143 Statistical annex The tables in this annex present new concepts and measures which lay the 1 basis for pe the report provides detail on the different goals for health systems and the measures of performance. The material in these tables will be presented on an annual basis in each World health report. As with any innovative approach, methods and data sources can be refined and improved. It is hoped that careful scrutiny and use of results will lead to progressively better meas- urement of health system performance in the coming World health re ports. All the main results are reported with uncertainty intervals in order to communicate to the user the plausible range of estimates for each country on each measure
143 tatistical nnex The tables in this annex present new concepts and measures which lay the empirical basis for assessing health system performance. The main body of the report provides detail on the different goals for health systems and the measures of performance. The material in these tables will be presented on an annual basis in each World health report. As with any innovative approach, methods and data sources can be refined and improved. It is hoped that careful scrutiny and use of results will lead to progressively better measurement of health system performance in the coming World health reports. All the main results are reported with uncertainty intervals in order to communicate to the user the plausible range of estimates for each country on each measure. 143
144 STATISTICAL ANNEX EXPLANATORY NOTES The tables in this technical annex present new concepts and measures which lay the empirical basis for assessing health system performance. The main body of the report provides detail on the different goals for health systems and the measures of performance Both the text of the report and the annex are based on the WHO framework for health stem performance assessment. The work leading to these annex tables was undertaken mostly by the WHo Global Programme on Evidence for Health policy in collaboration with counterparts from the Regional Offices of WHO. This analytical effort was organized in eleven working groups. Membership of these working groups is listed in the Appendix. The material in these tables will be presented on an annual basis in each World health report. Because this is the first year of presentation for the material in Annex Tables 1 and 5-10, working papers have been prepared which provide details on the concepts, methods and results that are only briefly mentioned here. The footnotes to these technical notes include a complete listing of the detailed working papers As with any innovative approach, methods and data sources can be refined and im proved. It is hoped that careful scrutiny and use of the results will lead to progressively better measurement of performance in the coming World health reports. All the main results are reported with uncertainty intervals in order to communicate to the user the plausibl range of estimates for each country on each measure Although not provided in any table, extensive use has been made of estimates of in ome per capita in international dollars, average years of schooling for the population over age 15 years, percentage of the population in absolute poverty and the income Gini coeffi cient. In all cases, there are multiple and often conflicting sources of information from inter- national agencies on these indicators; in addition, there are many countries for which there are no published estimates To facilitate the analyses presented here, consistent and com plete estimates of these key indicators have been developed through a variety of tech- niques including factor analysis, multiple imputation methods for missing data, remote sensing data from public use satellites and systematic reviews of household survey data The details on methods and data sources for the final figures on income per capita, educa tional attainment, poverty and income distribution are outlined elsewhere. ANNEX TABLE 1 Annex Table 1 is designed as a guide for using Annex Tables 5-7, 9 and 10. Each measure of goal attainment and performance -disability-adjusted life expectancy, health equality in terms of child survival, responsiveness level, responsiveness distribution, fairness of finan- cial contribution, performance on level of health, and overall health system performance reported as a league table ranked from the highest level of achievement or performance to the lowest level. Annex Table 1 lists countries alphabetically and provides the ranks on each of the measures reported in the other tables. The reader can use Annex Table 1 to identify quickly where a particular country falls in each table
144 The World Health Report 2000 STATISTICAL ANNEX EXPLANATORY NOTES T he tables in this technical annex present new concepts and measures which lay the empirical basis for assessing health system performance. The main body of the report provides detail on the different goals for health systems and the measures of performance. Both the text of the report and the annex are based on the WHO framework for health system performance assessment.1 The work leading to these annex tables was undertaken mostly by the WHO Global Programme on Evidence for Health Policy in collaboration with counterparts from the Regional Offices of WHO. This analytical effort was organized in eleven working groups. Membership of these working groups is listed in the Appendix. The material in these tables will be presented on an annual basis in each World health report. Because this is the first year of presentation for the material in Annex Tables 1 and 5-10, working papers have been prepared which provide details on the concepts, methods and results that are only briefly mentioned here. The footnotes to these technical notes include a complete listing of the detailed working papers. As with any innovative approach, methods and data sources can be refined and improved. It is hoped that careful scrutiny and use of the results will lead to progressively better