia |
||||
0–14 |
48.5 |
37.1 |
18.5 |
10.7 |
15–64 |
27.9 |
30.6 |
34.2 |
34.2 |
65+ |
23.5 |
32.3 |
47.4 |
55.1 |
Total |
99.9 |
100.0 |
100.1 |
100.0 |
Life expectancy at birth (years) |
(59.0) |
(64.0) |
(68.0) |
(70.0) |
Middle East and North Africa |
||||
0–14 |
51.6 |
54.5 |
34.7 |
25.8 |
15–64 |
28.2 |
23.9 |
33.9 |
37.2 |
65+ |
20.3 |
21.6 |
31.4 |
36.9 |
Total |
100.1 |
100.0 |
100.0 |
99.9 |
Life expectancy at birth (years) |
(53.0) |
(60.0) |
(64.5) |
(66.5) |
Sub-Saharan Africa |
||||
0–14 |
60.3 |
58.8 |
50.2 |
41.9 |
15–64 |
26.5 |
27.0 |
31.4 |
34.6 |
65+ |
13.2 |
14.2 |
18.4 |
23.5 |
Total |
100.0 |
100.0 |
100.0 |
100.0 |
Life expectancy at birth (years) |
(45.0) |
(51.5) |
(57.0) |
(61.0) |
SOURCE: Bulatao and Stephens (1989:44, 61–62). |
TABLE 1B Death by Cause (percent)
|
Year |
|||
Region and Cause of Death |
1970 |
1985 |
2000 |
2015 |
Less Developed Countries |
||||
Infectious and parasitic diseases |
42.1 |
36.2 |
25.9 |
19.4 |
Neoplasms and circulatory disorders |
21.6 |
26.0 |
39.6 |
48.9 |
Other |
36.4 |
37.7 |
34.6 |
31.8 |
Total |
100.1 |
99.9 |
100.1 |
100.1 |
Life expectancy at birth (years) |
(57.5) |
(62.0) |
(66.0) |
(68.5 |
Latin America and the Caribbean |
||||
Infectious and parasitic diseases |
33.3 |
24.4 |
15.1 |
9.3 |
Neoplasms and circulatory disorders |
30.3 |
35.9 |
51.6 |
61.2 |
Other |
36.4 |
39.7 |
33.3 |
29.5 |
Total |
100.0 |
100.0 |
100.0 |
100.1 |
Life expectancy at birth (years) |
(62.5) |
(66.5) |
(70.5) |
(72.5 |
Asia |
||||
Infectious and parasitic diseases |
41.2 |
33.6 |
21.3 |
14.6 |
Neoplasms and circulatory disorders |
22.8 |
28.5 |
45.5 |
55.7 |
Other |
36.0 |
37.9 |
33.2 |
29.7 |
Total |
100.0 |
100.0 |
100.0 |
100.0 |
Life expectancy at birth (years) |
(59.0) |
(64.0) |
(68.0) |
(70.0 |
Middle East and North Africa |
||||
Infectious and parasitic diseases |
41.4 |
40.4 |
29.5 |
23.2 |
Neoplasms and circulatory disorders |
19.1 |
22.0 |
32.7 |
40.8 |
Other |
39.5 |
37.6 |
37.8 |
36.0 |
Total |
100.0 |
100.0 |
100.0 |
100.0 |
Life expectancy at birth (years) |
(53.0) |
(60.0) |
(64.5) |
(66.5) |
Sub-Saharan Africa |
||||
Infectious and parasitic diseases |
48.2 |
47.2 |
41.8 |
36.5 |
Neoplasms and circulatory disorders |
13.9 |
16.0 |
20.7 |
26.7 |
Other |
37.9 |
36.8 |
37.5 |
36.8 |
Total |
100.0 |
100.0 |
100.0 |
100.0 |
Life expectancy at birth (years) |
(45.0) |
(51.5) |
(57.0) |
(61.0) |
SOURCE: Bulatao and Stephens (1989:44, 15–16). |
15 years of age would decline from 50 to 18 between 1970 and 2015, while that attributable to people more than 65 years old would rise from 22 to 47. Similarly, the percentage of deaths caused by infectious and parasitic diseases would fall from 42 to 19, and the percentage resulting from neoplasms and circulatory disorders would climb from 22 to 49.
These overall shifts represent only part of the story, however. Of greater interest from a distributional perspective are the interregional differences in the numbers presented, especially differences between high- and low-mortality regions. When these differences are examined, a clear pattern emerges: The poorer a region’s health status, the greater is the importance of deaths at earlier relative to older ages, and from communicable relative to chronic diseases. This pattern is clearly visible in each of the time periods shown in Tables 1A and 1B. In general, when the life expectancy of a region is 50 years, about 60 percent of deaths occur among children under 15 years, about 20 percent at age 65 or above. When a region’s life expectancy is 70 years, the percentage of deaths occurring at less than 15 years is only 10–12, compared to almost 50 percent at age 65 or above. The interregional differences in the cause of death are equally sharp: when a region’s life expectancy is 50 years, approximately 50 percent of deaths are caused by infectious/parasitic diseases and 10–15 percent by neoplasms/circulatory disorders. When a region’s life expectancy is 70 years, the proportions are about 20 percent and 45–50 percent, respectively.2
This pattern suggests that a health strategy focusing on chronic diseases among adults and the elderly might be considerably more relevant for regions with low mortality levels than for high-mortality regions. For example, an emphasis on adult mortality would seem quite sensible for a region in which overall life expectancy is 70 years or so, in which almost 90 percent of deaths take place among people over 15. But its likely efficacy would appear questionable when life expectancy is only 50 and the distinct majority of deaths occurs among infants and children younger than 15.
This point appears to have been at least implicitly accepted within the international health profession in that calls for a review of health priorities seem to arise much more frequently with reference to the more advanced Latin American and Asian countries than with respect to high-mortality sub-Saharan African nations. What seems much less well recognized is the possibility that differences analogous to those just noted with respect to
regions might also exist within regions and within the individual countries of which regions are composed.
Should analogous intracountry differences exist, a health strategy oriented toward communicable diseases among infants and children would seem much more relevant for a society’s disadvantaged groups than a strategy focusing on chronic diseases of older ages. This raises the possibility that shifting from the former to the latter could benefit the better-off more than the poor, thereby increasing the degree of inequality between them.
How great is such a possibility? To find out, it is necessary to turn to the information available about individual countries and about differences in health conditions among population groups within them.
