Howo Well do Health Systems Perform? MEASURING GOAL ACHIEVEMENT To assess a health system, one must measure five things: the overall level of health; the distribution of health in the population; the overall level of responsiveness; the distribution of responsiveness; and the distribution of financial contribution For each one, WHO has used existing sources or newly generated data to calculate measures of attainment for the countries where information could be obtained. These data were also used to estimate values when particular numbers were judged unreliable, and to estimate attainment and performance for all other Member States. Several of these measures are novel and are explained in detail in the Statistical Annex, where all the estimates are given, along with ntervals expressing the uncertainty or degree of confidence in the point estimate. The cor- rect value for any indicator is estimated to have an 80% probability of falling within the uncertainty interval, with chances of 10% each of falling below the low value or above the high one. This recognition of inexactness underscores the importance of getting more and better data on all the basic indicators of population health, responsiveness and faimess in financial contribution, a task which forms part of WHO's continuing programme of work. The achievements with respect to each objective are used to rank countries, as are the overall measures of achievement and performance described below. Since a given coun or health system may have very different ranks on different attainments, Annex Table shows the complete ranking for all Member States on all the measures. In several subse quent tables, countries are ranked in order of achievement or performance, and the order varies from one table to another. Since the ranking is based on estimates which include uncertainty as to the exact values, the rank assigned also includes uncertainty: a health system is not always assigned a specific position relative to all others but is estimated to lie somewhere within a narrower or broader range, depending on the uncertainties in the calculation. The ranks of different health systems therefore sometimes overlap to a greater or lesser degree, and two or more countries may have the same rank. Health is the defining objective for the health system. This means making the health status of the entire population as good as possible over people's whole life cycle, taking account of both premature mortality ar Annex Table 2 presents three conven tional and partial measures of health status, by country, without ranking: these are the probability of dying before age five years or between ages 15 and 59 years, and life expect ancy at birth. For the first time, these measures are presented with estimates of uncertainty nd these uncertainties carry over to subsequent calculations On the basis of the mortality figures, five strata are identified, ranging from low child and adult mortality to high chi mortality and very high adult death rates. Combining these strata with the six WHO Regions gives 14 subregions defined geographically and epidemiologically(see the list of Member States by WHO Region and mortality stratum) Annex Table 3 presents estimates of mortality by cause and sex in 1999 in each of these subregions(not by country), and Annex Table 4 combines these death rates with information about disability mates of one measure of overall population health: the burden of disease, that is, the num- bers of disability-adjusted life years(DALYs) lost. To assess overall population health and thus to judge how well the objective of good health is being achieved, WHO has chosen to use disability-adjusted life expectancy (ALe), which has the advantage of being directly comparable to life expectancy estimated from mortality alone and is readily compared across populations. Annex Table 5 provides esti mates for all countries of disability-adjusted life expectancy. DALE is estimated to equal or exceed 70 years in 24 countries, and 60 years in over half the Member States of WHO. At
How Well do Health Systems Perform? 27 MEASURING GOAL ACHIEVEMENT To assess a health system, one must measure five things: the overall level of health; the distribution of health in the population; the overall level of responsiveness; the distribution of responsiveness; and the distribution of financial contribution. For each one, WHO has used existing sources or newly generated data to calculate measures of attainment for the countries where information could be obtained. These data were also used to estimate values when particular numbers were judged unreliable, and to estimate attainment and performance for all other Member States. Several of these measures are novel and are explained in detail in the Statistical Annex, where all the estimates are given, along with intervals expressing the uncertainty or degree of confidence in the point estimate. The correct value for any indicator is estimated to have an 80% probability of falling within the uncertainty interval, with chances of 10% each of falling below the low value or above the high one. This recognition of inexactness underscores the importance of getting more and better data on all the basic indicators of population health, responsiveness and fairness in financial contribution, a task