measurement of performance in the coming World health reports. All the main results are reported with uncertainty intervals in order to communicate to the user the plausible range of estimates for each country on each measure. Although not provided in any table, extensive use has been made of estimates of income per capita in international dollars, average years of schooling for the population over age 15 years, percentage of the population in absolute poverty and the income Gini coefficient. In all cases, there are multiple and often conflicting sources of information from international agencies on these indicators; in addition, there are many countries for which there are no published estimates. To facilitate the analyses presented here, consistent and complete estimates of these key indicators have been developed through a variety of techniques including factor analysis, multiple imputation methods for missing data, remote sensing data from public use satellites and systematic reviews of household survey data. The details on methods and data sources for the final figures on income per capita, educational attainment, poverty and income distribution are outlined elsewhere.2 ANNEX TABLE 1 Annex Table 1 is designed as a guide for using Annex Tables 5-7, 9 and 10. Each measure of goal attainment and performance - disability-adjusted life expectancy, health equality in terms of child survival, responsiveness level, responsiveness distribution, fairness of financial contribution, performance on level of health, and overall health system performance -is reported as a league table ranked from the highest level of achievement or performance to the lowest level. Annex Table 1 lists countries alphabetically and provides the ranks on each of the measures reported in the other tables. The reader can use Annex Table 1 to identify quickly where a particular country falls in each table
Statistical annex ANNEX TABLE 2 To assess the performance of health systems in terms of health achievement, it was crucial to develop the best possible assessment of the life table for each country. New life tables have been developed for all 191 Member States starting with a systematic review of all available evidence from surveys, censuses, sample registration systems, population labo ratories and vital registration on levels and trends in child mortality and adult mortality This review benefited greatly from the work undertaken on child mortality by UNICEPS and the UN Population Division 1998 demographic assessment. To aid in demographic, cause of death and burden of disease analysis, the 191 Member States have been divided into 5 mortality strata on the basis of their level of child(5q0) and adult male mortality (45q15) The matrix defined by the six WHO Regions and the 5 mortality strata leads to 14 ubregions, since not every mortality stratum is represented in every Region. These subregic are used in Tables 3 and 4 for presentation of results. Because of increasing heterogeneity of patterns of adult and child mortality, WHO has developed a system of two-parameter logit life tables for each of the 14 subregions. This system of model life tables has been used extensively in the development of life tables for each Member State and in projecting life tables to 1999 when the most recent data avail able are from earlier years. Details on the data, methods and results by country of this life table analysis are available in the corresponding technical paper. A major innovation that WHO is introducing this year to demographic and other analy- ses is the reporting of uncertainty intervals To capture the uncertainty due to sampling indirect estimation technique or projection to 1999, a total of 1000 life tables have been developed for each Member State. Uncertainty bounds are reported in Annex Table 1 by giving key life table values at the 10th percentile and the 90th percentile. This uncertainty analysis was facilitated by the development of new methods and software tools. In coun tries with a substantial HIV epidemic, recent estimates of the level and uncertainty range of the magnitude of the HiV epidemic have been incorporated into the life table uncertainty ANNEX TABLES 3 AND 4 Causes of death for the 14 subregions and the world have been estimated based on data from national vital registration systems that capture 16.7 million deaths annually. In addi tion, information from sample registration systems, population laboratories and epidemic logical analyses of specific conditions have been used to produce better estimates of the cause of death pattems Cause of death data have been carefully analysed to take into account incomplete cov- erage of vital registration in countries and the likely differences in cause of death pattens that would be expected in the uncovered and often poorer sub-populations. Techniques to undertake this analysis have been developed based on the global burden of disease study? nd further refined using a much more extensive database and more robust modelling techniques Special attention has been paid to problems of misattribution or miscoding of causes of death in cardiovascular diseases, cancer, injuries and general ill-defined categories. A cor- rection algorithm for reclassifying ill-defined cardiovascular codes has been developed Cancer mortality by site has been evaluated using both vital registration data and popula tion based cancer incidence registries. The latter have been analysed using a complete age, period cohort model of cancer survival in each region. 2