RANGE OF INTRACOUNTRY MORTALITY DIFFERENCES
The best way to begin the examination is with a look at intergroup differences in overall mortality levels—that is, at differences in mortality at all ages and from all causes together. These differences are of interest both in themselves and, as will be seen later, because of the use to which they can be put in assessing age- and cause-specific patterns. Also, the data that exist about differences in overall levels are considerably better than those available for differences in the distribution of deaths by age and cause; these data make it possible to begin by drawing upon observations taken directly from the countries concerned before proceeding to the necessarily more synthetic data featured in later stages of the discussion.
Data based on such direct observations are presented in the appendix to this paper. They represent 14 countries that are broadly representative of the widely diverse conditions found in the Third World, for which relatively recent information is available. For each of the countries there is information on one of four attributes commonly used in discussions of mortality differentials: income, place of residence, women’s educational status, and race. To facilitate comprehension, the figures presented in the appendix focus on one particularly important dimension of intergroup mortality differences: the range between a society’s healthiest and least healthy groups. To the extent feasible, the figures seek to compare the situation of the top 10–20 percent with that of the bottom 10–20 percent of the population in the country concerned.3
The many limitations of the data significantly restrict the uses to which
the figures may legitimately be put.4 The figures can, however, serve to substantiate the ubiquitous presence of significant intrasocietal differences in mortality and to give an initial sense of the general orders of magnitude typically involved. For example, the figures show that in all but one of the fourteen societies covered, infant mortality in the most disadvantaged group is more than twice as high as that of the most advantaged. In four of the fourteen, the difference is three times or more; in two, it is more than fourfold. Similarly, life expectancy of the most privileged groups is 25 percent or more than that of the most disadvantaged in all but three of the societies. In at least one case, the difference is greater 50 percent.5
Beyond this are suggestions with respect to differences in mortality status according to place of residence and education, the two attributes for which the most information is available:
-
The information with respect to place of residence gives rise to a “10–20” rule of thumb. That is, in most settings the life expectancy of the top 10–20 percent of the population as measured by place of residence appears to be somewhere on the order of 10–20 years higher than that of the lowest 10–20 percent of the population defined in a similar manner.6 This rule of thumb, which is broad enough to cover all of the six situations based
-
on place of residence in section B of the appendix, implies a two- to fourfold difference in infant mortality between the uppermost and lowest 10–20 percent of the population.
-
The children of the relatively few women with secondary or college education, which qualifies them for consideration as members of their countries’ social and economic elite, can expect to live 10–20 years longer than the children of women with no education, who belong to the core of the most disadvantaged groups in those societies. Here, too, the implied difference in infant mortality is roughly two- to fourfold. This variation of the 10–20 hypothesis suggested above would cover five of the six groups for which figures by education are presented in section C of the appendix.
FRAMEWORK
Placed in an appropriate framework, this knowledge about differences in overall conditions between a society’s healthiest and least healthy groups can play a central role in the examination of intergroup differences in age-and cause-specific patterns of mortality. The most appropriate framework, of course, is that of the developing country population within which the observed differences exist; however, fully satisfying studies can be undertaken only at a national (or subnational) level. However, an initial look at illustrative country situations based on composite data can be instructive both in introducing an approach of potential relevance for those national studies and in providing findings to guide policies in the interim prior to the studies’ completion.
For this purpose, two illustrative populations can serve as the basis for an examination of age- and cause-specific mortality patterns: one population with mortality and fertility levels characteristically found in societies of sub-Saharan Africa, the area of the Third World where overall mortality is highest; the other broadly representative of Latin America, the developing world’s lowest-mortality region. Within each, the age- and cause-specific mortality patterns of the healthiest and least healthy 10–20 percent of the population can be compared.
The overall mortality levels of the two societies can be established through reference to the World Bank’s figures cited in Tables 1A and 1B, which showed 1985–1990 life expectancy to be 51.5 years in sub-Saharan Africa, 66.5 years in Latin America and the Caribbean. An upward rounding to allow for increases that have presumably taken place since 1985–1990 leads to the selection of 55.0 years as the average life expectancy of the high-mortality society, 67.5 years as that of the low-mortality society.
Based on the 10–20 hypothesis established through the figures for developing countries presented in the appendix, the difference between the most-advantaged 10–20 percent and least-privileged 10–20 percent of the
population is set at 15.0 years7, assumed to be centered at the mean. These assumptions yield population groups with life expectancies of 47.5 and 62.5 years in the high-mortality society and life expectancies of 60.0 and 75.0 years in the low-mortality society.
The fertility levels are determined through reference to World Bank data. An examination of these data suggests that on average, a woman in an African population group with a life expectancy of 47.5 years will bear about 6.7 children. When life expectancy is 62.5 years, the average number of children will be around 5.1. In Latin America, the corresponding figures for children born are 4.9 (when life expectancy is 60.0 years) and 2.3 (when life expectancy is 75.0 years), respectively (Bulatao et al., 1989, 1990).
The same World Bank data are also used to establish the age distribution of each population group. This estimation is done with reference to the three or four national populations whose life expectancies are closest to that of the group concerned: two below, two above. Thus, for example, the age structure for the least healthy group in the high-mortality society is the unweighted average of those of the four African countries whose 1985–1990 life expectancies are closest to the 47.5 years of that group: Mali and Burkina Faso, whose life expectancies of 47.2 and 47.3 years, respectively, are just below 47.5 years; and Senegal and Mozambique, whose life expectancies of 47.6 and 48.1 years are just above. The population distributions of the other groups are determined in an analogous manner, by using data from Africa for the other group in the high-mortality society, and from Latin America and the Caribbean for the two groups in the low-mortality society.8
The resulting population distribution figures appear in Table 2, which also presents the mortality and fertility levels referred to earlier. The higher fertility and mortality of the least healthy groups combine to produce populations that are younger than those of the most healthy groups, particularly in the low-mortality society, with only around 25 percent of the healthiest population being under 15 years of age, compared with more than 40 percent of the least healthy population. At the upper end of the age spectrum, the situation is reversed: more than 9 percent of the healthiest population is
TABLE 2 Demographic Characteristics of Population Groups With Differing Life Expectancies and Fertility Levels
|
High-Mortality Country |
Low-Mortality Country |
||
Characteristic |
Least Healthy 10–20% of Population |
Healthiest 10–20% of Population |
Least Healthy 10–20% of Population |
Healthiest 10–20% of Population |
Life expectancy (years) |
47.5 |
62.5 |
60.0 |
75.0 |
Total fertility rate |
6.7 |
5.1 |
4.9 |
2.3 |
Distribution by age group |
||||
0–14 |
45.3 |
42.0 |
42.6 |
25.5 |
15–44 |
40.9 |
42.9 |
42.7 |
49.0 |
45–64 |
10.7 |
11.3 |
11.3 |
16.4 |
65+ |
3.1 |
3.8 |
3.4 |
9.2 |
Total |
100.0 |
100.0 |
100.0 |
100.1 |
65 years of age or older, compared with much less than 4 percent of the least healthy population.