which forms part of WHO’s continuing programme of work. The achievements with respect to each objective are used to rank countries, as are the overall measures of achievement and performance described below. Since a given country or health system may have very different ranks on different attainments, Annex Table 1 shows the complete ranking for all Member States on all the measures. In several subsequent tables, countries are ranked in order of achievement or performance, and the order varies from one table to another. Since the ranking is based on estimates which include uncertainty as to the exact values, the rank assigned also includes uncertainty: a health system is not always assigned a specific position relative to all others but is estimated to lie somewhere within a narrower or broader range, depending on the uncertainties in the calculation. The ranks of different health systems therefore sometimes overlap to a greater or lesser degree, and two or more countries may have the same rank. Health is the defining objective for the health system. This means making the health status of the entire population as good as possible over people’s whole life cycle, taking account of both premature mortality and disability. Annex Table 2 presents three conventional and partial measures of health status, by country, without ranking: these are the probability of dying before age five years or between ages 15 and 59 years, and life expectancy at birth. For the first time, these measures are presented with estimates of uncertainty, and these uncertainties carry over to subsequent calculations. On the basis of the mortality figures, five strata are identified, ranging from low child and adult mortality to high child mortality and very high adult death rates. Combining these strata with the six WHO Regions gives 14 subregions defined geographically and epidemiologically (see the list of Member States by WHO Region and mortality stratum). Annex Table 3 presents estimates of mortality by cause and sex in 1999 in each of these subregions (not by country), and Annex Table 4 combines these death rates with information about disability to create estimates of one measure of overall population health: the burden of disease, that is, the numbers of disability-adjusted life years (DALYs) lost. To assess overall population health and thus to judge how well the objective of good health is being achieved, WHO has chosen to use disability-adjusted life expectancy (DALE), which has the advantage of being directly comparable to life expectancy estimated from mortality alone and is readily compared across populations. Annex Table 5 provides estimates for all countries of disability-adjusted life expectancy. DALE is estimated to equal or exceed 70 years in 24 countries, and 60 years in over half the Member States of WHO. At
The World Health Report 2000 the other extreme are 32 countries where disability-adjusted life expectancy is estimated to be less than 40 years. Many of these are countries with major epidemics of HiviAIds among other causes Box 2.1 describes how these summary measures of population health are constructed and how they are related Figure 2.2 summarizes the relation between DALE and life expectancy without adjust- ment, for each of the 14 subregions, for both men and women. The adjustment is nearly uniform, at about seven years of healthy life equivalent lost to disability. Both absolute and relatively this loss is slightly less for richer, low-mortality subregions, despite the fact that people live longer there and so have more opportunity to acquire non-fatal disabilities Disability makes a substantial difference in poorer countries because some limitations injury, blindness, paralysis and the debilitating effects of several tropical diseases such as malaria and shistosomiasis-strike children and young adults. Separating life expectancy into years in good health and years lived with disability therefore widens rather than nar- rows the difference in health status between richer and poorer populations. This is most evident in the share of life expectancy which is lost to disability: it ranges from less than 9% in the healthiest subregions to more than 14% in the least healthy. Annex Table 5 shows these shares for individual countries, where the range is even wider. Annex Table 5 also provides estimates of health inequality. The distributional measure of health ranges from 1 for the case of perfect equality to zero for extreme inequality, which corresponds to a fraction of the population having an expectancy of 100 years and the rest Box 2. 