Statistical Annex 145 ANNEX TABLE 2 To assess the performance of health systems in terms of health achievement, it was crucial to develop the best possible assessment of the life table for each country. New life tables have been developed for all 191 Member States starting with a systematic review of all available evidence from surveys, censuses, sample registration systems, population laboratories and vital registration on levels and trends in child mortality and adult mortality. This review benefited greatly from the work undertaken on child mortality by UNICEF3 and the UN Population Division 1998 demographic assessment.4 To aid in demographic, cause of death and burden of disease analysis, the 191 Member States have been divided into 5 mortality strata on the basis of their level of child (5q0) and adult male mortality (45q15). The matrix defined by the six WHO Regions and the 5 mortality strata leads to 14 subregions, since not every mortality stratum is represented in every Region. These subregions are used in Tables 3 and 4 for presentation of results. Because of increasing heterogeneity of patterns of adult and child mortality, WHO has developed a system of two-parameter logit life tables for each of the 14 subregions.5 This system of model life tables has been used extensively in the development of life tables for each Member State and in projecting life tables to 1999 when the most recent data available are from earlier years. Details on the data, methods and results by country of this life table analysis are available in the corresponding technical paper.6 A major innovation that WHO is introducing this year to demographic and other analyses is the reporting of uncertainty intervals. To capture the uncertainty due to sampling, indirect estimation technique or projection to 1999, a total of 1000 life tables have been developed for each Member State. Uncertainty bounds are reported in Annex Table 1 by giving key life table values at the 10th percentile and the 90th percentile. This uncertainty analysis was facilitated by the development of new methods and software tools.7 In countries with a substantial HIV epidemic, recent estimates of the level and uncertainty range of the magnitude of the HIV epidemic have been incorporated into the life table uncertainty analysis.8 ANNEX TABLES 3 AND 4 Causes of death for the 14 subregions and the world have been estimated based on data from national vital registration systems that capture 16.7 million deaths annually. In addition, information from sample registration systems, population laboratories and epidemiological analyses of specific conditions have been used to produce better estimates of the cause of death patterns. Cause of death data have been carefully analysed to take into account incomplete coverage of vital registration in countries and the likely differences in cause of death patterns that would be expected in the uncovered and often poorer sub-populations. Techniques to undertake this analysis have been developed based on the global burden of disease study9 and further refined using a much more extensive database and more robust modelling techniques.10 Special attention has been paid to problems of misattribution or miscoding of causes of death in cardiovascular diseases, cancer, injuries and general ill-defined categories. A correction algorithm for reclassifying ill-defined cardiovascular codes has been developed.11 Cancer mortality by site has been evaluated using both vital registration data and population based cancer incidence registries. The latter have been analysed using a complete age, period cohort model of cancer survival in each region.12
The World Health Report 2000 Annex Table 4 provides estimates of the burden of disease using disability-adjusted life years (DALYs)as a measure of the health gap in the world in 1999. DALYs along with disability-adjusted life expectancy are summary measures of population health. DALYS are a type of health gap that measures the difference between a populations health and normative goal of living in full health. For a review of the development of DALYs and recent advances in the measurement of the burden of disease see murray lopez. DALYs have been estimated based on cause of death information for each Region and regional assess ments of the ep ology of major disabling conditions ANNEX TABLE 5 Annex Table 5 provides measurements of health attainment in terms of the average level of population health and the distribution of population health or health equality. iwo meas- ures are reported by WHo for the first time at the country level: disability-adjusted life expectancy and the index of equality of child survival Achievement of the average level of population health is reported in terms of disability. djusted life expectancy (DALE). daLE is most easily understood as the expectation of life lived in equivalent full health. As a summary measure of the burden of disability from all causes in a population, DALE has two advantages over other summary measures. The first that it is relatively easy to explain the concept of a lifespan without disability to a non- technical audience. The second is that it is easy to calculate DALE using the Sullivan method based on age-specific information on the prevalence of non-fatal health outcomes. In the global burden of disease study, dale was estimated at the regional level, based on the estimates of all disabling sequelae included in the study. Disability weights were measured for each of these sequelae for five standard age groups, sex and eight regions. National estimates of dale are based on the life tables for each member state summa rized in Annex Table 2, population representative sample surveys assessing physical and cognitive disability and general health status, and detailed information on the epidemiol ogy of major disabling conditions in each country. Use of household surveys is complicated by the variation in self-assessed health for a given level of observed health as a function of ex, age, socioeconomic status, exposure to health services, and culture. 5, I6 The methodo- logical details for national estimates of dale and the uncertainty in these estimates are provided elsewhere.17 Measurement of achievement in the distribution of health is based on the who frame work for measuring health inequality. s The intention is ultimately to measure the distribu tion of health using the distribution of DALe across individuals. However, the analysis of the distribution of DALE in each country has not yet been completed For selected coun- ies, the distribution of life expectancy across small areas has been completed and reveals lat there is often much greater variation in life expectancy and probably in DALE than expected. 