AGE-SPECIFIC MORTALITY
Within societies such as those just described, how different are the age-and cause-specific patterns of mortality? What are the implications of these differences for the distributional consequences of alternative health improvement strategies?
These questions may be more easily addressed with respect to age, because age at death has been the subject of intense study by demographers over the past several decades. The results of this study cannot be considered fully definitive because it has not yet proceeded to the point at which it can provide reliable direct information about the Third World’s higher-mortality countries. There are, however, model life tables or standardized compilations of age-specific mortality data from developed and advanced developing countries that are generally considered trustworthy. These have long been routinely used by the United Nations, the World Bank, and national statistical offices to assist demographic analyses in a wide range of developing countries with insufficient data of their own. Although obviously less suitable than reliable direct observations, they have withstood the test of time well enough to justify at least a modest degree of confidence in their applicability for the development of broad illustrations such as those pictured here.
The figures in Table 3 show what one of the most frequently used sets
of model life tables has to say about age-specific mortality in populations with life expectancies corresponding to those of the four groups presently under consideration. Age-specific mortality displays the well-known “U-shaped” pattern featuring high death rates at the youngest and oldest ages, separated by a period of much lower mortality during the intervening years. In each case, death rates for the least healthy are higher at every age. Of particular interest in the present context is the magnitude of the difference between the two groups, which is not constant.
In the high-mortality society, for example, the death rate of the high-mortality group’s children aged 1–5 is about 3.6 times that of children in the low-mortality group. This difference declines sharply with age: it is only
TABLE 3 Age-Specific Mortality Rates (per 1,000 population) of Population Groups With Differing Life Expectancies
|
High-Mortality Country |
Low-Mortality Country |
||||
Age (1) |
Least Healthy 10–20% of Population (2) |
Healthiest 10–20% of Population (3) |
Ratio (2)/(3) (4) |
Least Healthy 10–20% of Population (5) |
Healthiest 10–20% of Population (6) |
Ratio (5)/(6) (7) |
0 |
148.80 |
62.49 |
2.38 |
74.45 |
13.31 |
5.59 |
1 |
20.15 |
5.58 |
3.61 |
7.37 |
0.44 |
16.79 |
5 |
4.50 |
1.61 |
2.80 |
2.00 |
0.23 |
8.54 |
10 |
3.37 |
1.24 |
2.72 |
1.53 |
0.20 |
7.58 |
15 |
4.73 |
1.96 |
2.42 |
2.35 |
0.39 |
6.04 |
20 |
6.39 |
2.71 |
2.36 |
3.24 |
0.55 |
5.95 |
25 |
7.13 |
3.00 |
2.38 |
3.60 |
0.58 |
6.18 |
30 |
8.15 |
3.44 |
2.37 |
4.12 |
0.70 |
5.92 |
35 |
9.43 |
4.19 |
2.25 |
4.96 |
0.97 |
5.13 |
40 |
11.16 |
5.42 |
2.06 |
6.27 |
1.55 |
4.04 |
45 |
13.42 |
7.43 |
1.81 |
8.34 |
2.77 |
3.01 |
50 |
17.97 |
10.75 |
1.67 |
11.87 |
4.70 |
2.53 |
55 |
24.00 |
15.76 |
1.52 |
17.05 |
8.10 |
2.11 |
60 |
35.09 |
24.06 |
1.46 |
25.80 |
13.42 |
1.92 |
65 |
50.31 |
37.03 |
1.36 |
39.16 |
23.19 |
1.69 |
70 |
75.49 |
58.64 |
1.29 |
61.35 |
40.08 |
1.53 |
75 |
115.04 |
93.60 |
1.23 |
97.06 |
69.09 |
1.40 |
80 |
175.81 |
146.80 |
1.20 |
151.54 |
112.60 |
1.35 |
85 |
270.64 |
231.02 |
1.17 |
237.54 |
183.27 |
1.30 |
90 |
413.43 |
360.62 |
1.15 |
369.37 |
295.78 |
1.25 |
95 |
626.93 |
557.93 |
1.12 |
569.41 |
472.27 |
1.21 |
100 |
949.35 |
860.99 |
1.10 |
875.70 |
751.40 |
1.17 |
SOURCE: Average figures for males and females from West model life tables in Coale and Demeny (1983:46, 50, 53). |
one-half as large among people 45 years of age in the two groups, and the mortality prospects of the few people in each group who attain 100 years are almost identical.
The differences in the low-mortality society follow the same general trend, reaching a maximum in the same 1- to 5-year age group and then falling. However, the magnitude of the differences is notably greater at each age. Between 1 and 5 years of age, for example, the death rate in the least healthy group is more than 16 times that of the most healthy group in the low-mortality society, compared with a difference of three-plus times in the high-mortality society.
The implications of this pattern can be seen by weighting the age-specific mortality rates that lie behind these curves by the proportion of the population in each age group shown in Table 2, to provide an estimate of the percentage of total deaths occurring at each age. These results appear in Table 4.
The results show that infant and child deaths occupy a far more prominent place in society’s least healthy groups than in its most healthy ones. In the least healthy group of the high-mortality society, for instance, there are nearly four times as many deaths in the age group 0–14 as in the age group 65+; in the healthiest group of that society, the number of deaths in each age category is approximately equal. The situation in the low-mortality society is similar: there are 50 percent more deaths at ages 0–14 than at ages 65 and over, less than one-tenth as many in the healthiest group.
The potential consequences of these differences from a distributional perspective may be illustrated through a simple example. Suppose it were possible to reduce mortality equally in all of a society’s population groups by a given amount—one-third, for example—in any specified age category
TABLE 4 Total Deaths by Age (percent)
|
High-Mortality Country |
Low-Mortality Country |
||
Age Group |
Least Healthy 10–20% of Population |
Healthiest 10–20% of Population |
Least Healthy 10–20% of Population |
Healthiest 10–20% of Population |
0–14 |
56.5 |
35.0 |
40.4 |
4.2 |
15–44 |
16.5 |
15.5 |
16.5 |
6.2 |
45–64 |
12.2 |
17.0 |
16.7 |
15.2 |
65+ |
14.8 |
32.5 |
26.4 |
74.4 |
Total |
100.0 |
100.0 |
100.0 |
100.0 |
Ratio of deaths at ages 0–14 to deaths at age 65+ |
3.8 |
1.1 |
1.5 |
0.1 |
through the introduction of additional resources.9 What would be the distributional consequences of such a mortality reduction in any one of the four broad age categories presented in Table 4, relative to those of reducing it in another age category?