1 Summary measures of population health population; each way of estimating it violates ono ming up the health of a vival plus part of that for disability No measure is perfect for the purpose of sumI or another desirable cri- DALE is estimated from three kinds of information the fraction of the ion. The two principal approaches are the burden of disease, which meas- population surviving to each age, calculated from birth and death rates ures losses of good health compared to a long life free of disability, and the prevalence of each type of disability at each age; and the weight as- some measure of life expectancy, adjusted to take account of time lived signed to each type of disability, which may or may not vary with age. with a disability. Both ways of summarizing health use the same informa- Survival at each age is adjusted downward by the sum of all the disabi tion about mortality and disability, and Summarizing population health from mortality ffects, each of which is the product of a and disabil ight and the complement of a preva- curve, such as the bold line between 1 lence(the share of the population not suf- the areas labelled Disability and Mor- fering that disability). These adjusted survival shares are then divided by the The area labelled Mortality repre- initial population, before any mortality sents losses due to death, compared ccurred, to give the average number of to a high standard of life expectancy uivalent healthy life years that a new- the burden of disease corresponds to born member of the population could ex all of that area plus a fraction of the a ect to live. One important difference between the disability. Ihe fraction depends on the 8 burden of disease estimation using dis- disability weights assigned to various rs(DALYs)and that states between death and perfect of dale is that the former do but the lat- health. Life expectancy without any ter do not, distinguish the contribution of adjustment corresponds to the areas each diseaseto the overall result dale has I free of disability an Survival free of disabil Disability together, the whole area many choices of parameters for the cal- under the survivorship curve Disabil- culation, and it is directly comparable to ity-adjusted life expectancy(DALE) the more familiar notion of life expectancy then corresponds to the area for sur- 0 Source: Murray CJL, Salomon JA, Mathers CA critical examination of summary measures of population health. Geneva, World Health Organization, 1999(GPE Disaussion paper No. 1
28 The World Health Report 2000 the other extreme are 32 countries where disability-adjusted life expectancy is estimated to be less than 40 years. Many of these are countries with major epidemics of HIV/AIDS, among other causes. Box 2.1 describes how these summary measures of population health are constructed and how they are related. Figure 2.2 summarizes the relation between DALE and life expectancy without adjustment, for each of the 14 subregions, for both men and women. The adjustment is nearly uniform, at about seven years of healthy life equivalent lost to disability. Both absolutely and relatively this loss is slightly less for richer, low-mortality subregions, despite the fact that people live longer there and so have more opportunity to acquire non-fatal disabilities. Disability makes a substantial difference in poorer countries because some limitations – injury, blindness, paralysis and the debilitating effects of several tropical diseases such as malaria and shistosomiasis – strike children and young adults. Separating life expectancy into years in good health and years lived with disability therefore widens rather than narrows the difference in health status between richer and poorer populations. This is most evident in the share of life expectancy which is lost to disability: it ranges from less than 9% in the healthiest subregions to more than 14% in the least healthy. Annex Table 5 shows these shares for individual countries, where the range is even wider. Annex Table 5 also provides estimates of health inequality. The distributional measure of health ranges from 1 for the case of perfect equality to zero for extreme inequality, which corresponds to a fraction of the population having an expectancy of 100 years and the rest Box 2.1 Summary measures of population health No measure is perfect for the purpose of summing up the health of a population; each way of estimating it violates one or another desirable criterion. The two principal approaches are the burden of disease, which measures losses of good health compared to a long life free of disability, and some measure of life expectancy, adjusted to take account of time lived with a disability. Both ways of summarizing health use the same information about mortality and disability, and both are related to a survivorship curve, such as the bold line between the areas labelled Disability and Mortality in the figure. The area labelled Mortality represents losses due to death, compared to a high standard of life expectancy: the burden of disease corresponds to all of that area plus a fraction of the area corresponding to time lived with disability. The fraction depends on the disability weights assigned to various states between death and perfect health. Life expectancy without any adjustment corresponds to the areas labelled Survival free of disability and Disability together, the whole area under the survivorship curve. Disability-adjusted life expectancy (DALE) then corresponds to the area for survival plus part of that for disability. DALE is estimated from three kinds of information: the fraction of the population surviving to each age, calculated from birth and death rates; the prevalence of each type of disability at each age; and the weight assigned to each type of disability, which may or may not vary with age. Survival at each age is adjusted downward by the sum of all the disability effects, each of which is the product of a weight and the complement of a prevalence (the share of the population