9 In this World health report, the analysis of achievement in the distribution of health, presented in Annex Table 5, is the index of equality of child survival. It is based on the distribution of child survival across countries, and takes advantage of the widely ava able and extensive information on complete birth histories in the demographic and health surveys and small area vital registration data on child mortality Statistical methods based on maximum likelihood estimation of the extended beta binomial distribution have been developed to distinguish between variation across moth ers in the number of children who have died due to chance and that due to differences in the underlying risks of death. 2 This statistical method has been applied to demographic and health survey data and small area data from more than 60 countries to estimate the
146 The World Health Report 2000 Annex Table 4 provides estimates of the burden of disease using disability-adjusted life years (DALYs) as a measure of the health gap in the world in 1999. DALYs along with disability-adjusted life expectancy are summary measures of population health.13 DALYs are a type of health gap that measures the difference between a population’s health and a normative goal of living in full health. For a review of the development of DALYs and recent advances in the measurement of the burden of disease see Murray & Lopez.14 DALYs have been estimated based on cause of death information for each Region and regional assessments of the epidemiology of major disabling conditions. ANNEX TABLE 5 Annex Table 5 provides measurements of health attainment in terms of the average level of population health and the distribution of population health or health equality. Two measures are reported by WHO for the first time at the country level: disability-adjusted life expectancy and the index of equality of child survival. Achievement of the average level of population health is reported in terms of disabilityadjusted life expectancy (DALE). DALE is most easily understood as the expectation of life lived in equivalent full health. As a summary measure of the burden of disability from all causes in a population, DALE has two advantages over other summary measures. The first is that it is relatively easy to explain the concept of a lifespan without disability to a nontechnical audience. The second is that it is easy to calculate DALE using the Sullivan method based on age-specific information on the prevalence of non-fatal health outcomes. In the global burden of disease study, DALE was estimated at the regional level, based on the estimates of all disabling sequelae included in the study. Disability weights were measured for each of these sequelae for five standard age groups, sex and eight regions. National estimates of DALE are based on the life tables for each Member State summarized in Annex Table 2, population representative sample surveys assessing physical and cognitive disability and general health status, and detailed information on the epidemiology of major disabling conditions in each country. Use of household surveys is complicated by the variation in self-assessed health for a given level of observed health as a function of sex, age, socioeconomic status, exposure to health services, and culture.15, 16 The methodological details for national estimates of DALE and the uncertainty in these estimates are provided elsewhere.17 Measurement of achievement in the distribution of health is based on the WHO framework for measuring health inequality.18 The intention is ultimately to measure the distribution of health using the distribution of DALE across individuals. However, the analysis of the distribution of DALE in each country has not yet been completed. For selected countries, the distribution of life expectancy across small areas has been completed and reveals that there is often much greater variation in life expectancy and probably in DALE than expected.19 In this World health report, the analysis of achievement in the distribution of health, presented in Annex Table 5, is the index of equality of child survival. It is based on the distribution of child survival across countries, and takes advantage of the widely available and extensive information on complete birth histories in the demographic and health surveys and small area vital registration data on child mortality. Statistical methods based on maximum likelihood estimation of the extended betabinomial distribution have been developed to distinguish between variation across mothers in the number of children who have died due to chance and that due to differences in the underlying risks of death.20 This statistical method has been applied to demographic and health survey data and small area data from more than 60 countries to estimate the
Statistical annex underlying distribution of the risk of child death. For the purposes of calculating the index of equality of child survival, child mortality distributions have been transformed into distri- butions of expected survival time under age 5 years. The resulting distributions of survival time have been summarized for the creation of a composite index using the following Equality of child survival 1-isl j-l here x is the survival time of a given child and x is the mean survival time across children The particular form of this summary measure of inequality has been selected on the basis of a survey of preferences for measuring health inequality of over one thousand re- spondents. 22 Because all measures of goal achievement are intended to be positive meas- ures, the inequality index has been transformed into an index of equality by calculating one minus child survival inequality, as shown above. As the measure of inequality has a maxi mum value that can be greater than 1, in theory this transformed measure of equality of child survival could be negative. However, across the range of countries, no country has a egree of inequality that would lead to a measurement of equality less than zero. The value of 1 can be interpreted as complete equality and zero can be interpreted as a degree of inequality that is worse than has been seen in any country measured directly or estimated indirectly to date For countries without a demographic and health survey or small area data, the index of the distribution of health for child survival has been estimated using indirect techniques and information on important covariates of health inequality such as poverty, educational attainment and the level of child mortality. ANNEX TABLE 6 The measurement of achievement in the level of responsiveness was based on a survey of nearly two thousand key informants in selected countries. Key informants were aske to evaluate the performance of their health system regarding seven elements of respon siveness: dignity, autonomy and confidentiality jointly termed respect of persons); and ompt attention, quality of basic amenities, access to social support networks during care nd choice of care provider(encompassed by the term client orientation). The elements were scored from 0 to 10. Scores on each component were combined into a composite score for responsiveness based on results of the survey on preferences for health system performance assessment. For other countries, achievement in the level of responsivenes has been estimated using indirect techniques and information on important covariates of responsiveness. To enhance the measurement of responsiveness, WHO is actively devel oping and field testing instruments to measure responsiveness from household respond ents. This strategy of using household surveys will be supplemented with facility surveys to re directly some components of responsiveness he measurement of achievement in the distribution of responsiveness reflected in Annex 6 is based on a very simple approach. Respondents in the key informants survey were asked to identify groups who were disadvantaged with regard to responsiveness.The number of times a particular group was identified as disadvantaged was used to calculate a key informant intensity score. Four groups had high key informant intensity scores: poor peo- ple, women, old people, and indigenous groups or racially disadvantaged groups (in most stances minorities). The key informant intensity scores for these four groups were multi
Statistical Annex 147 underlying distribution of the risk of child death.21 For the purposes of calculating the index of equality of child survival, child mortality distributions have been transformed into distributions of expected survival time under age 5 years. The resulting distributions of survival time have been summarized for the creation of a composite index using the following formula: where x is the survival time of a given child and x is the mean survival time across children. The particular form of this summary measure of inequality has been selected on the basis of a survey of preferences for measuring health inequality of over one thousand respondents.22 Because all measures of goal achievement are intended to be positive measures, the inequality index has been transformed into an index of equality by calculating one minus child survival inequality, as shown above. As the measure of inequality has a maximum value that can be greater than 1, in theory this transformed measure of equality of child survival could be negative. However, across the range of countries, no country has a degree of inequality that would lead to a measurement of equality less than zero. The value of 1 can be interpreted as complete equality and zero can be interpreted as a degree of inequality that is worse than has been seen in any country measured directly or estimated indirectly to date. For countries without a demographic and health survey or small area data, the index of the distribution of health for child survival has been estimated using indirect techniques and information on important covariates of health inequality such as poverty, educational attainment and the level of child mortality. ANNEX TABLE 6 The measurement of achievement in the level of responsiveness was based on a survey of nearly two thousand key informants in selected countries.23 Key informants were asked to evaluate the performance of their health system regarding seven elements of responsiveness: dignity, autonomy and confidentiality (jointly termed respect of persons); and prompt attention, quality of basic amenities, access to social support networks during care and choice of care provider (encompassed by the term client orientation). The elements were scored from 0 to 10. Scores on each component were combined into a composite score for responsiveness based on results of the survey on preferences for health system performance assessment. For other countries, achievement in the level of responsiveness has been estimated using indirect techniques and information on important covariates of responsiveness.24 To enhance the measurement of responsiveness, WHO is actively developing and field testing instruments to measure responsiveness from household respondents. This strategy of using household surveys will be supplemented with facility surveys to observe directly some components of responsiveness.25 The measurement of achievement in the distribution of responsiveness reflected in Annex Table 6 is based on a very simple approach. Respondents in the key informants survey were asked to identify groups who were disadvantaged with regard to responsiveness. The number of times a particular group was identified as disadvantaged was used to calculate a key informant intensity score. Four groups had high key informant intensity scores: poor people, women, old people, and indigenous groups or racially disadvantaged groups (in most instances minorities). The key informant intensity scores for these four groups were multi- ΣΣ 1 – n i=1 j=1 xi – xj 3 2 –– 0.5 Equality of child survival = n 2n x