The answer to this question appears in Table 5. If mortality among infants and children under 15 years were to decline by one-third in all groups, the result would be distinctly progressive in that the least healthy would benefit more than the healthiest, and the gap between the two groups would narrow. The number of deaths in the least healthy group of the high-mortality society would fall by 18.8 percent, a reduction 1.6 times as large as the 11.7 percent decline of the healthiest group. In the low-mortality society, the differences would be greater: deaths would fall by 13.4 percent in the least healthy group, nearly 10 times as much as the 1.4 percent reduction in the healthiest one.
The result would become steadily less progressive and more regressive as the age of those affected by the mortality decline increases. An emphasis on ages 65+ would be clearly regressive in each society, benefiting the healthiest twice or more as much as the least healthy.10
9 |
This formulation produces what might be called an “equal-output” (or, more precisely, an “equiproportionate-output”) scenario, where output is defined in terms of mortality reduction. Such a scenario needs to be distinguished carefully from an “equal-input” scenario, in which the per capita allocation of resources to each health status or age group is equal. The equal-output and equal-input scenarios will produce the same results only if the output-input ratios (i.e., the amount of mortality reduction produced per unit of resource invested) are the same for all groups. The equal-output scenario has been selected for presentation here partly to permit ready comprehension, and partly out of concern for the reliability of the available information about input-output relationships that an input-based scenario requires. A next step in the line of inquiry initiated by this paper is to be an input-based approach, allowing variation in both the level and the effectiveness of inputs by group. In the meantime, brief references to the results that can be expected from use of this approach appear in footnotes 10 and 11. |
10 |
The distinctions would be even sharper if the resources needed to reduce mortality in any particular age category were to be drawn from activities benefiting people in some other age category, instead of being introduced exogenously as assumed in the calculations underlying the figures in Table 5. (For instance, diverting resources from children’s programs to serve people in older ages under the conditions specified in the example would reduce the progressive influence of those programs in addition to introducing the more regressive effect of the new ones.) The distinctions would also become sharper under an input-based approach such as that alluded to in footnote 9 incorporating the view, held by many, that infant and child deaths can be prevented less expensively than deaths among adults and the elderly under present technology. Use of an input-based approach incorporating the assumption that deaths at all ages are less expensive to prevent among the least healthy than among the healthiest would increase the progressive effect of funds allocated to the least healthy at each age. This would raise the age above which there would be a regressive effect in absolute terms. In the absence of variations in the age-specific cost-effectiveness differential between groups, however, there would be no change in relative terms because the progressive effect of dealing with unhealthy infants and children would be increased by an amount comparable to that of the increase at older ages. |
TABLE 5 Change in Number of Deaths Attributable to One-Third Reduction of Mortality in a Particular Age Group (percent)
|
High-Mortality Country |
Low-Mortality Country |
||||
Age Group |
Least Healthy 10–20% of Population (1) |
Healthiest 10–20% of Population (2) |
Ratio (1)/(2) (3) |
Least Healthy 10–20% of Population (4) |
Healthiest 10–20% of Population (5) |
Ratio (4)/(5) (6) |
One-third reduction in deaths from all causes |
|
|||||
0–14 |
−18.8 |
−11.7 |
1.6 |
−13.4 |
−1.4 |
9.6 |
15–44 |
−5.5 |
−5.2 |
1.1 |
−5.5 |
−2.1 |
2.6 |
45–64 |
−4.1 |
−5.7 |
0.7 |
−5.6 |
−5.1 |
1.1 |
65+ |
−4.9 |
−10.8 |
0.5 |
−8.8 |
−24.8 |
0.4 |
This result holds even though, as noted earlier, mortality rates among the least healthy are higher than among the healthiest at every age level—including age 65 and older—because of a situation somewhat analogous to that of the famous international trade example that David Ricardo presented nearly 200 years ago. That is, what matters is not so much the absolute difference in mortality levels between two groups at a given age level, but rather the comparative difference (i.e., the magnitude of the difference at that age level relative to the magnitude of the difference at another age level) considered in conjunction with differences in the age structure of the groups concerned resulting (primarily) from differing fertility levels.
The example also shows that the regressive effects of shifting attention toward older ages would by no means be limited to high-mortality societies. The low-mortality societies in which the case for a shift in emphasis seems particularly strong would also be affected, and under some circumstances and by some measures, the effect would be considerably more significant than in high-mortality populations. As indicated above, for example, a focus on infants and children in the illustrative low-mortality society would benefit the least healthy groups almost 10 times as much as the healthiest group, compared with less than twice as much in the high-mortality one. An upward movement in the age-specific focus would bring a far greater fall in this ratio in the high- than in the low-mortality society because of the age structure effects of the larger fertility differences between groups in the low-mortality society and from the much greater intergroup differences in child mortality.
As such an outcome indicates, what matters in determining the distributional impact of a policy change is not a society’s overall health status, but rather the pattern of intergroup differences in demographic and health conditions within the society. The distributional consequences of changing priorities may be considerably more severe in a low-mortality society marked by large intergroup differences than in a high-mortality society where intergroup differences are relatively small.
None of the quantitative findings reported here can be considered definitive, of course. The example from which they emerged has been consciously oversimplified to facilitate communication about a general point, and as a result, the numbers presented cannot be legitimately applied to any particular situation without far more careful analysis than is possible here. This being said, however, the logic that underlies the figures suggests that the concern raised in the earlier investigation of regional data is well founded. That is, the health problems of infants and children appear likely to loom much larger among the less healthy than among the healthier groups of a national population, suggesting that shifting priorities away from them toward the health problems of older people would divert attention toward issues of primary relevance for the better-off and away from those of greatest concern to the least healthy.
CAUSE-SPECIFIC MORTALITY
The approach just used to look at age-specific mortality can also be employed to examine the distributional implications of alternative disease-oriented program foci, although the results that it produces must be viewed with considerably greater caution for several reasons. One is the inherent complexity of disease patterns, which defy easy categorization. A second is the smaller amount and less certain reliability of the information about cause of death. A third is the more limited experience in inferring cause-specific mortality patterns in the Third World from developed country data.
Even so, reliance on developed country data when assessing the Third World’s evolution appears to be becoming an increasingly accepted practice in leading professional institutions, as demonstrated by the central role it has played in such major works as the two World Bank studies that represent the most careful extant examination of developing country trends (Feachem et al., 1992; Jamison and Mosley, 1993). This reliance and the importance of the topic suffice to justify at least a passing look at what the figures in current use have to say about the distributional consequences of alternate disease-specific foci.