not suffering that disability). These adjusted survival shares are then divided by the initial population, before any mortality occurred, to give the average number of equivalent healthy life years that a newborn member of the population could expect to live. One important difference between the burden of disease estimation using disability-adjusted life years (DALYs) and that of DALE is that the former do, but the latter do not, distinguish the contribution of each disease to the overall result. DALE has the advantage that it does not require as many choices of parameters for the calculation, and it is directly comparable to the more familiar notion of life expectancy without adjustment. Source: Murray CJL, Salomon JA, Mathers C. A critical examination of summary measures of population health. Geneva, World Health Organization, 1999 (GPE Discussion paper No. 12). Summarizing population health from mortality and disability Age Percentage surviving Mortality Disability Survival free of disability 0 100
Howo Well do Health Systems Perform? If everyone had the same life for disability, the system would be perfectly fair with respect to health, even though people would actually die at different ages. For a small number of countries it has been possible to estimate the distribution of life expectancy within the population using information on both child Figure 2.2 Life expectancy and disability-adjusted life expectancy for males and females, by WHO Region and stratum defined by child mortality and adult mortality, 1999 75 Male life expectancy The dotted line represents a situation of no time lived with disability, so that life expectancy and disability-adjusted life expectancy coincide. 55 WHo Mortality stratum Males Females Region Child Life expectancy Disability adjusted Life expectancy Disability adjusted 446 470 ry low 74 610 70.7 633 62.3 EUR Very low Very low SEA WPR ry low Very low 680
How Well do Health Systems Perform? 29 having no expectation of surviving infancy. If everyone had the same life expectancy, adjusted for disability, the system would be perfectly fair with respect to health, even though people would actually die at different ages. For a small number of countries it has been possible to estimate the distribution of life expectancy within the population using information on both child Mortality stratum Males Females Region Child Adult Life expectancy Disability adjusted Life expectancy Disability adjusted AFR High High 52.0 44.6 54.9 47.0 High Very high 45.6 38.0 48.0 40.0 AMR Very low Very low 73.9 67.5 80.4 73.2 Low Low 67.3 60.6 74.1 66.8 High High 63.6 56.7 68.6 61.1 EMR Low Low 67.7 61.0 70.7 63.3 High High 60.0 53.0 62.3 54.7 EUR Very low Very low 74.5 68.1 80.8 73.7 Low Low 67.3 60.6 73.9 66.6 Low High 62.3 55.4 73.4 66.1 SEAR Low Low 67.2 60.5 73.1 65.7 High High 62.6 55.7 64.0 56.4 WPR Very low Very low 76.7 70.3 82.7 75.6 Low Low 68.0 61.3 72.3 65.0 Figure 2.2 Life expectancy and disability-adjusted life expectancy for males and females, by WHO Region and stratum defined by child mortality and adult mortality, 1999 35 40 45 50 55 60 65 70 75 80 40 45 50 55 60 65 70 75 80 85 Disability-adjusted life expectancy at birth (years) Life expectancy at birth (years) Male life expectancy Female life expectancy WHO The dotted line represents a situation of no time lived with disability, so that life expectancy and disability-adjusted life expectancy coincide
The World Health Report 2000 and adult mortality; these results are presented below. For most countries, however, it has so far been possible to use only child mortality data. Because high-income countries have largely eliminated child mortality, the highest ranking countries in Annex Table 5 nearly all have relatively high incomes; most are European. A few Latin American countries which Figure 2.3 Inequality in life expectancy at birth, by sex, in six countries Australia 1992 Chile 1993 出 0 Life expectancy at birth(years) Japan 1990 Mexico 1993 40 Females Females Life expectancy at birth(years) Norway 1996 USA 1990 Males 20 Life expectancy at birth (years) Life expectancy at birth(years)
30 The World Health Report 2000 and adult mortality; these results are presented below. For most countries, however, it has so far been possible to use only child mortality data. Because high-income countries have largely eliminated child mortality, the highest ranking countries in Annex Table 5 nearly all have relatively high incomes; most are European. A few Latin American countries which 54 60 66 72 78 84 90 0 10 20 30 40 50 Percentage of population Life expectancy at birth (years) Mexico 1993 54 60 66 72 78 84 90 0 10 20 30 40 50 Percentage of population Life expectancy at birth (years) Males Females Australia 1992 54 60 66 72 78 84 90 0 10 20 30 40 50 Percentage of population Life expectancy at birth (years) USA 1990 54 60 66 72 78 84 90 0 10 20 30 40 50 Percentage of population Life expectancy at birth (years) Japan 1990 54 60 66 72 78 84 90 0 10 20 30 40 50 Percentage of population Life expectancy at birth (years) Chile 1993 54 60 66 72 78 84 90 0 10 20 30 40 50 Percentage of population Life expectancy at birth (years) Norway 1996 Figure 2.3 Inequality in life expectancy at birth, by sex, in six countries Males Females Males Females Males Females Males Females Males Females