Table 6 presents estimates of cause-specific mortality rates for our four populations that are based on estimates from Bulatao and Stephens (1992). As can be seen from columns 4 and 7 of the table, mortality from most causes is higher in the least healthy groups than in the healthiest ones. However, there is one exception: neoplasm, a noncommunicable disease particularly important at older ages. For the single most important set of noncommunicable diseases in adults and the elderly—circulatory and certain degenerative diseases—death rates are only modestly (15–20 percent) higher among the least healthy than among the healthiest. In contrast, people in the least healthy groups are 2.2 to 3.6 times as likely to die as are those in the healthiest from infections and parasitic diseases, the largest category of communicable ailments.
The implications of these patterns, when viewed in conjunction with the inter-group differences in age structure, can be seen in from Tables 7 and 8. These two tables, developed in a manner analogous to that used to produce Tables 5 and 6, show the cause-specific distribution of deaths and the effect on each group of a society-wide one-third reduction of mortality in a particular disease category.
As can be seen from Table 7, the principal communicable diseases are of far greater importance than the principal noncommunicable diseases for the least healthy than for the healthiest segments of society. In the high-mortality society, communicable diseases kill three times as many of the least healthy as do noncommunicable diseases; among the healthiest, about the same number of people die from each type of disease. When overall mortality is low, the least healthy die around 40 percent more often from
TABLE 6 Cause-Specific Mortality Rates (per 1,000 population) of Groups With Differing Life Expectancies
|
High-Mortality Country |
Low-Mortality Country |
||||
Cause of Death (1) |
Least Healthy 10–20% of Population (2) |
Healthiest 10–20% of Population (3) |
Ratio (2)/(3) (4) |
Least Healthy 10–20% of Population (5) |
Healthiest 10–20% of Population (6) |
Ratio (5)/(6) (7) |
Circulatory system and certain degenerative diseases |
5.70 |
4.91 |
1.16 |
6.13 |
5.25 |
1.17 |
Infectious and parasitic diseases |
7.99 |
3.65 |
2.19 |
4.64 |
1.30 |
3.57 |
Neoplasms |
1.33 |
1.44 |
0.92 |
1.68 |
1.83 |
0.92 |
Injury and/or poisoning |
0.84 |
0.69 |
1.22 |
0.76 |
0.64 |
1.19 |
Perinatal conditions |
0.70 |
0.37 |
1.89 |
0.35 |
0.16 |
2.19 |
Complications of pregnancy |
0.14 |
0.06 |
2.33 |
0.07 |
0.01 |
7.00 |
Other causes |
5.51 |
3.50 |
1.57 |
4.47 |
2.43 |
1.84 |
SOURCE: Calculated by using functional relationships between overall life expectancy and age-cause-specific mortality rates prepared by Bulatao and Stephens (1989: Supplementary Tables). (Figures for both groups in high-mortality countries are based on stationary age structures at country mean life expectancy (55.0 years) in order to neutralize intergroup fertility differences. For the same reason, figures for both groups in low-mortality countries are calculated for stationary population age structures at country mean life expectancy of 62.5 years.) |
TABLE 7 Total Deaths by Cause (percent)
|
High-Mortality Country |
Low-Mortality Country |
||
Cause of Death |
Least Healthy 10–20% of Population |
Healthiest 10–20% of Population |
Least Healthy 10–20% of Population |
Healthiest 10–20% of Population |
Infectious and parasitic diseases |
47.2 |
32.8 |
36.6 |
10.7 |
Circulatory system diseases and neoplasms |
15.5 |
30.0 |
26.0 |
59.4 |
Other causes |
37.3 |
37.2 |
37.4 |
30.0 |
Deaths from infectious and parasitic diseases relative to deaths from circulatory system diseases and neoplasms |
3.0 |
1.1 |
1.4 |
0.2 |
communicable than from noncommunicable diseases; the healthiest die only about 20 percent as frequently.
Accordingly, it should come as no surprise to find that the least healthy would be the principal beneficiaries of a reduction of mortality from communicable diseases produced in a manner comparable to that described earlier in the section on age-specific mortality. As indicated by the figures in columns 3 and 6 of Table 8, a society-wide decline in mortality from communicable diseases would benefit the least healthy around 1.4 times as much as the healthiest in the high-mortality society, and more than three time as much in the low-mortality society. A decline in mortality from noncommunicable diseases would produce the opposite effect: the least healthy would gain only about one-half as much as the healthiest in each society, and the gap between the two groups would widen. According to the figures in Table 8, in other words, an emphasis on communicable diseases is progressive; a focus on the noncommunicable diseases is regressive.11
Although figures like these can be considered no more than suggestive,
TABLE 8 Change in Number of Deaths Attributable to One-Third Reduction of Mortality from Particular Causes (percent)
|
High-Mortality Country |
Low-Mortality Country |
||||
|
Least Healthy 10–20% of Population (1) |
Healthiest 10–20% of Population (2) |
Ratio (1)/(2) (3) |
Least Healthy 10–20% of Population (4) |
Healthiest 10–20% of Population (5) |
Ratio (4)/(5) (6) |
One-third reduction in deaths from infectious and parisitic diseases |
−15.7 |
−10.9 |
1.4 |
−12.2 |
−3.6 |
3.4 |
One-third reduction in deaths from circulatory system diseases and neoplasms |
−5.2 |
−10.0 |
0.5 |
−8.6 |
−19.8 |
0.4 |
they are nonetheless of interest both in themselves and for the reinforcement they give to and derive from the age-specific mortality figures presented earlier. The fact that an emphasis on reducing mortality from leading communicable diseases produces the same distributional effect as a focus on reducing deaths at early ages is what one would expect in light of what is known about the generally direct relationships that exist between the two, as is the fact that emphasizing noncommunicable diseases has the same distributional consequences as stressing mortality reduction among adults and the elderly.
OTHER CONSIDERATIONS
In moving from the suggestions emerging from the simple arithmetic exercises presented here toward the definitive conclusions needed for effective policy formulation, many other considerations will have to be taken into account. An adequate treatment of any of them would require far more space than presently available and will thus have to await another time. It is, however, both possible and valuable to note some of the most important.
Among the most significant considerations is one that has already been mentioned: the wide variations among countries that could limit the relevance of generalizations such as those developed here in any particular setting. However, the generalizations’ illustrative value should not be denigrated. The demographic and epidemiological parameters developed for the high- and low-mortality societies presented in the preceding sections are far from hypothetical. Rather, they are based on information about sub-Saharan Africa and Latin America drawn from standard sources and used routinely for a wide range of analytical purposes. The results presented can thus be reasonably considered at least broadly representative of each of those regions as a whole. There are, however, wide differences within each region; and many important developing countries lie in neither one. The strengths of the effects of a shift in age- and disease-specific focus in particular settings could thus well vary from those presented above in a manner that can be determined only through the fuller country-level investigations that this initial exploration hopes to inspire.
A second consideration is the limited nature of the indicator of health status used in the illustrations (see Preston in this volume for a fuller discussion). The number of deaths suffered or prevented in a group is obviously important, but it is by no means a complete representation of the group’s health status or of changes in it. Compared to a more appealing indicator such as the number of quality-adjusted years lost or added, a focus on the number of deaths prevented is deficient in two respects. First, it ignores differences in the number of years of life added per death prevented at different ages or from different causes. A death prevented in childhood
obviously adds many more years of life than does a death averted at an advanced age. To overlook this difference is to bias the arguments advanced against infant and child health programs and against many communicable disease initiatives. Second, it overlooks morbidity. If the morbidity/mortality ratio differs across age and/or disease groups, and if people in the groups concerned place a high value on morbidity relative to mortality reduction, a failure to take morbidity into account could lead to distorted results. Whether any such distortion actually exists is difficult to determine. Little is known about the morbidity/mortality ratios of diseases suffered at different ages and caused by different agents.12 Even less knowledge exists about the relative importance that different population groups in developing countries attach to morbidity relative to mortality.
The third consideration is the question of the interrelationships among deaths (and illnesses) at different ages and from different causes. The examples presented here assume that no such relationships exist. But in fact, they clearly do. The two-way relationship between death and poor health at different ages has been a topic of particular attention. Deaths among adults are thought to produce a significant decrease in the survival probabilities of their children. Conversely, deaths among infants and children have well-known effects on fertility and thus on maternal health; more generally, unhealthy children are believed to become less healthy adults who are more prone to early death (see Mosley and Gray in this volume). The effects work in both directions, and it is possible that they are of adequately similar magnitudes to cancel each other. Should this be the case, the failure to account for the effects in the illustrations would not bias the illustrations’ results, but one cannot be certain without a careful investigation.
The fourth consideration is the previously noted possibility of differences in the cost-effectiveness of interventions available to prevent deaths at different ages and from different causes. The conclusions reached in the illustrations are valid only if the cost of preventing a death in the most
prevalent category is equal to or less than the cost of preventing a death in the less prevalent categories.13 In view of what is currently known about the cost-effectiveness of dealing with different age- and disease-specific groups (Jamison and Mosley, 1993), the inclusion of cost-effectiveness considerations would seem unlikely to upset many of the conclusions reached in situations where communicable diseases and deaths at an early age appear to deserve highest priority. However, the costs of the traditional approaches to averting deaths among older people from chronic diseases are notoriously high, which raises the possibility that the illustrations in which such diseases emerge dominant overstate the case for according priority to them.14
A fifth consideration concerns the existence of wide differences in the distributional implications of dealing with specific problems within each of the broad age and disease categories covered. Even if a general focus on adults and the elderly or on noncommunicable diseases is less progressive or more regressive than an overall emphasis on communicable diseases or on infants and children, such will not necessarily be the case with respect to each and every noncommunicable condition that occurs primarily at older ages. The figures in Table 6 on deaths resulting from complications of pregnancy illustrate the point. Deaths from this cause, a condition of young adults that is not communicable in the standard sense of the term, represent too small a proportion of total deaths at older ages to justify their use as the basis of any generalization (Bulatao and Stephens, 1989). Unlike other, more frequent causes of death from noncommunicable diseases, however, they clearly affect the least healthy groups far more severely than the healthiest groups: more than twice as severely in the high-mortality population and seven times as severely in the low-mortality one. If comparable figures available were for malnutrition, the results would probably be similar. Such examples serve as important reminders that, although a primary overall focus on communicable diseases and an overall emphasis on infants and children might well be justified, the careful examinations necessary for the formulation of intelligent policies are likely to find at least some activities
outside those rubrics that deserve inclusion in a program oriented toward disparity reduction.
The sixth consideration is the possibility of modifying the distributional effects of efforts to reduce mortality by targeting specific population groups to benefit from them. By careful targeting, a program with any age- or disease-specific focus can be shaped to benefit the least healthy more than the healthiest. For example, even a tertiary-care facility specializing in rare cardiovascular conditions will benefit the least healthy more than the healthiest and thereby reduce inequalities if it is located in a poor neighborhood and its services are made available only to the poorest residents of that neighborhood. However, the magnitude of the distributional improvement thereby achieved would obviously be far smaller than that which brought about by applying the same resources to an approach more closely aligned with the epidemiological conditions prevailing in the poor community being served. As the example demonstrates, targeting has an extremely important role to play in the development of programs to help the disadvantaged, but it cannot by itself lead to the efficient reduction of disparities that represents the logical objective of such efforts. To reduce disparities, targeting must be used in conjunction with a clear appreciation of the target population’s demographic-epidemiological situation and with cost-effectiveness considerations.
CONCLUSION
The importance of the considerations presented in the preceding section point to a clear need for further research in many areas to assist in the development of effective approaches to the reduction of differences in health status. One can only urge and hope that such research will proceed as rapidly as possible.
In the meantime, thousands of health policymakers will be making hundreds of thousands of decisions concerning health policies that will benefit different groups in different manners. Can guidance, however preliminary and provisional, be drawn from what is now known and presented here in order to help them decide how to respond to the demographic and epidemiological transitions?
At the heart of any such guidance would have to be a warning against responding to overall trends such as those portrayed by the data presented in tables 1A and 1B by a general shift in health priorities toward a greater emphasis on problems caused by noncommunicable diseases among adults and the elderly. These data are societal averages. To rely on them is to overlook the differences that exist among groups within society and, in the process, to give as much weight to problems concentrated in society’s most privileged groups as to those of greatest relevance for the least healthy. The
information presented here suggests that the least healthy can be much better served by a strategy based on a careful study of their particular needs and that such a strategy is likely to give highest priority to communicable diseases among the young.
REFERENCES
Bobadilla, J.L., J.Frenk, T.Frejka, R.Lozano, and C.Stern 1993 The epidemiological transition and health priorities. In D.T.Jamison and W.H. Mosley, eds., Disease Control Priorities in Developing Countries. New York: Oxford University Press for the World Bank.
Bulatao, R.A., and P.W.Stephens 1989 Estimates and projections of mortality by cause: A global overview, 1990–2015. Unpublished manuscript, World Bank, Washington, D.C.
1992 Estimates and projections of mortality by cause: A global overview, 1970–2015. Policy Research Working Papers. Washington, D.C.: World Bank.
Bulatao, R.A., E.Bos, P.W.Stephens, and M.T.Vu 1989 Africa regional population projections (1989–1990 edition). Policy Planning and Research Working Paper. No. WPS 330 (November). Population and Human Resources Department of the World Bank, Washington, D.C.
Bulatao, R.A., E.Bos, P.W.Stephens, and M.T.Vu 1990 Latin America and the Caribbean region population projections, 1989–1990 edition. Policy Planning and Research Working Paper. No. WPS 329 (November). Population and Human Resources Department of the World Bank, Washington, D.C.
Coale, A.J., and P.Demeny, with B.Vaughan 1983 Regional Model Life Tables and Stable Populations, 2nd ed. New York: Academic Press.
Feachem, R., T.Kjellstrom, C.J.L.Murray, M.Over, and M.A.Phillips, eds. 1992 The Health of Adults in the Developing World. New York: Oxford University Press for the World Bank.
Jamison, D.T., and W.H.Mosley. 1993 Disease control priorities in developing countries. In D.T.Jamison and W.H. Mosley, eds., Disease Control Priorities in Developing Countries. New York: Oxford University Press for the World Bank.
APPENDIX
Intergroup Differences in Mortality Within Developing Countries
A. Populations differentiated by income status
Brazil: |
The highest mortality group for 1970 is based on households with monthly incomes below Cr$100, representing 21.0 percent of the total population. Infant mortality rate (IMR) was 127.4 per 1,000 live births and life expectancy at birth (e0) was 48.9 years. The lowest mortality group is based on households with monthly incomes above Cr$1,000, representing 22.0 percent of the total population. IMR was 60.8 per 1,000 live births and e0 was 62.2 years. IMR relative difference: 2.10 times; IMR absolute difference: 66.6 deaths per 1,000 live births. e0 relative difference: 1.27 times; e0 absolute difference: 13.3 years. |
|
Charles H.Wood and José Alberto Magno de Carvalho, The Demography of Inequality in Brazil, (Cambridge, New York, New Rochelle, Melbourne, Sydney: Cambridge University Press, 1988), p. 190. (Infant mortality derived through application of model life tables to life expectancy figures provided in text.) |
B. Populations differentiated by place or residence
Kenya: |
The highest mortality group for 1974 is based on residents of Coast and Nyanza Provinces, representing 26.1 percent of the total population. IMR was 140.3 per 1,000 live births and e0 was 46.7 years. The lowest mortality group is based on residents of the Central Province, representing 15.3 percent of the total population. IMR was 58.0 per 1,000 live births and e0 was 62.9 years. IMR relative difference: 2.42 times; IMR absolute difference: 82.3 deaths per 1,000 live births. e0 relative difference: 1.36 times; e0 absolute difference: 16.2 years. |
|
W.Henry Mosley, “Will Primary Care Reduce Infant and Child Mortality? A Critique of Some Current Strategies, with Special Reference to Africa and Asia,” in Jacques Vallin and Alan D.Lopez, eds. Health Policy, Social Policy, and Mortality Prospects: Proceedings of a Seminar at Paris, France, February 28-March 4, 1983, (Place of publication not indicated: Ordina Publications for the Institut National d’Etudes Démographiques and the International Union for the Scientific Study of Population, 1985), p. 109. (Infant mortality and life expectancy derived through application of model life tables to figures for child deaths by age 2 per 1000 live births presented in text.) |
Sudan: |
The highest mortality group for 1973 is based on residents of Bahr El Ghazal Province, representing 9.4 percent of the total population. IMR was 227.5 per 1,000 live births and e0 was 34.2 years. The lowest mortality group is based on residents of the Khartoum Province, representing 7.8 percent of the total population. IMR was 107.6 per 1,000 live births and e0 was 52.5 years. IMR relative difference: 2.11 times; IMR absolute difference: 119.9 deaths per 1,000 live births. e0 relative difference: 1.54 times; e0 absolute difference: 18.3 years. |
|
Abdul-Aziz Farah and Samuel H.Preston, “Child Mortality Differentials in Sudan,” Population and Development Review, vol. 8, no. 2 (June 1982). (Infant mortality derived through application of model life tables to life expectancy figures presented in text. Figures for percentage of total population calculated from 1973 census.) |
India: |
The highest mortality group for 1986 is based on residents of Uttar Pradesh State, representing 16.2 percent of the total population. IMR was 132.0 per 1,000 live births and e0 was 48.0 years. The lowest mortality group is based on residents of Kerala, Maharashtra, and Punjab States, representing 15.3 percent of the total population. IMR was 54.9 per 1,000 live births and e0 was 63.5 years. IMR relative difference: 2.40 times; IMR absolute difference: 77.1 deaths per 1,000 live births. e0 relative difference: 1.32 times; e0 absolute difference: 15.6 years. |
|
Office of the Registrar General, Ministry of Home Affairs, Government of India, Registrar General’s Newsletter, vol. XIX, no. 1 (January 1988), p. 16. (Life expectancy derived through application of model life tables to infant mortality figures provided in text. Figures for percentage of total population calculated from 1981 census.) |
Philippines: |
The highest mortality group for 1986 is based on residents of Western and Central Mindanao Regions, representing 10.0 percent of the total population. IMR was 101.4 per 1,000 live births and e0 was 53.2 years. The lowest mortality group is based on residents of the National Capital Region, representing 12.9 percent of the total population. IMR was 36.3 per 1,000 live births and e0 as 67.5 years. IMR relative difference: 2.79 times; IMR absolute difference: 65.1 deaths per 1,000 live births. e0 relative difference: 1.27 times; e0 absolute difference: 14.3 years. |
|
Panfila Ching, “Factors Affecting the Demand for Health Services in the Philippines,” (unpublished manuscript, 1989), pp. 10–21, 29. |
Mexico: |
The highest mortality group for the period 1982–1988 is based on residents of eight southern states, representing 16.6 percent of the total population. IMR was 92.0 per 1,000 live births and e0 was 55.5 years. The lowest mortality group is based on residents of the seven northern states, representing 14.8 percent of the total population. IMR was 28.0 per 1,000 live births and e0 was 70.2 years. IMR relative difference: 3.29 times; IMR absolute difference: 64.0 deaths per 1,000 live births. e0 relative difference: 1.26 times; e0 absolute difference: 14.7 years. |
|
José-Luis Bobadilla et al., “The Epidemiological Transition and Health Priorities,” draft manuscript prepared for the World Bank Disease Control Priorities Review, December 1989, p. 14a. (Life expectancy derived through application of model life tables to infant mortality figures presented in text. Figures for percentage of total population calculated from 1980 census.) |
Peru: |
The highest mortality group for 1967–1968 is based on residents of the Southern Region, representing 18.6 percent of the total population. IMR was 173.0 per 1,000 live births and e0 was 42.0 years. The lowest mortality group is based on residents of the Metropolitan Region, representing 24.4 percent of the total population. IMR was 78.0 per 1,000 live births and e0 was 58.2 years. IMR relative difference: 2.22 times; IMR absolute difference: 96.0 deaths per 1,000 live births. e0 relative difference: 1.39 times; e0 absolute difference: 16.2 years. |
|
Hugo Behm and Alfredo Ledesma, La Mortalidad en Los Primeros Años de Vida en Paises de La America Latina: Peru, 1967–1968, (Serie A. No. 1029) (San José: Centro Latinoamericano de Demografia, Mayo de 1977), pp. 11, 38. |
C. Populations differentiated by women’s educational status
Burundi: |
The highest mortality group for the period 1981–1987 is based on children of women with no education, representing 80.2 percent of the total population. IMR was 90.0 per 1,000 live births and e0 was 55.8 years. The lowest mortality group is based on children of women with secondary or higher education, representing 2.2 percent of the total population. IMR was 32.0 per 1,000 live births and e0 was 69.2 years. IMR relative difference: 2.81 times; IMR absolute difference: 58.0 deaths per 1,000 live births. e0 relative difference: 1.24 times; e0 absolute difference: 13.4 years. |
|
“Burundi 1987: Results from the Demographic and Health Survey,” Studies in Family Planning, Vol. 20, No. 3 (May/ June 1989). (Life expectancy derived through application of model life tables to infant mortality figures provided in text.) |
Dominican Republic |
The highest mortality group for the period 1982–1985 is based on children of women with no education, representing 4.8 percent of the total population. IMR was |
|
102.0 per 1,000 live births and e0 was 53.5 years. The lowest mortality group is based on children of women with higher education, representing 8.4 percent of the total population. IMR was 34.0 per 1,000 live births and e0 was 68.5 years. IMR relative difference: 3.00 times; IMR absolute difference: 68.0 deaths per 1,000 live births. e0 relative difference: 1.26 times; e0 absolute difference: 15.0 years. |
|
“Dominican Republic 1986: Results from the Demographic and Health Survey,” Studies in Family Planning, Vol. 19, No. 2 (March/April 1988). (Life expectancy derived through application of model life tables to infant mortality figures provided in text.) |
Ecuador: |
The highest mortality group for the period 1982–1987 is based on children of women with no education, representing 7.8 percent of the total population. IMR was 106.0 per 1,000 live births and e0 was 52.8 years. The lowest mortality group is based on children of women with higher education, representing 9.2 percent of the total population. IMR was 22.0 per 1,000 live births and e0 was 71.9 years. IMR relative difference: 4.82 times; IMR absolute difference: 84.0 deaths per 1,000 live births. e0 relative difference: 1.36 times; e0 absolute difference: 19.1 years. |
|
“Ecuador 1987: Results from the Demographic and Health Survey,” Studies in Family Planning, Vol. 20, No. 2 (March/April 1989). (Life expectancy derived through application of model life tables to infant mortality figures provided in text.) |
Indonesia: |
The highest mortality group for the period 1984–1987 is based on children of women with no education, representing 23.2 percent of the total population. IMR was 99.0 per 1,000 live births and e0 was 54.1 years. The lowest mortality group is based on children of women with secondary or higher education, representing 13.1 percent of the total population. IMR was 34.0 per 1,000 live births and e0 was 65.6 years. |
|
IMR relative difference: 2.91 times; IMR absolute difference: 65.0 deaths per 1,000 live births. e0 relative difference: 1.27 times; e0 absolute difference: 14.4 years. |
|
“Indonesia 1987: Results from the Demographic and Health Survey,” Studies in Family Planning, Vol. 20, No. 5 (September/October 1989). (Life expectancy derived through application of model life tables to infant mortality figures provided in text.) |
Senegal: |
The highest mortality group for the period 1981–1986 is based on children of women with no education, representing 77.2 percent of the total population. IMR was 96.0 per 1,000 live births and e0 was 54.7 years. The lowest mortality group is based on children of women with higher education, representing 9.3 percent of the total population. IMR was 50.0 per 1,000 live births and e0 was 64.7 years. IMR relative difference: 1.92 times; IMR absolute difference: 46.0 deaths per 1,000 live births. e0 relative difference: 1.16 times; e0 absolute difference: 10.0 years. |
|
“Senegal 1986: Results from the Demographic and Health Survey,” Studies in Family Planning, Vol. 19, No. 6 (November/December 1988). (Life expectancy derived through application of model life tables to infant mortality figures provided in text.) |
Thailand: |
The highest mortality group for the period 1982–1987 is based on children of women with no education, representing 9.7 percent of the total population. IMR was 54.03 per 1,000 live births and e0 was 63.7 years. The lowest mortality group is based on children of women with secondary or higher education, representing 7.7 percent of the total population. IMR was 19.0 per 1,000 live births and e0 was 72.9 years. IMR relative difference: 2.84 times; IMR absolute difference: 35.0 deaths per 1,000 live births. e0 relative difference: 1.14 times; e0 absolute difference: 9.2 years. |
|
“Thailand 1987: Results from the Demographic and Health Survey,” Studies in Family Planning, Vol. 20, No. 1 (January/February 1989). (Life expectancy derived through application of model life tables to infant mortality figures provided in text.) |
D. Populations differentiated by race
South Africa: |
The highest mortality group for the period 1981–1985 is based on the black population, representing 68.0 percent of the total population. IMR was ranged from 94.0 to 124.0 per 1,000 live births and e0 ranged from was 49.4 to 55.1 years. The lowest mortality group is based on the white popuation, representing 18.2 percent of the total population. IMR was 12.3 per 1,000 live births and e0 was 75.3 years. IMR relative difference: 7.64–10.08 times; IMR absolute difference: 81.7–111.7 deaths per 1,000 live births. e0 relative difference: 1.37–1.52 times; e0 absolute difference: 20.2–25.9 years. |
|
D. Yach, “Infant Mortality Rates in Urban Areas of South Africa, 1981–1985,” South African Medical Journal, vol. 73 (1988), p. 234. (Life expectancy derived through application of model life tables to infant mortality figures presented in text. Figures for percentage of total population calculated from 1980 census.) |
NOTES
The data here are drawn from the 14 studies referred to in the text. These studies employed a wide range of approaches that required conversion into a standardized format. Three aspects of this conversion process are worthy of note: