W O R L D B A N K W O R K I N G P A P E R N O . 3 3 Millennium Development Goals for Health in Europe and Central Asia Relevance and Policy Implications Kara Sea Barents Sea R U S S I A N F E D E R A T I O N Baltic ESTONIA Sea LATVIA LITHUANIA RUSSIAN FED. BELARUS POLAND CZECH REP. SLOVAK UKRAINE REP. K A Z A K H S T A N HUNGARY MOLDOVA ROMANIA Sea of Azov SLOVENIA Aral CROATIA Sea BOSNIA/HERZ. BULGARIA B l a c k S e a SERBIA & GEORGIA Caspian UZBEKISTAN MONTENEGRO Sea KYRGYZ REP. ALBANIA ARMENIAAZERBAIJAN FYR MACEDONIA TURKEY TURKMENISTAN TAJIKISTAN Mediterranean Sea THE WORLD BANK W O R L D B A N K W O R K I N G P A P E R N O . 3 3 Millennium Development Goals for Health in Europe and Central Asia Relevance and Policy Implications THE WORLD BANK Washington, D.C. Copyright © 2004 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing: June 2004 printed on recycled paper 1 2 3 4 06 05 04 World Bank Working Papers are published to communicate the results of the Bank's work to the development community with the least possible delay. The manuscript of this paper therefore has not been prepared in accordance with the procedures appropriate to formally-edited texts. Some sources cited in this paper may be informal documents that are not readily available. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and do not necessarily reflect the views of the Board of Executive Directors of the World Bank or the governments they represent. The World Bank cannot guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply on the part of the World Bank any judgment of the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this publication is copyrighted. The World Bank encourages dissemination of its work and normally will grant permission for use. Permission to photocopy items for internal or personal use, for the internal or personal use of specific clients, or for educational classroom use, is granted by the World Bank, provided that the appropriate fee is paid. Please contact the Copyright Clearance Center before photocopying items. Copyright Clearance Center, Inc. 222 Rosewood Drive Danvers, MA 01923, U.S.A. Tel: 978-750-8400 · Fax: 978-750-4470. For permission to reprint individual articles or chapters, please fax your request with complete information to the Republication Department, Copyright Clearance Center, fax 978-750-4470. All other queries on rights and licenses should be addressed to the World Bank at the address above, or faxed to 202-522-2422. ISBN: 0-8213-5878-2 eISBN: 0-8213-5879-0 ISSN: 1726-5878 Library of Congress Cataloging-in-Publication Data has been requested. CONTENTS Abstract vii Acknowledgments ix Abbreviations and Acronyms xi Executive Summary 1 1. Introduction 3 2. Literature Review 7 3. Data Quality 9 4. Methods 15 5. Results 17 Life Expectancies in the ECA Region 17 Scenario 1: Achieving MDG Targets for Infant, Child and Maternal Mortality 17 Scenario 2: Reducing Infant, Child and Maternal Mortality to EU Levels 19 Scenario 3: Reducing Infant, Child and Maternal Mortality to the Lowest Subregional Levels 23 Scenario 4: Reducing Deaths from Cardiovascular Disease and External Causes of Death to EU Levels 25 Comparisons of Age-specific Deaths from Cardiovascular Disease and External Causes to Sweden 30 Overview of Scenarios 30 6. Discussion 39 HIV/AIDS 39 Moving Forward 40 7. Recommendations 43 Appendix 47 Impact on Years of Potential Life Lost 47 References 55 LIST OF TABLES Table 1 Different Estimates of Infant Mortality in Central Asia and Caucasus 10 Table 2 Sources of Childhood Mortality Estimates 11 Table 3 Millennium Development Goals for Infant Mortality 20 Table 4 Millennium Development Goals for Infant Mortality (Central Asia and Caucasus) 20 iii iv CONTENTS Table 5 Millennium Development Goals for 1­4 Mortality .21 Table 6 Millennium Development Goals for 1­4 Mortality (Central Asia and Caucasus) .21 Table 7 Millennium Development Goals for Maternal Mortality .22 Table 8 Scenario 1: Impact on Life Expectancy of Reaching Millennium Development Goals 4 and 5 .23 Table 9 Scenario 1: Impact on Life Expectancy of Reaching Millennium Development Goals 4 and 5 (Central Asia and Caucasus) .24 Table 10 Scenario 2: Impact on Life Expectancy of Reducing Infant, Child and Maternal Mortality to EU Levels .25 Table 11 Scenario 2: Impact on Life Expectancy of Reducing Infant, Child and Maternal Mortality to EU Levels (Central Asia and Caucasus) .26 Table 12 Lowest Subregional Levels of Infant, Child and Maternal Mortality .27 Table 13 Scenario 3: Impact on Life Expectancy of Reducing Infant, Child and Maternal Mortality to the Lowest Levels Achieved in the Subregion .27 Table 14 Scenario 3: Impact on Life Expectancy of Reducing Infant, Child and Maternal Mortality to the Lowest Levels Achieved in the Subregion (Central Asia and Caucasus) .28 Table 15 Scenario 4: Impact on Life Expectancy of Reducing Adult Mortality from Injuries and Violence and Cardiovascular Disease to EU Levels .29 Table 16 Age-specific Death Rates from Cardiovascular Disease per 100,000 Population in Sweden and Selected ECA Countries (latest available years) .30 Table 17 Age-specific Death Rates from External Causes of Death per 100,000 Population in Sweden and Selected ECA Countries (latest available years) .33 Table 18 Overall Impact on Life Expectancy According to the Different Scenarios .36 Table 19 Overall Impact on Life Expectancy According to the Different Scenarios (Central Asia and Caucasus) .37 Table 20 Impact on YPLL (age 0­64 per 10,000) of Reaching the Millennium Development Goals 4 and 5 .48 Table 21 Impact on YPLL (age 0­64 per 10,000) of Reducing Infant, Child and Maternal Mortality to EU Levels .49 Table 22 Impact on YPLL (age 0­64 per 10,000) of Reaching the Lowest Infant, Child and Maternal Mortality in the Subregion .50 Table 23 Impact on YPLL (age 0­64 per 10,000) of Reducing Adult Mortality from Injuries and Violence and Cardiovascular Disease to EU Levels .51 Table 24 Impact on YPLL (age 0­64 per 10,000) in the Four Scenarios .52 LIST OF FIGURES Figure 1 Infant Mortality Rates in Tajikistan According to WHO, WB, and UNICEF .13 Figure 2 Life Expectancies in the ECA Region (except Central Asia and Caucasus) .18 Figure 3 Life Expectancies in Central Asia and the Caucasus .19 Figure 4 Age-specific CVD Death Rates in Russia as Compared to Sweden .31 Figure 5 Age-specific CVD Death Rates in Kazakhstan as Compared to Sweden .31 Figure 6 Age-specific CVD Death Rates in Bulgaria as Compared to Sweden .32 CONTENTS v Figure 7 Age-specific CVD Death Rates in Macedonia as Compared to Sweden .32 Figure 8 Age-specific Death Rates from External Causes of Death in Russia as Compared to Sweden .33 Figure 9 Age-specific Death Rates from External Causes of Death in Kazakhstan as Compared to Sweden .34 Figure 10 Age-specific Death Rates from External Causes of Death in Lithuania as Compared to Sweden .34 Figure 11 Age-specific Death Rates from External Causes of Death in Croatia as Compared to Sweden .35 Figure 12 Overall Impact on Life Expectancy According to the Different Scenarios: Regional and Subregional averages .38 Figure 13 Overall Impact on YPLL According to the Different Scenarios (age 0­64 per 10,000): Regional and Subregional averages .53 LIST OF BOXES Box 1 The Millennium Development Goals 4 Box 2. Data quality in Central Asia and the Caucasus .13 v ABSTRACT T his study aims to contribute to the debate about the appropriateness of health-related Millen- nium Development Goals (MDGs) for the countries of the Europe and Central Asia region. It is primarily addressed at policy advisors and senior analysts at the local, regional, and global level, who could influence the strategic directions of policymakers. The study is of particular relevance to international development institutions and UN organizations, who might have to reconsider the focus of their health-related development assistance to the countries of the ECA region. The study examines how appropriate the health-related MDGs are for the countries of the ECA region by analysing the impact of the following four scenarios on life expectancy at birth: I achieving MDG targets for infant, child, and maternal mortality; I reducing infant, child, and maternal mortality to EU levels; I reducing infant, child, and maternal mortality to the lowest subregional levels; I reducing deaths from cardiovascular disease and external causes of death to EU levels. Because national registration data on infant and child mortality in countries of Central Asia and the Caucasus underestimate the true figure, the first three scenarios were recalculated for these countries in an additional calculation, using survey-based World Bank data. vii ACKNOWLEDGMENTS T his work was financed by the World Bank, from a trust fund provided by the Government of the Netherlands. However the findings, interpretations, and conclusions expressed herein are those of the authors and do not necessarily reflect the views of the Board of Executive Directors of the World Bank or the governments they represent. Bernd Rechel, Laidon Shapo, and Martin McKee, of the London School of Hygiene and Tropical Medicine, were the principal authors of the report. They worked with a counterpart team in the Health, Nutrition and Population Group of the Europe and Central Asia Region of the World Bank, which included Olusoji Adeyi, Enis Baris, Sarbani Chakraborty, Armin Fidler, Dominic Haazen, Loraine Hawkins, Akiko Maeda, Daniel Miller, Michael Mills, Toomas Palu, Silviu Radulescu, and Alan Wright. The authors are especially grateful to Olusoji Adeyi, who initiated and encouraged this work, and to Eduard Bos (of the Health, Nutrition and Population Central Unit in the World Bank) for information on mortality estimates and his comments on earlier versions of this report, to Marc Suhrcke of the World Health Organization (formerly of UNICEF) who provided valuable informa- tion on data quality, and to Karen Lock of the London School of Hygiene and Tropical Medicine, Vladimir Shkolnikov, of the Max Planck Institute for Demographic Research, and Evgueni Andreev, of the Russian Academy of Sciences, who contributed to the development of the method- ology used here. Vice President Shigeo Katsu Sector Director Michal Rutkowski Sector Manager Armin Fidler Team Leader Olusoji Adeyi ix ABBREVIATIONS AND ACRONYMS AIDS Acquired Immune Deficiency Syndrome CVD Cardiovascular Disease DHS Demographic and Health Survey DOTS Directly Observed Treatment, Short Course ECA Europe and Central Asia EU European Union HIV Human Immunodeficiency Virus MDG Millennium Development Goal OECD Organization for Economic Cooperation and Development UNDP United Nations Development Program UNICEF United Nations Children's Fund WHO World Health Organization YPLL Years of Potential Life Lost xi EXECUTIVE SUMMARY I n 2000 the international community adopted the Millennium Development Goals (MDGs) as the basis for international development policies. Covering a range of sectors, they have the advantage of providing common, objectively verifiable measures of progress in world devel- opment. Yet the application of this common framework to diverse settings can create challenges. A particular example is the application of health goals to the countries of the Europe and Central Asia (ECA) region. Although this region faces formidable health challenges, with low levels of life expectancy relative to European Union (EU) levels, this is driven largely by high adult mortality, while levels of infant and child mortality (which feature prominently in the health Millennium Development Goals) are relatively low, compared to countries at a similar level of economic performance in other parts of the world. This report examines how appropriate the health-related MDGs are for the countries of the ECA region by assessing the impact of achieving them on one of the widely used measures of aggre- gate health status: life expectancy at birth. It looks at the consequences of reaching the Millennium Development Goals 4 and 5, relating to infant, child (1­4 years-old), and maternal mortality. The report also explores other indicators that might be more relevant to the countries of this region. Achievement of Goal 6 of the MDGs (combating HIV/AIDS, malaria, and tuberculosis) has not been included in the analysis, because the impact of these diseases on life expectancy at birth is still relatively small. However, this may change in the near future due to HIV/AIDS and tuberculosis. Four scenarios were considered: (1) reducing infant, child, and maternal mortality rates to the levels envisaged by the MDGs; (2) reducing infant, child, and maternal mortality rates to average EU levels; (3) reducing infant, child, and maternal mortality rates to the lowest levels in the sub- region; and (4) keeping infant, child, and maternal mortality rates constant, and reducing mortality rates from cardiovascular disease and external causes of death to EU levels. The effects of the calcu- lations were then compared with regard to life expectancy at birth. Because it is well known that national registration data on infant and child mortality consider- ably underestimate the true figures in Central Asia and the Caucasus, survey-based data were used in an additional calculation of the first three scenarios in these countries. The results showed the extent to which the underestimation of under-5 mortality in official data seriously overestimates life 1 2 WORLD BANK WORKING PAPER expectancy at birth. Consequently, achieving reductions of infant and child mortality in the first three scenarios would result in much more pronounced gains in life expectancy, when compared to national registration data. For the whole of the ECA region, Scenario 4 (focusing on adult mortality) had the greatest im- pact on life expectancy at birth, resulting in an average gain of 7.75 years, and reaching 10.09 years in the Russian Federation. In contrast, reaching the MDG targets for infant, child, and maternal mortality resulted in an average gain of only 0.68­1.24 years, depending on whether national reg- istration data or World Bank estimates of infant and child mortality were used for the countries of Central Asia and the Caucasus. Reducing infant, child, and maternal mortality to EU levels resulted on average in a gain of 0.85­1.97 years; while reaching the best subregional values for these indicators resulted in an average gain of 0.58­1.09 years. The MDGs are important because they drive the choice of policies supported by the inter- national community. Overall, the study confirmed the importance of adapting the health-related MDGs Goals to the countries of the ECA region, placing greater emphasis on adult mortality. The choice of goals at a global level is largely determined by what data are available. The absence, in many developing countries, of data on adult mortality thus precludes the use of life expectancy at birth as a global measure. It does not, however, preclude its use where data do exist, and, as this analysis shows, specific measures of adult mortality such as deaths from cardiovascular disease and external causes are appropriate additional measures of progress in improving health in this region. The study also highlighted how different sources of data yield different values for commonly cited variables, raising important questions about how, in practice, progress towards the MDGs will actually be assessed. Perhaps equally important, the study demonstrated the great variety in challenges to health among the countries of the ECA region. Although it might be tempting to formulate common goals for all countries of the ECA region, ultimately, development targets and policy interventions need to be adapted to the country context to become meaningful. Implications for Policy Choices The study has important implications for policy choices at the regional, subregional, and country levels. Despite the fact that averages mask variations that exist within countries, certain conclusions stand out: I For the ECA region as a whole, proportionately more gains in life expectancy would accrue from the control of non-communicable diseases than from achieving the targets in the classic MDGs. This pattern holds in all subregions. I There are proportionately more gains to be had from achieving the classic MDGs in Central Asia and the Caucasus subregions. For priority setting, policy formulation, programs, and development assistance, the key mes- sages from this study are: I the importance of reducing morbidity and premature mortality from NCDs and external causes across the entire region; I the need for particular attention to the classic MDG indicators (infant mortality rates, under-5 mortality rates, and maternal mortality ratios) in the countries of Central Asia and the Caucasus; I the importance of establishing effective and sustainable surveillance, vital registration, and health information systems to provide valid data for local decisions and international comparisons, and I the continued need for interagency collaboration in support of all these objectives. CHAPTER 1 INTRODUCTION I n September 2000, the world's leaders adopted the United Nations Millennium Declaration, calling for stronger global efforts to reduce poverty, improve health, and promote peace, human rights and environmental sustainability. The UN General Assembly asked the Secretary-General to prepare a road map for achieving the commitments made in the Declaration--resulting in the Millennium Development Goals (MDGs). The Goals reflect key aims of various UN development conferences in the 1990s. They also built on the International Development Goals created by the Organization for Economic Co-operation and Development (OECD) in 1996. The Millennium Development Goals include all but one of the OECD International Development Goals (UNDP, 2003). The Millennium Development Goals were affirmed at the March 2002 Monterrey Confer- ence on Financing for Development, the September 2002 Johannesburg Declaration on Sustainable Development, and the June 2003 G8 summit in Evian (UNDP, 2003). The Goals are now widely accepted as a framework for measuring development progress. Bilateral and multilateral institutions, including the World Bank, have made the MDGs a central focus of their development assistance. The overall aim of the MDGs is to reverse the spread of poverty and disease by 2015. The eight goals are backed by a plan of action that sets out 18 quantifiable targets, each using specified indicators. Three goals and four targets are directly related to health (Box 1). The focus of the health-related Goals, as of the MDGs in general, is on poor countries in the trop- ics. For example, 99 percent of worldwide maternal deaths occur in these countries (IMF et al., 2000). Problems are particularly grave in sub-Saharan Africa, which has the highest rates of under-five mortal- ity, maternal mortality, and HIV/AIDS infection in the world. The incidence of HIV/AIDS and hunger in this region is still increasing instead of decreasing (UNDP, 2003) In sub-Saharan Africa, progress so far on achieving the Millennium Development Goals has been very limited. If progress continues at the current pace, this region will not reach the Goals for under-five mortality until 2165. Concerns, however, have been raised as to whether the health-related Millennium Develop- ment Goals are appropriate to the transition countries of Eastern Europe and Central Asia. In these countries, under-five and maternal mortality are much lower than in countries at similar levels of economic development in other parts of the world. Adult mortality, on the other hand, 3 4 WORLD BANK WORKING PAPER BOX 1. THE MILLENNIUM DEVELOPMENT GOALS Goal 1: Eradicate extreme poverty and hunger Goal 2: Achieve Universal Primary Education Goal 3: Promote Gender Equality and Empower Women Goal 4*: Reduce child mortality Target 5: Reduce by two-thirds, between 1990 and 2015, the under-five mortality rate Indicators: Under-five mortality rate Infant mortality rate Proportion of one-year-old children immunized against measles Goal 5*: Improve maternal health Target 6: Reduce by three-quarters, between 1990 and 2015, the maternal mortality rate Indicators: Maternal mortality ratio Proportion of births attended by skilled health personnel Goal 6*: Combat HIV/AIDS, malaria, and other diseases Target 7: Have halted by 2015 and begun to reverse the spread of HIV/AIDS Indicators: HIV prevalence among 15- to 24-year-old pregnant women Contraceptive prevalence rate Number of children orphaned by HIV/AIDS Target 8: Have halted by 2015 and begun to reverse the incidence of malaria and other major diseases Indicators: Prevalence and death rates associated with malaria Proportion of population in malaria-risk areas using effective malaria prevention and treatment mea- sures Prevalence and death rates associated with tuberculosis Proportion of tuberculosis cases detected and cured under DOTS Goal 7: Ensure environmental sustainability Goal 8: Build a global partnership for development Source: World Bank Group. http://www.developmentgoals.org/About_the_goals.htm. Accessed on March 12, 2004. *Examined in this study of "Appropriate health-related Millennium Development Goals for the Europe and Central Asia Region." contributes most to the burden of disease and premature mortality. An analysis of life expectancy in Russia has shown that life expectancy would improve only slightly if the International Develop- ment Targets for under-five and maternal mortality were reached. Slight reductions of adult mor- tality, on the other hand, could result in more pronounced improvements in life expectancy (Lock et al., 2002). The situation may be similar in many other transition countries of the ECA region. Although infant, child, and maternal mortality rates are generally many times higher than the European Union averages, in particular in Central Asia and the Caucasus, the impact of adult mortality on life expectancies in the region is greater. This study aims to contribute to the debate about the appropriateness of health-related MDGs for the countries of the ECA region. It is primarily addressed at policy advisors and senior analysts at the local, regional, and global level, who could influence the strategic directions of policy makers. The study will be of particular relevance to international development institutions and UN organizations, who might have to reconsider the focus of their health-related development assistance to the countries of the ECA region. The study examines how appropriate the health-related MDGs are for the countries of the ECA region by analysing the impact of the following four scenarios on life expectancy at birth: 1) achieving MDG targets for infant, child, and maternal mortality; 2) reducing infant, child, and maternal mortality to EU levels; MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 5 3) reducing infant, child, and maternal mortality to the lowest sub-regional levels; 4) reducing deaths from cardiovascular disease and external causes of death to EU levels. Because national registration data on infant and child mortality in countries of Central Asia and the Caucasus underestimate the true figure, the first three scenarios were recalculated for these countries in an additional calculation, using survey-based World Bank data. The following chapter 2 summarizes the findings of a literature review on the appropriateness of the health-related MDGs for the countries of the Europe and Central Asia region. Chapter 3 discusses the quality of the data used. Chapter 4 describes the methods used in the analytical part of the study. Chapter 5 details and compares the results of this analysis. Chapter 6 discusses the importance and the impact of HIV/AIDS on mortality patterns for the future and looks at how to move forward with more appropriate targets within the health-related MDGs for the countries for the ECA region. The concluding chapter 7 makes recommendations for policies, programs, and development assistance. The Appendix details the impact of the four scenarios on another measure of population health, Years of Potential Life Lost (YPLL). CHAPTER 2 LITERATURE REVIEW A part form the work of Lock et al. on Russia (2002), surprisingly little has been published on the appropriateness of global development targets for health in the ECA region. A search using PubMed/Medline in June 2003 could not identify any other publication on the subject. The analysis by Lock et al. illustrated the limited relevance of some of the health-related Inter- national Development Targets to Russia. The analysis covered the population of the Russian Fed- eration in 1995­1999 and used life expectancy as the measure of population health. A reduction of infant and child mortality by two-thirds, and maternal mortality by three-quarters, was found to result in an increase of 0.96 years in life expectancy. In contrast, a 20 percent reduction in adult mortality resulted in an increase of 1.86 years in male life expectancy. A reduction of adult mortal- ity to UK levels for cardiovascular diseases and external causes of death resulted in an increase in life expectancy at birth of 5.97 years for both sexes (Lock et al., 2002). In many parts of the ECA region, adult mortality rates increased markedly during the early 1990s. Most of this excess mortality affected males before retirement age. The main causes of the increase in premature deaths were circulatory diseases and external causes of deaths, including injuries, homicides, and suicide. The reasons were complex but the stress of the social costs of tran- sition and an increased consumption of alcohol, tobacco, and drugs were major contributing fac- tors to increased mortality rates (UNICEF, 2001). Deaths due to cardiovascular diseases exceed the EU average in all countries of the ECA region, reaching levels of 807/100,000 in the Russian Federation in 2001, as compared to an EU level of 258/100,000 in 1999 (WHO, 2003). The rapid spread of HIV/AIDS and the re-emergence of tuberculosis are two of the most seri- ous threats to health in the ECA region, particularly in the countries of the former Soviet Union. Tuberculosis re-emerged in the ECA region during the 1990s after 40 years of steady decline. In the Russian Federation, for example, the incidence of tuberculosis increased from 34 per 100,000 in 1991 to 92 per 100,000 in 2001 (WHO, 2003). The spread of HIV/AIDS is no less worrying. By the end of 1995, only 8,616 HIV infections had been officially recorded in the whole of Central and Eastern Europe. By June 2001, this number had increased to over 200,000 (Kelly and Amirkhanian, 2003). Most of these cases were recorded in Ukraine and Russia, where the number 7 8 WORLD BANK WORKING PAPER of HIV infections has grown exponentially over the last years. These two countries have currently the highest growth rates of HIV infections worldwide (Ruehl, Pokrovsky, and Vinogradov, 2002). Reported cases, however, are believed to be considerable underestimates. In Russia, the actual number of persons living with HIV was estimated to be 700,000 by the end of December 2002, while in Ukraine this number was 250,000 (Kelly and Amirkhanian, 2003), corresponding to about 1 percent of the population aged 15­49 in these two countries (UNAIDS 2002a and 2002b). The HIV epidemic serves as a strong reminder of the relevance of Goal 6 of the MDGs (com- bating HIV/AIDS, malaria, and other diseases) to the countries of the ECA region. However, achievement of Goal 6 has not been included in this analysis, because the impact on mortality and thus life expectancy at birth is still relatively low, when compared with non-communicable diseases and external causes of death. In Russia, the estimated number of deaths from HIV in 2001 was 9,000 (UNAIDS, 2002b), while in Ukraine in the same year an estimated 11,000 people died from HIV (UNAIDS, 2002a). Given the large and increasing number of HIV infections and the very limited availability of antiretroviral treatment, it is clear that the unchecked HIV epidemic in the ECA region, in particular in Russia and Ukraine, will likely have a devastating impact on mor- tality patterns in the near future, especially when combined with the growth in tuberculosis. In spite of the relevance of Goal 6 to the ECA region, the MDGs in general focus primarily on the problems of developing countries whose health policies have long focused on child and maternal health rather than adult health. There are two main reasons for this. One is that levels of childhood and maternal mortality have been, and continue to be, very high. The second is that these levels are visible. Data on infant, maternal, and to a lesser extent childhood mortality are collected in many countries where almost nothing is documented about adult mortality, either through routine collec- tion systems or surveys, in particular the series of Demographic and Health Surveys. Thus, the 2000 World Health Report was forced to use childhood mortality to assess equity of outcome in its measure of health care system performance, because these data were much more widely available than data on adult mortality (WHO, 2000). This point has implications that go beyond the ECA region: in those few settings in the developing world where adult mortality has been studied it has also been found to be very high, with some, at least, of the causes increasingly resembling those seen in the former Soviet Union, in particular cardiovascular diseases and injuries (Moshiro et al., 2001; Walker et al., 2000). A more critical problem with the health-related MDGs is that comparisons between countries are problematic because of significantly different starting points. Where infant and maternal mortal- ity rates are very high, basic improvements in social services and primary health care can considerably reduce the numbers of deaths. As the mortality rates decline, however, more complex clinical inter- ventions become necessary to further reduce infant and maternal mortality rates. Where these rates are already very low, it would be unrealistic to expect major reductions to be achieved so rapidly. This also has implications for those seeking to apply the principles set out by the WHO Commission on Macroeconomics and Health. This envisages the potential for rapid progress to be made through scaling up basic interventions, citing examples such as the supply of certain pharmaceuticals or insec- ticide treated bed nets. In the ECA region, however, while more needs to be done to scale up such basic interventions, much progress has already been made. For many health problems the challenge is to put in place complex packages that will move the health system on to the next step. A final problem with measuring progress on achieving the health-related MDGs is that they are formulated in terms of country averages. Countries as a whole can progress well towards the Goals, but leave certain groups or areas behind when policies are not directly targeted at them. In some countries, the health-related Goals can be achieved more easily by improving the circumstances of those groups in the population that are already better off. This is well illustrated by the experience in developed countries, such as the United Kingdom, where improvements in overall health were accompanied by either stagnation or actual deterioration for the worst off, while gains were con- centrated among those who already had the best health (Acheson, 1998). The danger is that the most disadvantaged and poor, such as women, rural populations or minorities such as the Roma are left behind (UNDP, 2003). CHAPTER 3 DATA QUALITY T here are major variations in the completeness, validity and reliability of health data across countries in the Europe and Central Asia Region. A major problem in some countries is the absence of accurate population data. Calculation of health-related rates depends on accurate data on population size. In some countries there is consid- erable uncertainty about population size and composition. Since 1989, wars and conflicts, com- bined with economic and social turmoil and the opening of borders, have triggered large-scale population movements that have often remained poorly quantified. Even within the most recent data as reported by the countries to WHO, significant differ- ences in population estimates can be found. When comparing the latest WHO population data with the mid-year population given for the same year in the WHO HFA database, differences in estimates of current population size amounted to about 670,000 for the Russian Federation, 630,000 for Moldova, 495,000 for Georgia, and 195,000 for Latvia. This emphasizes that popu- lation denominators in many countries of the region have to be treated with caution. Because most health indicators are based on population data, this also has implications for the accuracy of these indicators. Another problem is that health information systems deteriorated or, in some cases, collapsed. Deaths resulting from conflicts were often not recorded. In some countries of Central Asia and the Caucasus, fees were introduced for vital registration such as death or birth, compromising the qual- ity of official statistics (Badurashvili et al., 2001; UNICEF 2003; McKee et al., 2002). A recalcula- tion of life expectancy in Georgia concluded that official data may overestimate life expectancy at birth by as much as eight years. Even where health information systems are in place, there is often limited capacity to under- take meaningful analysis of health data. A study in Croatia showed that official mortality data were unreliable for the 1990s. One of the main problems was that the population denominator used before 1996 included not only people living in Croatia, but also Croatian citizens living perma- nently abroad (Bozicevic 2001). 9 10 WORLD BANK WORKING PAPER In particular, serious concerns have been raised about the quality of data on infant and child mortality, in particular in Central Asia and the Caucasus. For child mortality, this can be illustrated by the case of Tajikistan. The reported level of 1.16 deaths for ages 1­4 in 1999, as compared to an EU level of 1.02 per 1,000 live births of the same year and a level in Tajikistan of 20.56 in 1990, is almost certainly a significant underestimate. It is apparent that, given the sequelae of civil war in Tajikistan and the breakdown in many governmen- tal functions, including presumably vital registration, the recent data on child mortality must be treated with considerable scepticism. It is well known that in the countries of Central Asia and the Caucasus, official infant mortality rates, as recorded by national registration systems, are considerable underestimates. Three main factors that contribute to this discrepancy have been identified (UNICEF 2001; UNICEF 2003; McKee et al., 2002; Government of Tajikistan 2003; Aleshina and Redmond, 2003). The first fac- tor is that in many countries of Central Asia and the Caucasus, the Soviet definition of live birth is still in use, reducing the infant mortality rate by 22­25 percent (Aleshina and Redmond, 2003). Armenia, Georgia, and Kyrgyzstan have now adopted the WHO definition of live birth, but in practice implementation has been only partial. In the remaining countries of Central Asia and the Caucasus, the Soviet definition is still in use (Aleshina and Redmond, 2003). The second factor contributing to official underestimates is the misreporting of births and infant deaths by medical staff. The two main problems are that live births are reported as stillbirths or miscarriages and that post-neonatal deaths are reported as child deaths, both resulting in an underestimation of the num- ber of infant deaths. The third factor contributing to the discrepancy between official and survey data is that births and deaths of children are not reported by parents to the authorities, due to fees for registration. In Kazakhstan, Kyrgyzstan, and Tajikistan, fees for registration have been abol- ished in the last two years, but in Georgia there are still fees for the registration of births. Where mothers are not in a registered marriage, the charge is doubled, resulting in an increased risk of under-registration of children of single mothers (Aleshina and Redmond, 2003). In view of these problems with the quality of data, UN organizations such as UNICEF and the World Bank have produced alternative data sets on infant and child mortality, using models that include data from surveys and other sources. The health-related MDGs, established on the basis of estimates of 1990 levels, and measurement of progress towards these targets by World Bank and UNDP are based on World Bank estimates of infant and child mortality and not on national regis- tration data. Table 1 shows infant mortality rates according to national registration data, recent surveys, and World Bank estimates. TABLE 1. DIFFERENT ESTIMATES OF INFANT MORTALITY IN CENTRAL ASIA AND CAUCASUS Infant mortality Infant mortality Official infant mortality according to survey data estimated by (per 1,000 live births), (per 1,000 live births), World Bank Country Source: [6] Source: [16] (per 1,000 live births) Armenia 15.4 (1996­2000) 36 (1996­2000) 37.1 (1996­2000) Azerbaijan 17.2 (1996­2000) 74 (1996­2000) 78.9 (1996­2000) Georgia 15.2 (1995­1999) 43 (1995­1999) 24 (1995­1999) Kazakhstan 24.8 (1994­1999) 62 (1994­1999) 60 (1994­1999) Kirgyzstan 29.1 (1993­1997) 61 (1993­1997) 61 (1993­1997) Tajikistan 24.1 (1996­2000) 89 (1996­2000) 92.9 (1996­2000) Turkmenistan 38.9 (1995­2000) 74 (1995­2000) 74 (1995­2000) Uzbekistan 30.2 (1992­1996) 49 (1992­1996) 54.4 (1992­1996) Subregional average 25.83 59.42 60.58 MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 11 As the surveys generally estimate infant mortality rates based on reproductive histories of the five years before the survey, they are compared with the average figures for the corresponding peri- ods where figures are given for individual years. In all countries, surveys and World Bank estimates for infant mortality show much higher values than the official data reported to WHO. In Azerbai- jan, these differences are most pronounced. Survey data are 4.3 times and World Bank estimates 4.6 times higher than figures from national registration data. The sources used to estimate infant and child mortality rates in the ECA region are listed in Table 2. Surveys used in models aim to be representative of national populations or of women of repro- ductive age. The Demographic and Health Surveys (DHS) and the Tajikistan Living Standards and TABLE 2. SOURCES OF CHILDHOOD MORTALITY ESTIMATES Eastern European EU candidate countries Bulgaria National registration data only Czech Republic National registration data only Estonia National registration data only Hungary National registration data only Latvia National registration data only Lithuania National registration data only Poland National registration data only Romania Reproductive Health Survey 1999 Slovakia National registration data only Slovenia National registration data only Southeastern Europe Albania Multiple Indicator Cluster Survey 2000 Bosnia and Herzegovina Multiple Indicator Cluster Survey 2000 Croatia National registration data only Macedonia FYR National registration data only Serbia and Montenegro National registration data only Turkey Census 70; Census 75; Census 80; Census 85; Census 90; Demographic and Health Survey 93; Demographic and Health Survey 98; National Demographic Survey 66; Population and Health Survey 83; Population and Health Survey 88; World Fertility Survey 78 Remaining countries of the former Soviet Union Belarus National registration data only Moldova Reproductive Health Survey 1997, Multiple Indicator Cluster Survey 2000 Russian Federation Reproductive Health Survey 1996 Ukraine Reproductive Health Survey 1999, Multiple Indicator Cluster Survey 2000 Countries of Central Asia and Caucasus Armenia Demographic and Health Survey 2000 Azerbaijan Multiple Indicator Cluster Survey 2000, Reproductive Health Survey 2001 Georgia Multiple Indicator Cluster Survey 2000, Reproductive Health Survey 2000 Kazakhstan Demographic and Health Survey 1995, Demographic and Health Survey 1999 Kyrgyzstan Demographic and Health Survey 1997 Tajikistan Multiple Indicator Cluster Survey 2000, Living Standard Measurement Survey 1999 Turkmenistan Demographic and Health Survey 2000 Uzbekistan Demographic and Health Survey 1996, Multiple Indicator Cluster Survey 2000 Note: not all of the surveys contained questions about infant and child mortality, and all use national registration data in addition to other sources noted 12 WORLD BANK WORKING PAPER Measurement Survey (LSMS) in 1999 used direct estimates of infant and child mortality, asking participants to report in detail about pregnancy outcomes. The Multiple Indicator Cluster Surveys (MICS) used indirect ("Brass") estimates, based on the reported age of the mother, total number of children, and number of children who have died. Although estimates based on survey data appear to be better reflections of real infant mortality rates in Central Asia and the Caucasus than national registration data, they cannot be considered to be entirely accurate. Surveys are associated with a number of errors, which can be distinguished as non-sampling and sampling errors. Non-sampling errors result from misunderstandings and misreporting by respondents and mis- takes by interviewers and researchers. The death of children is a difficult topic and mothers might be reluctant to answer survey questions or they might not report all deaths. This omission of dead children from birth histories is aggravated for female infant deaths, resulting in an underestimation of female infant mortality rates in some surveys. Other non-sampling errors resulting in an under- estimation of infant deaths in surveys include the selection bias towards surviving women whose children had better chances of survival. Sampling errors are a result of sample size and procedures. The width of confidence intervals shows the resulting uncertainty in survey estimates. As confidence intervals increase with low sam- ple sizes and the number of births in Central Asia and the Caucasus tends to be lower than in developing countries, this is a problem especially relevant for this subregion. The Demographic and Health Survey in Kazakhstan in 1999, for example, provides a confidence interval of 45.3 to 78.5 per 1,000 live births. All values within this confidence interval have the same statistical validity at the specified level of confidence. The central values of confidence intervals, which are usually given as the survey estimates, might therefore be considerable underestimates or overestimates (Aleshina and Redmond, 2003). Where indirect estimates of infant mortality are used, as in the Multiple Indicator Cluster Sur- veys, extra potential sources of error are introduced. The most widely used model, developed by Brass, rests on a number of assumptions that might not accurately reflect mortality and fertility trends in the ECA region (Aleshina and Redmond, 2003). Going beyond the errors associated with surveys, different methods have been used to establish estimates of childhood mortality that make use of surveys and other data, and these methods are not always explicit. In effect, different UN organizations are presently using different datasets, and it is often unclear, or at least very difficult to discover, how exactly they were established. This illustrates the role that interpretation and subjective judgments play in establishing estimates. Figure 1, using the example of Tajikistan, shows how UNICEF and World Bank data on infant mortality rates differ. Both UNICEF and the World Bank use an updated version of the methodology developed by Hill for the 1993 World Development Report. In this methodology, a series of data points are sub- jected to a regression analysis, based on the approach of weighted least squares. This model assumes a linear rate of infant and child mortality changes, allowed to change at "knot" points, depending on the number of independent observations available. A knot, defining a particular scope, is defined for every five years of vital registration or one Demographic and Health Survey or other similar sur- vey. Although this methodology aims to provide a transparent means of estimation based on regres- sion analysis, subjective judgements are involved at every step of the analysis (World Bank, 2003). Another factor increasing the uncertainty about real infant and child mortality rates in Central Asia and the Caucasus is the scarcity of available surveys. As was noted in Table 2, coverage of this region by DHS surveys, for example, is limited. There has been no DHS conducted in Azerbaijan, Georgia, and Tajikistan, and only one in Armenia (2000), Kyrgyzstan (1997), Turkmenistan (2000), and Uzbekistan (1996), and two in Kazakhstan (1995, 1999). A more fundamental constraint on the use of World Bank data is that data on some indicators, such as maternal mortality rates in 1990, are entirely missing and data on other indicators, such as infant mortality, are incomplete. Furthermore, given the almost complete absence of surveys in this region in the first half of the 1990s, those data for 1990 that are included are derived ultimately MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 13 FIGURE 1. INFANT MORTALITY RATES IN TAJIKISTAN ACCORDING TO WHO, WB, AND UNICEF 100 80 60 40 20 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 National registration data World Bank data UNICEF data Sources: National Registration data: June 2003 version of the WHO HFA database; World Bank data: World Development Indicators 2003; UNICEF: UNICEF End of Decade Database (http://www.childinfo.org) BOX 2. DATA QUALITY IN CENTRAL ASIA AND THE CAUCASUS Validity The problems of data quality in the ECA region are greatest in the eight countries of Central Asia and the Caucasus, where official infant and child mortality rates, as reported to WHO, are considered to be substantial underestimates. Surveys in these countries have revealed much higher rates of childhood mortality. Conse- quently, estimates of infant and child mortality differ substantially, with implications for the 1990 baselines used to establish the health targets for the Millennium Development Goals. Although surveys appear to provide a more accurate picture, they are associated with a number of errors that limit their capacity to gauge current levels of infant and child mortality. When calculating datasets based on survey data, additional subjective judg- ments are involved. In the absence of complete vital registration, the exact current levels of childhood mortality in the region remain unknown. Credibility and ownership There is currently no consensus on infant and child mortality rates in Central Asia and the Caucasus. Estimates dif- fer not only between national governments and international organizations, but between UN organizations them- selves. Although international organizations agree that survey data are more accurate than national registration data, it is often not transparent how exactly their estimates of current rates were derived. On a national level, vital registration data remain in general the official baseline against which countries are tracking their progress. Availability of data Despite the major problems associated with national registration data in this subregion, they are at present the only complete source of data on infant, child, and maternal mortality, and the only source of data on adult mortality. 14 WORLD BANK WORKING PAPER from surveys undertaken 7 to 10 years later, so that the degree of uncertainty is considerable. Thus, in practice, if the intention is to track changes, WHO, recording national registration data, is the only complete source of the data required. However, because the quality of data in many countries is either poor or doubtful, the data have limited usefulness for serious policy decisions. Similar observations and concerns were made by Miller and Ryskulova (2004), who examined the com- pleteness and quality of official data in a study of the epidemiologic surveillance systems in the region. They noted many critical gaps, and concluded that it is important for the achievement of the MDGs to have supplementary data that provides information for better understanding of the determinants of birth outcomes and interventions to improve maternal and child health. Due to the uncertainties associated with surveys and survey-based estimates, surveys are only partial solutions to the problem of incomplete vital registration. As long as national registration is incomplete, the exact levels of childhood mortality in Central Asia and the Caucasus will remain unknown. In this context, the analysis of the impact of the health-related Millennium Develop- ment Goals on these countries in the current study was confronted with the choice of whether to use national registration data or survey-based estimates. It was decided to use both datasets in two parallel calculations. National registration data were used for all scenarios considered in this study. Where, in addition, survey-based estimates by the World Bank were available for the countries of Central Asia and the Caucasus, as was the case for infant and child mortality, an additional analysis based on these estimates was undertaken. Although neither of these two approaches may be entirely accurate, the aim was to make the best use of currently available data. CHAPTER 4 METHODS T his analysis is primarily based on population and mortality data by age, sex, and cause of death, as reported by national statistical offices to WHO in June 2003. The latest updates by WHO at this time were from February 10, 2003 for the population data and from December 31, 2002 for the mortality data. Using these data, life tables were calculated based on standard methods (Chiang 1968), the same as used in the study on the Russian Federation by Lock et al. (2002). The analysis covered all countries of the Europe and Central Asia region, with the exceptions of Bosnia and Herzegovina, Serbia and Montenegro, and Turkey, for which no recent population or mortality data were available. Using the official data reported by the countries to WHO, life expectancies were tabulated. The effects on life expectancy at birth were then calculated for the following four scenarios: 1) achieving MDG targets for infant, child, and maternal mortality; 2) reducing infant, child, and maternal mortality to EU levels; 3) reducing infant, child, and maternal mortality to the lowest subregional levels; 4) keeping infant, child, and maternal mortality constant, and reducing deaths from cardiovascular disease and external causes of death to EU levels. The regional and subregional averages were population-weighted. In countries that have already achieved or overachieved the target levels of the scenarios, the lower of the two figures was taken for the calculation. By this, no country was assumed to backslide behind their achievements and to make negative gains in life expectancy. Zero values in the subsequent tables on gains in life expectancy thus indicate that the target levels were achieved or overachieved. In the calculation of maternal mortality, current numbers of maternal deaths in the age groups 15­49 were reduced by the ratio between current maternal mortality rates and target rates. As has been noted previously, achievement of MDG Goal 6 (combating HIV/AIDS, malaria, and tuberculosis) has not been included in the analysis, because the impact of these diseases on life expectancy at birth is still relatively small. However, it is important to recognize that this may change dramatically in the near future. 15 16 WORLD BANK WORKING PAPER Scenario 1 explored the impact on life expectancy at birth of reaching the Millennium Devel- opment Goals for infant, child, and maternal mortality. The MDGs consider progress from a desig- nated baseline, the year 1990. Infant mortality rates, child mortality rates, and maternal mortality ratios for 1990 were taken from the January 2003 version of the WHO HFA database to establish the MDG targets for 2015. The most recent available data for infant, child, and maternal mortality from the WHO HFA database were used to analyze the effects on life expectancy, if these indica- tors could be reduced to MDG levels. In Scenario 2, life tables were recalculated on the basis of achieving the most recent EU levels for infant, child, and maternal mortality, showing the effects on life expectancy. In Scenario 3, infant, child, and maternal mortality of each country were reduced to the lowest levels achieved within four subregional groupings: 1) Eastern European EU candidate countries; 2) Southeastern Europe (Albania, Croatia, Macedonia, FYR) 3) Central Asia and Caucasus; 4) The remaining countries of the former Soviet Union (Belarus, Moldova, Russia, Ukraine). Scenario 4 kept infant, child, and maternal mortality rates constant, and reduced the deaths from cardiovascular disease and external causes of deaths to the current EU average. Because of the known problems with data from Central Asia and the Caucasus, the analyzes for these countries were undertaken separately in the first three scenarios for infant and child mortality, using both national vital registration data as reported to WHO and World Bank estimates. In the latter case, live tables were adjusted with World Bank estimates of infant and child mortality rates. For Scenario 1, the targets for Goal 4 were established on the basis of WB estimates of infant and mortality rates for 1990. For Scenario 3, the lowest subregional rates of infant and child mortality were similarly established on the basis of World Bank estimates. CHAPTER 5 RESULTS Life Expectancies in the ECA Region Using national registration data supplied to WHO, life expectancies were tabulated for the coun- tries of the ECA region. In the case of Central Asia and the Caucasus, this analysis was comple- mented by an adjustment of life tables with World Bank estimates on infant and child mortality. Figures 2 and 3 show the results of these tabulations, grouping the countries into the following four subregions: 1) Eastern European EU candidate countries, 2) Central Asia and Caucasus, 3) remaining countries of the former Soviet Union (Belarus, Moldova, Russia, Ukraine), 4) Southeastern Europe (Albania, Croatia, Macedonia). Regional averages for Figure 2 were taken from the WHO HFA database. In all countries of the ECA region, life expectancy is lower than the EU average. Figure 3 shows the results of recal- culations of life tables according to different estimates of infant and child mortality. If World Bank data reflect actual infant and child mortality rates in Central Asia and the Caucasus, life expectan- cies would be considerably lower. The difference is most dramatic for Tajikistan, where life expectancy would be reduced by 13.4 years. Scenario 1: Achieving MDG Targets for Infant, Child, and Maternal Mortality In this scenario, life tables were recalculated on the basis of achieving the MDG targets for infant, child, and maternal mortality. As a first step, the MDG targets were established on the basis of 1990 levels, taken from the WHO HFA Database, and, in the case of Central Asia and the Cauca- sus, from World Bank estimates. Tables 3 and 4 show current levels for infant mortality, the levels as they were in 1990 and the resulting MDG targets. Tables 5 and 6 show the related levels for 1­4 mortality and Tables 7 and 8 show the levels for maternal mortality. 17 18 WORLD BANK WORKING PAPER FIGURE 2. LIFE EXPECTANCIES IN THE ECA REGION (EXCEPT CENTRAL ASIA AND CAUCASUS) NIS 2001 67.21 CEE 2001 73.29 EU 1999 78.37 Latvia 70.48 Estonia 70.85 Romania 71.2 Bulgaria 71.59 Hungary 72.45 Lithuania 72.81 Slovakia 73.34 Poland 73.83 Czech Republic 75.08 Slovenia 75.63 Russian Federation 65.35 Moldova 67.71 Ukraine 67.8 Belarus 68.9 Croatia 72.94 Macedonia 73.19 Albania 74.82 60 62 64 66 68 70 72 74 76 78 80 Years Note: Data refer to latest available years. MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 19 FIGURE 3. LIFE EXPECTANCIES IN CENTRAL ASIA AND THE CAUCASUS Armenia 74.2 75.6 Azerbaijan 66.3 71.5 Georgia 72.6 74.3 Kazakhstan 61.1 66 Kyrgyzstan 66.2 68.2 Tajikistan 58.5 71.9 Turkmenistan 64.0 66.2 Uzbekistan 64.9 68.7 55 60 65 70 75 80 Life expectancy (World Bank data) Life expectancy (national registration data) Note: Data refer to latest available years. Tables 8 and 9 show the changes in life expectancies after the recalculation of Scenario 1. The largest gains were generally achieved by reducing infant mortality, followed by gains in reducing child mortality. The reduction of maternal mortality resulted in only slight gains in life expectancy. The average gain in life expectancy in the ECA region was between 0.68 and 1.24 years. According to national registration data, on a subregional level, Central Asia and the Caucasus showed maximum gains (0.98 years), followed by the remaining countries of the former Soviet Union (0.75 years), the Eastern European EU candidate countries (0.39 years) and Southeastern Europe (0.26 years). According to World Bank estimates of infant and child mortality, gains in Central Asia and the Caucasus would be considerably higher, reaching 4.04 years for the subregion and almost 8 years for Tajikistan. Scenario 2: Reducing Infant, Child, and Maternal Mortality to EU Levels In this scenario, life tables were recalculated on the basis of achieving the most recent EU levels for infant, child, and maternal mortality. The used EU levels were: I Infant mortality: 4.89/1,000 live births (1999) I Child (1­4) mortality: 1.02/1,000 live births (1999) I Maternal mortality: 5.54/100,000 live births (2000) Gains in life expectancy are shown in Tables 10 and 11. The overall average gain in life expectancy in the ECA region would be between 0.85 and 1.57 years, thus slightly higher than 20 WORLD BANK WORKING PAPER TABLE 3. MILLENNIUM DEVELOPMENT GOALS FOR INFANT MORTALITY Infant mortality per 1,000 live births (Source: WHO 2003) MDG target 1990 most recent year (one third of 1990 level) Eastern European EU candidate countries Slovenia 8.27 4.25 2.76 Lithuania 10.22 7.92 3.41 Czech Republic 10.8 3.97 3.60 Slovakia 11.99 8.58 4.00 Estonia 12.37 8.79 4.12 Latvia 13.74 10.37 4.58 Bulgaria 14.77 14.4 4.92 Hungary 14.82 8.13 4.94 Poland 15.95 8.11 5.32 Romania 26.91 18.41 8.97 Southeastern Europe Serbia and Montenegro 13.25 Croatia 10.67 7.68 3.56 Bosnia and Herzegovina 14.85 15 4.94 Macedonia FYR 31.60 11.81 10.53 Albania 33.81 11.62 11.27 Turkey 58 40 19.33 Remaining countries of the former Soviet Union Belarus 12.08 9.15 4.03 Ukraine 12.97 11.96 4.32 Russian Federation 17.64 14.57 5.88 Moldova 19.23 16.38 6.41 EU 7.6 4.89 Note: Data are incomplete for Serbia and Montenegro. TABLE 4. MILLENNIUM DEVELOPMENT GOALS FOR INFANT MORTALITY (CENTRAL ASIA AND CAUCASUS) National registration data, Source: [6] World Bank data, Source: [2] Infant mortality per 1,000 live births Infant mortality per 1,000 live births MDG target MDG target most recent (one third of most recent (one third of 1990 year 1990 level) 1990 year 1990 level) Georgia 15.83 10.43 5.28 24 24 8 Armenia 18.34 15.75 6.11 50 33.5 16.7 Azerbaijan 22.91 12.52 7.64 84 74 28 Kazakhstan 26.69 20.72 8.90 42 81 14 Kyrgyzstan 30.19 22.68 10.06 68 53.7 22.7 Uzbekistan 34.28 18.38 11.43 53 53.8 17.7 Tajikistan 40.41 27.9 13.47 98 91 32.7 Turkmenistan 45.15 32.78 15.05 80 69 26.7 Note: Data are incomplete for Serbia and Montenegro. MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 21 TABLE 5. MILLENNIUM DEVELOPMENT GOALS FOR 1­4 MORTALITY 1­4 mortality per 1,000 live births (Source: WHO 2003) MDG target 1990 most recent year (one third of 1990 level) Eastern European EU candidate countries Czech Republic 1.62 1.01 0.54 Slovenia 1.63 0.44 0.54 Slovakia 1.95 1.54 0.65 Hungary 1.97 1.24 0.66 Poland 2.33 1.24 0.78 Lithuania 3.08 2.53 1.03 Bulgaria 3.5 2.78 1.17 Latvia 3.89 1.99 1.3 Estonia 4.4 2.19 1.47 Romania 7.4 3.27 2.47 Southeastern Europe Serbia and Montenegro 2.51 Macedonia, FYR 0.15 1.84 0.05 Bosnia and Herzegovina 1.62 0.54 Croatia 1.71 1.27 0.57 Albania 8.25 5.18 2.75 Remaining countries of the former Soviet Union Belarus 3.19 2.46 1.06 Ukraine 3.74 3.53 1.25 Russian Federation 3.86 3.82 1.29 Moldova 5.13 3.53 1.71 EU 1.55 1.02 Note: No data were available for Turkey. Data are incomplete for Bosnia and Herzegovina and Serbia and Montenegro. TABLE 6. MILLENNIUM DEVELOPMENT GOALS FOR 1­4 MORTALITY (CENTRAL ASIA AND CAUCASUS) National registration data, Source: [6] World Bank data, Source: [2] 1­4 mortality per 1,000 live births 1­4 mortality per 1,000 live births MDG target MDG target most recent (one third of most recent (one third of 1990 year 1990 level) 1990 year 1990 level) Georgia 4.08 1.52 1.36 5 5 1.7 Armenia 5.81 3.47 1.94 8 4 2.7 Kazakhstan 7.56 6.04 2.52 10 18 3.3 Kyrgyzstan 12.03 11.59 3.86 13 9 4.3 Uzbekistán 13.49 8.05 4.5 12 16 4 Azerbaijan 18.2 11.08 6.07 22 19 7.3 Turkmenistan 20.4 19.67 6.56 18 18 6 Tajikistan 21.76 1.16 7.25 29 25 9.7 22 WORLD BANK WORKING PAPER TABLE 7. MILLENNIUM DEVELOPMENT GOALS FOR MATERNAL MORTALITY Maternal mortality per 100,000 live births (Source: WHO 2003) MDG target 1990 most recent year (one fourth of 1990 level) Eastern European EU candidate countries Slovakia 6.25 15.64 1.56 Czech Republic 8.42 3.31 2.11 Slovenia 8.84 17.22 2.21 Poland 12.78 7.93 3.20 Hungary 20.69 5.15 5.17 Bulgaria 20.92 19.07 5.23 Lithuania 22.86 12.68 5.72 Latvia 23.74 24.69 5.94 Estonia 31.38 7.92 7.85 Romania 83.56 34.03 20.89 Southeastern Europe Bosnia and Herzegovina 10.46 2.62 Croatia 10.83 2.44 2.71 Serbia and Montenegro 10.97 5.56 2.74 Macedonia, FYR 11.48 14.81 2.87 Albania 37.75 22.69 9.44 Turkey 180 130 45 Central Asia and Caucasus Azerbaijan 9.29 25.37 2.32 Georgia 20.47 58.69 5.12 Uzbekistan 34.12 34.12 8.53 Armenia 40.06 21.83 10.02 Turkmenistan 42.28 44.03 10.57 Kazakhstan 54.77 46.74 13.69 Kyrgyzstan 62.88 46.12 15.72 Tajikistan 97.2 44.6 24.3 Remaining countries of the former Soviet Union Belarus 21.81 14.17 5.45 Ukraine 32.41 23.91 8.10 Moldova 44.11 43.9 11.03 Russian Federation 47.41 36.52 11.85 EU 7.83 5.54 Note: No recent data were available for Bosnia and Herzegovina. for reaching the Millennium Development Goals for infant, child, and maternal mortality. The subregion of Central Asia and the Caucasus would achieve maximum gains (between 1.58 and 5.60 years), followed by the remaining countries of the former Soviet Union (0.81 years), Southeastern Europe (0.47 years), and the Eastern European EU candidate countries (0.46 years). On the basis of World Bank estimates for Central Asia and the Caucasus, maximum gains would be achieved in Tajikistan (12.29 years), Kazakhstan (5.74 years), and Azerbaijan (5.71 years). MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 23 TABLE 8. SCENARIO 1: IMPACT ON LIFE EXPECTANCY OF REACHING MILLENNIUM DEVELOPMENT GOALS 4 AND 5 Change due to reducing Overall gain in Infant Maternal life expectancy mortality 1­4 mortality mortality Eastern European EU candidate countries Czech Republic 0.07 0.03 0.04 0.0006 Slovenia 0.12 0.12 0 0.0026 Poland 0.24 0.20 0.03 0.0013 Hungary 0.27 0.23 0.04 0 Estonia 0.39 0.33 0.06 0.0001 Slovakia 0.40 0.33 0.06 0.0006 Lithuania 0.45 0.34 0.11 0.0014 Latvia 0.49 0.44 0.05 0.0047 Romania 0.73 0.67 0.06 0.0034 Bulgaria 0.76 0.65 0.11 0.0033 Subregional average 0.39 0.34 0.05 0.0018 Southeastern Europe Albania 0.20 0.02 0.18 0.0046 Macedonia, FYR 0.24 0.11 0.13 0.0047 Croatia 0.31 0.27 0.04 0 Subregional average 0.26 0.15 0.10 0.0025 Remaining countries of the former Soviet Union Belarus 0.48 0.36 0.11 0.0010 Ukraine 0.67 0.52 0.15 0.0013 Russian Federation 0.79 0.62 0.17 0.0058 Moldova 0.90 0.76 0.14 0.0053 Subregional average 0.75 0.59 0.16 0.0045 Overall ECA average 0.68 0.52 0.16 0.0037 Overall ECA average (1) 1.24 0.89 0.35 Note: (1) based on World Bank data on infant and child mortality in Central Asia and Caucasus; data refer to latest available years; no recent data available for Bosnia and Herzegovina, Serbia and Montenegro, and Turkey. Scenario 3: Reducing Infant, Child and Maternal Mortality to the Lowest Subregional Levels The analyzes were repeated using the lowest subregional levels of infant, child, and maternal mor- tality as targets. For all countries, national registration data were used to establish lowest sub- regional levels of infant, child, and maternal mortality. In the additional analysis for the countries of Central Asia and the Caucasus, based on World Bank data on infant and child mortality, the subregional targets were established on the basis of World Bank estimates. The lowest levels in the four subregions are shown in Table 12. Tables 13 and 14 show the gains in life expectancy that would be achieved in this scenario. The average gain in the ECA region would be between 0.58 and 1.09 years. On a subregional level and on the basis of national registration data, the maximum gains would be reached in Central Asia and the Caucasus (1.18 years), followed by the Eastern European EU candidate countries (0.55 years), the remaining countries of the former Soviet Union (0.41 years) and Southeastern Europe (0.27 years). On the basis of World Bank data on infant and child mortality in Central Asia and the 24 W ORLD B ANK W ORKING TABLE 9. SCENARIO 1: IMPACT ON LIFE EXPECTANCY OF REACHING MILLENNIUM DEVELOPMENT GOALS 4 AND 5 P (CENTRAL ASIA AND CAUCASUS) APER National registration data World Bank data Change due to reducing Change due to reducing Overall gain in Infant Maternal Overall gain in Infant life expectancy mortality 1­4 mortality mortality life expectancy mortality 1­4 mortality Tajikistan 0.48 0.48 0 0.0003 7.97 1.57 6.4 Georgia 0.57 0.55 0.01 0.0026 1.92 1.68 0.24 Azerbaijan 0.73 0.36 0.37 0.0039 3.98 3.17 0.81 Armenia 0.86 0.74 0.11 0.0050 1.38 1.28 0.1 Uzbekistan 0.97 0.56 0.40 0.0011 3.72 2.41 1.31 Kazakhstan 1.02 0.78 0.23 0.0109 5.05 4.15 0.9 Kyrgyzstan 1.44 0.88 0.56 0.0055 2.41 2.1 0.31 Turkmenistan 2.05 1.11 0.93 0.0031 3.37 2.59 0.78 Subregional average 0.98 0.64 0.33 0.33 4.04 2.68 1.37 Note: Data refer to latest available years MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 25 TABLE 10. SCENARIO 2: IMPACT ON LIFE EXPECTANCY OF REDUCING INFANT, CHILD AND MATERNAL MORTALITY TO EU LEVELS Change due to reducing Overall gain in Infant Maternal life expectancy mortality 1­4 mortality mortality Eastern European EU candidate countries Czech Republic 0 0 0 0 Slovenia 0 0 0 0.0020 Poland 0.25 0.24 0.02 0.0007 Hungary 0.26 0.24 0.02 0 Slovakia 0.31 0.27 0.04 0.0004 Lithuania 0.34 0.23 0.11 0.0014 Estonia 0.37 0.28 0.09 0.0033 Latvia 0.48 0.41 0.07 0.0048 Bulgaria 0.77 0.65 0.12 0.0032 Romania 1.12 0.95 0.16 0.0074 Subregional average 0.46 0.40 0.06 0.0024 Southeastern Europe Croatia 0.19 0.18 0.01 0 Macedonia, FYR 0.65 0.59 0.06 0.0036 Albania 0.76 0.45 0.31 0.0059 Subregional average 0.47 0.36 0.12 0.0027 Remaining countries of the former Soviet Union Belarus 0.42 0.30 0.12 0.0010 Ukraine 0.65 0.48 0.17 0.0015 Russian Federation 0.88 0.69 0.18 0.0072 Moldova 1.07 0.87 0.19 0.0062 Subregional average 0.81 0.62 0.17 0.0055 Overall ECA average 0.85 0.63 0.22 0.0045 Overall ECA average (1) 1.57 1.12 0.45 Note: (1) based on World Bank data on infant and child mortality in Central Asia and Caucasus; data refer to latest available years; no recent data available for Bosnia and Herzegovina, Serbia and Montenegro and Turkey. Caucasus, gains in this subregion would be 4.02 years, while gains on the country level would be maximal in Tajikistan (10.55 years), Azerbaijan (4.48 years), and Kazakhstan (4.41 years). Scenario 4: Reducing Deaths from Cardiovascular Disease and External Causes of Death to EU Levels In this scenario, infant, child, and maternal mortality were kept constant and mortality rates from cardiovascular disease and external causes of death (injuries, violence, and poisoning) reduced to EU averages. The most recent EU mortality rates were 258/100,000 (1999) for cardiovascular disease and 40/100,000 (1999) for external causes of death. Table 15 shows the implications for life expectancy. The overall gain in life expectancy in the ECA region in this scenario would be 7.75 years. At a subregional level, maximum gains would be achieved in the remaining countries of the former Soviet Union (9.58 years), followed by Central Asia and the Caucasus (7.37 years), the Eastern European EU candidate countries (4.75 years), and Southeastern Europe (3.67 years). On a country level, gains in life expectancy were highest in the countries of the former Soviet Union. Maximum gains would be achieved in the Russian 26 W ORLD B ANK W ORKING TABLE 11. SCENARIO 2: IMPACT ON LIFE EXPECTANCY OF REDUCING INFANT, CHILD AND MATERNAL MORTALITY TO EU LEVELS P (CENTRAL ASIA AND CAUCASUS) APER National registration data World Bank data Change due to reducing Change due to reducing Overall gain in Infant Maternal Overall gain in Infant life expectancy mortality 1­4 mortality mortality life expectancy mortality 1­4 mortality Georgia 0.63 0.59 0.04 0.0026 2.29 2 0.29 Tajikistan 0.82 0.76 0.07 0.0005 12.29 2.31 9.98 Armenia 1.02 0.83 0.18 0.0068 2.38 2.16 0.22 Azerbaijan 1.31 0.56 0.74 0.0034 5.71 4.72 0.99 Kazakhstan 1.39 1.05 0.33 0.0136 5.74 4.7 1.04 Uzbekistan 1.89 1.09 0.80 0.0013 5.39 3.76 1.63 Kyrgyzstan 1.99 1.23 0.75 0.0074 3.82 3.29 0.53 Turkmenistan 3.05 1.74 1.30 0.0036 5 3.89 1.11 Subregional average 1.58 1.00 0.58 0.0050 5.60 3.72 1.88 Note: Data refer to latest available years. MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 27 TABLE 12. LOWEST SUBREGIONAL LEVELS OF INFANT, CHILD, AND MATERNAL MORTALITY Infant mortality Child (1­4) Maternal mortality (per 1,000 live mortality (per (per 100,000 live Subregion births) 1,000 live births) births) Central Asia and Caucasus 10.43 1.16 21.83 (national registration data) (Georgia 2001) (Tajikistan 1999) (Armenia 2001) Central Asia and Caucasus 24 4 (World Bank data) (Georgia 2001) (Armenia 2001) Eastern European EU 3.97 0.44 3.31 candidate countries (Czech Republic 2001) (Slovenia 2001) (Czech Republic 2001) Remaining countries of 9.15 2.46 14.17 the former Soviet Union (Belarus 2001) (Belarus 2001) (Belarus 2001) (Belarus, Moldova, Russia, Ukraine) Southeastern Europe 7.68 1.27 2.44 (Albania, Croatia, Macedonia) (Croatia 2001) (Croatia 2001) (Croatia 2001) TABLE 13. SCENARIO 3: IMPACT ON LIFE EXPECTANCY OF REDUCING INFANT, CHILD, AND MATERNAL MORTALITY TO THE LOWEST LEVELS ACHIEVED IN THE SUBREGION Change due to reducing Overall gain in Infant Maternal life expectancy mortality 1­4 mortality mortality Eastern European EU candidate countries Slovenia 0.02 0.02 0 0.0024 Czech Republic 0.05 0 0.04 0 Latvia 0.09 0.09 0 0.0026 Hungary 0.36 0.31 0.06 0 Poland 0.36 0.30 0.06 0.0013 Slovakia 0.42 0.34 0.08 0.0005 Lithuania 0.45 0.30 0.15 0.0019 Estonia 0.48 0.34 0.13 0.0063 Bulgaria 0.88 0.72 0.16 0.0038 Romania 1.22 1.02 0.20 0.0079 Subregional average 0.55 0.45 0.10 0.0028 Southeastern Europe Croatia 0 0 0 0 Macedonia, FYR 0.40 0.35 0.04 0.0049 Albania 0.55 0.26 0.29 0.0070 Subregional average 0.27 0.16 0.10 0.0033 Remaining countries of the former Soviet Union Belarus 0 0 0 0 Ukraine 0.26 0.19 0.07 0.0008 Russian Federation 0.48 0.38 0.09 0.0052 Moldova 0.64 0.55 0.08 0.0048 Subregional average 0.41 0.32 0.08 0.0039 Overall ECA average 0.58 0.40 0.17 0.0033 Overall ECA average (1) 1.09 0.73 0.36 Note: (1) based on World Bank data on infant and child mortality in Central Asia and Caucasus; data refer to latest available years; no recent data available for Bosnia and Herzegovina, Serbia and Montenegro and Turkey. 28 W ORLD B ANK W ORKING TABLE 14. SCENARIO 3: IMPACT ON LIFE EXPECTANCY OF REDUCING INFANT, CHILD, AND MATERNAL MORTALITY TO THE P APER LOWEST LEVELS ACHIEVED IN THE SUBREGION (CENTRAL ASIA AND CAUCASUS) National registration data World Bank data Change due to reducing Change due to reducing Overall gain in Infant Maternal Overall gain in Infant life expectancy mortality 1­4 mortality mortality life expectancy mortality 1­4 mortality Georgia 0.03 0 0.03 0 0.07 0 0.07 Armenia 0.58 0.41 0.17 0 0.72 0.72 0 Tajikistan 0.68 0.58 0.10 0.0003 10.55 1.8 8.75 Azerbaijan 0.89 0.15 0.73 0.0006 4.48 3.44 1.04 Kazakhstan 1.01 0.68 0.32 0.0082 4.41 3.55 0.86 Uzbekistan 1.43 0.65 0.78 0.0005 3.62 2.31 1.31 Kyrgyzstan 1.59 0.85 0.74 0.0044 2.35 2.02 0.33 Turkmenistan 2.69 1.40 1.29 0.0021 3.67 2.75 0.92 Subregional average 1.18 0.60 0.57 0.0024 4.02 2.44 1.58 Note: Data refer to latest available years. MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 29 TABLE 15. SCENARIO 4: IMPACT ON LIFE EXPECTANCY OF REDUCING ADULT MORTALITY FROM INJURIES AND VIOLENCE AND CARDIOVASCULAR DISEASE TO EU LEVELS Overall gain in Change due to reducing deaths from life expectancy Injuries and violence Cardiovascular disease Eastern European EU candidate countries Slovenia 1.36 0.73 0.63 Poland 3.30 0.56 2.74 Czech Republic 3.50 0.47 3.03 Hungary 4.29 0.73 3.56 Slovakia 4.65 0.39 4.26 Bulgaria 6.57 0.22 6.36 Estonia 6.72 2.28 4.44 Romania 6.85 0.51 6.34 Latvia 7.20 2.33 4.87 Lithuania 7.47 2.35 5.13 Subregional average 4.75 0.65 4.10 Southeastern Europe Albania 3.10 0.40 2.70 Croatia 3.54 0.40 3.14 Macedonia, FYR 4.84 0 4.84 Subregional average 3.67 0.31 3.36 Central Asia and Caucasus Armenia 4.36 0 4.36 Tajikistan 5.19 0 5.19 Kyrgyzstan 5.67 1.15 4.52 Georgia 6.39 0 6.39 Azerbaijan 6.47 0 6.47 Uzbekistan 7.98 0.23 7.75 Kazakhstan 8.63 1.83 6.81 Turkmenistan 9.70 0.34 9.36 Subregional average 7.37 0.56 6.81 Remaining countries of the former Soviet Union Moldova 7.36 1.20 6.16 Belarus 7.52 2.39 5.13 Ukraine 8.67 1.71 6.96 Russian Federation 10.09 3.40 6.70 Subregional average 9.58 2.91 6.68 Overall ECA average 7.75 1.82 5.93 Note: Data refer to latest available years; no recent data available for Bosnia and Herzegovina, Serbia and Montenegro and Turkey. Federation (10.09 years), Turkmenistan (9.70 years), and the Ukraine (8.67 years). Among the Eastern European EU candidate countries, Lithuania (7.47 years), and Latvia (7.20 years) stand out as the countries with the highest gains in life expectancy in this scenario. Most of these gains in life expectancy in the ECA region would be achieved from reducing mortality from cardiovascular disease (5.93 years), although in some countries gains from reducing adult mortality from external causes of death would be significant, as in the Russian Federation, reaching 3.40 years. 30 WORLD BANK WORKING PAPER Comparisons of Age-specific Deaths from Cardiovascular Disease and External Causes to Sweden In order to make the reductions of mortality in Scenario 4 more transparent, in a complementary analysis, age-specific mortality patterns of cardiovascular disease and external causes of death in selected countries of the ECA region were compared to levels in Sweden. From each subregion mortality patterns from one country were compared to mortality patterns in Sweden. As can be seen in Table 16, age-specific rates from cardiovascular disease are many times higher in the four countries of the ECA region than in Sweden. The difference is largest in younger age groups. The results of the comparisons for cardiovascular disease are graphically illustrated in the following four figures, which show the ratio of age-specific death rates in each country to that in Sweden. The comparisons for external causes of death are shown in Table 17. The differences of the four countries of the ECA region to Sweden are largest in the youngest age groups, illustrating the need for reducing childhood injuries in these countries. Figures 8 to 11 make the results of the comparisons more visible, showing the ratio of age-specific death rates from external causes of death in each country to that in Sweden. Overview of Scenarios The results of the four scenarios are summarized in the following tables and figures. Tables 18 and 19 and Figure 12 show the overall impact on life expectancy. In the whole of the ECA region, as well as in all four subregions and in most individual countries, the gain in life expectancy was by far greatest in Scenario 4. Reducing mortality levels from cardiovascular disease and external causes of death to EU levels would result in an average gain of 7.75 years. The reductions in infant, child, and maternal mortality in Scenarios 1­3, on the other hand, would result in lower gains in life expectancy. Scenario 1 (reaching the Millennium Development Goals for infant, child, and mater- nal mortality) resulted in an average ECA gain of between 0.68 and 1.24 years. Scenario 2 (reach- ing EU levels for infant, child, and maternal mortality) would result in an average gain of 0.85­1.57 years, while Scenario 3 (reaching the lowest subregional levels of infant, child, and maternal mortality) would result in an average gain of 0.58­1.09 years. TABLE 16. AGE-SPECIFIC DEATH RATES FROM CARDIOVASCULAR DISEASE PER 100,000 POPULATION IN SWEDEN AND SELECTED ECA COUNTRIES (LATEST AVAILABLE YEARS) Age group Sweden Russia Kazakhstan Macedonia Bulgaria 20­24 2 18 18 11 12 25­29 3 32 33 8 16 30­34 5 59 55 7 33 35­39 11 110 103 24 56 40­44 32 206 188 55 121 45­49 49 360 323 121 219 50­54 84 599 546 264 389 55­59 158 949 928 417 666 60­64 297 1476 1491 841 1105 65­69 552 2362 2417 1482 1830 70­74 1072 3694 3666 2898 3270 75­79 2060 5651 5684 5383 5823 80­84 3979 9083 8989 9388 10054 85+ 9332 16107 15380 13213 19169 MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 31 FIGURE 4. AGE-SPECIFIC CVD DEATH RATES IN RUSSIA AS COMPARED TO SWEDEN 12 10 8 6 4 2 0 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85+ Sweden Russia FIGURE 5. AGE-SPECIFIC CVD DEATH RATES IN KAZAKHSTAN AS COMPARED TO SWEDEN 12 10 8 6 4 2 0 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85+ Sweden Kazakhstan 32 WORLD BANK WORKING PAPER FIGURE 6. AGE-SPECIFIC CVD DEATH RATES IN BULGARIA AS COMPARED TO SWEDEN 7 6 5 4 3 2 1 0 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85+ Sweden Bulgaria FIGURE 7. AGE-SPECIFIC CVD DEATH RATES IN MACEDONIA, FYR AS COMPARED TO SWEDEN 6 5 4 3 2 1 0 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85+ Sweden Macedonia MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 33 TABLE 17. AGE-SPECIFIC DEATH RATES FROM EXTERNAL CAUSES OF DEATH PER 100,000 POPULATION IN SWEDEN AND SELECTED ECA COUNTRIES (LATEST AVAILABLE YEARS) Age group Sweden Lithuania Kazakhstan Croatia Russia <1 5 58 95 15 100 01­04 2 33 44 20 37 05­09 4 14 20 10 26 10­14 6 13 25 38 28 15­19 19 69 69 58 112 20­24 32 113 126 49 226 25­29 28 113 152 44 245 30­34 28 149 163 50 254 35­39 32 152 171 53 277 40­44 40 188 176 81 310 45­49 40 250 180 71 339 50­54 46 250 196 56 347 55­59 47 218 192 69 333 60­64 50 214 182 86 278 65­69 56 202 175 112 232 70­74 64 166 155 197 193 75­79 106 149 129 131 164 80­84 193 164 186 386 192 85+ 449 191 357 21 213 FIGURE 8. AGE-SPECIFIC DEATH RATES FROM EXTERNAL CAUSES OF DEATH IN RUSSIA AS COMPARED TO SWEDEN 25 20 15 10 5 0 <1 01- 05- 10- 15- 20- 25- 30- 35- 40- 45- 50- 55- 60- 65- 70- 75- 80- 85+ 04 09 14 19 24 29 34 39 44 49 54 59 64 69 74 79 84 Sweden Russia 34 WORLD BANK WORKING PAPER FIGURE 9. AGE-SPECIFIC DEATH RATES FROM EXTERNAL CAUSES OF DEATH IN KAZAKHSTAN AS COMPARED TO SWEDEN 25 20 15 10 5 0 <1 01- 05- 10- 15- 20- 25- 30- 35- 40- 45- 50- 55- 60- 65- 70- 75- 80- 85+ 04 09 14 19 24 29 34 39 44 49 54 59 64 69 74 79 84 Sweden Kazakhstan FIGURE 10. AGE-SPECIFIC DEATH RATES FROM EXTERNAL CAUSES OF DEATH IN LITHUANIA AS COMPARED TO SWEDEN 16 14 12 10 8 6 4 2 0 <1 01- 05- 10- 15- 20- 25- 30- 35- 40- 45- 50- 55- 60- 65- 70- 75- 80- 85+ 04 09 14 19 24 29 34 39 44 49 54 59 64 69 74 79 84 Sweden Lithuania MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 35 FIGURE 11. AGE-SPECIFIC DEATH RATES FROM EXTERNAL CAUSES OF DEATH IN CROATIA AS COMPARED TO SWEDEN 10 9 8 7 6 5 4 3 2 1 0 <1 01- 05- 10- 15- 20- 25- 30- 35- 40- 45- 50- 55- 60- 65- 70- 75- 80- 85+ 04 09 14 19 24 29 34 39 44 49 54 59 64 69 74 79 84 Sweden Crotia In Central Asia and the Caucasus, gains in the first three scenarios were considerably higher, when the calculations were based on World Bank estimates of infant and child mortality. In this case, average subregional gains would be 4.04 years in Scenario 1, 5.60 years in Scenario 2 and 4.02 years in Scenario 3. On a country level, Tajikistan would gain most (12.29 years) in Scenario 2. Countries of the former Soviet Union also stand out with maximum gains from Scenario 4 (based on national registration data). These countries are the Russian Federation (10.09 years), Turkmenistan (9.70 years), Ukraine (8.67 years), Kazakhstan (8.63 years), Lithuania (7.47 years), and Latvia (7.20 years). 36 WORLD BANK WORKING PAPER TABLE 18. OVERALL IMPACT ON LIFE EXPECTANCY ACCORDING TO THE DIFFERENT SCENARIOS Scenario 3: Scenario 4: Scenario 1: Scenario 2: Reaching best Reaching EU Reaching Reaching subregional levels for CVD and MDGs 4 and 5 EU levels levels external causes Eastern European EU candidate countries Bulgaria 0.76 0.77 0.88 6.57 Czech Republic 0.07 0 0.05 3.50 Estonia 0.39 0.37 0.48 6.72 Hungary 0.27 0.26 0.36 4.29 Latvia 0.49 0.48 0.09 7.20 Lithuania 0.45 0.34 0.45 7.47 Poland 0.24 0.25 0.36 3.30 Romania 0.73 1.12 1.22 6.85 Slovakia 0.40 0.31 0.42 4.65 Slovenia 0.12 0 0.02 1.36 Subregional average 0.39 0.46 0.55 4.75 Southeastern Europe Albania 0.20 0.76 0.55 3.10 Croatia 0.31 0.19 0 3.54 Macedonia, FYR 0.24 0.65 0.40 4.84 Subregional average 0.26 0.47 0.27 3.67 Remaining countries of the former Soviet Union Belarus 0.48 0.42 0 7.52 Moldova 0.90 1.07 0.64 7.36 Russian Federation 0.79 0.88 0.48 10.09 Ukraine 0.67 0.65 0.26 8.67 Subregional average 0.75 0.81 0.41 9.58 Overall ECA average 0.68 0.85 0.58 7.75 Overall ECA average (1) 1.24 1.57 1.09 Note: (1) based on World Bank data on infant and child mortality in Central Asia and Caucasus; data refer to latest available years; no recent data available for Bosnia and Herzegovina, Serbia and Montenegro and Turkey. M ILLENNIUM TABLE 19. OVERALL IMPACT ON LIFE EXPECTANCY ACCORDING TO THE DIFFERENT SCENARIOS (CENTRAL ASIA AND CAUCASUS) D EVELOPMENT National registration data World Bank data Scenario 4: Scenario 1: Scenario 2: Scenario 3: Reaching EU Scenario 1: Scenario 2: Scenario 3: Reaching Reaching Reaching best levels for CVD and Reaching Reaching Reaching best MDGs 4 and 5 EU levels subregional levels external causes MDGs 4 and 5 EU levels subregional levels G OALS Armenia 0.86 1.02 0.58 4.36 1.38 2.38 0.72 Azerbaijan 0.73 1.31 0.89 6.47 3.98 5.71 4.48 FOR Georgia 0.57 0.63 0.03 6.39 1.92 2.29 0.07 Kazakhstan 1.02 1.39 1.01 8.63 5.05 5.74 4.41 H EALTH Kyrgyzstan 1.44 1.99 1.59 5.67 2.41 3.82 2.35 Tajikistan 0.48 0.82 0.68 5.19 7.97 12.29 10.55 Turkmenistan 2.05 3.05 2.69 9.70 3.37 5 3.67 IN Uzbekistan 0.97 1.89 1.43 7.98 3.72 5.39 3.62 E UROPE Subregional average 0.98 1.58 1.18 7.37 4.04 5.60 4.02 Note: Data refer to latest available years. AND C ENTRAL A SIA 37 38 WORLD BANK WORKING PAPER FIGURE 12. OVERALL IMPACT ON LIFE EXPECTANCY ACCORDING TO THE DIFFERENT SCENARIOS: REGIONAL AND SUBREGIONAL AVERAGES 0.39 Eastern European EU 0.46 candidate countries 0.55 4.75 0.26 South Eastern Europe 0.47 0.27 3.67 0.98 Central Asia & Caucasus 1.58 1.18 7.37 4.04 Central Asia & Caucasus 5.6 (1) 4.02 7.37 0.75 Remaining FSU countries 0.81 0.41 9.58 0.68 Overall regional average 0.85 0.58 7.75 1.24 Overall regional average 1.57 (1) 1.09 7.75 0 1 2 3 4 5 6 7 8 9 10 Scenario 1: Reaching MDGs 4 and 5 Scenario 2: Reaching EU levels Scenario 3: Reaching best sub-regional levels Scenario 4: Reaching EU levels for CVD and external causes Note: (1) based on World Bank data on infant and child mortality in Central Asia and Caucasus. CHAPTER 6 DISCUSSION M aximal gains in life expectancy at birth in the ECA region were observed when re- ducing mortality levels from cardiovascular disease and external causes of death to EU levels (Scenario 4). In contrast, the impact of the first three scenarios on life expectancy at birth was much more limited, when using national registration data. Gains in life expectancy in the first three scenarios would be considerably higher for the countries of Central Asia and the Caucasus, if more survey-based data on infant and child mortality were used. How- ever, even in this case, adult mortality would still account for most premature mortality in the ECA region. HIV/AIDS As has already been noted, the impact of HIV/AIDS on mortality patterns, although still relatively small, may increase dramatically in the near future. In addition to the number of HIV positive indi- viduals who will die if they do not receive antiretroviral treatment, the number of infections is still growing. The HIV epidemic in the most affected countries, Russia and Ukraine, shows no signs of declining and there is considerable potential for a continued rapid increase of the number of HIV positive individuals in the future. A recent World Bank study on the economic impact of HIV in Russia modelled the future development of HIV infections, using cautious projections of the key parameters (Ruehl, Pokrovksy, and Vinogradov, 2002). The model assumed that no prevention or retroviral treatment would take place, using an "optimistic" and a "pessimistic" scenario. In the "optimistic" case, the cumulative number of HIV infected individuals would increase from 1.2 million in 2005 to 2.3 million in 2010 and 5.4 million in 2020. Mortality rates would increase from 500 per month in 2005 to 21,000 per month in 2020 (Ruehl, Pokrovksy, and Vinogradov, 2002). Obviously, the predicted rise in the number of HIV infected individuals and mortality rates will have a devastating impact on life expectancy. According to the scenario of Ruehl et al. on Russia and based on current WHO mortality data, the number of deaths from HIV/AIDS per year would increase 1,212 times between 2001 and 2020. On the basis of the 2001 life table, this increase in 39 40 WORLD BANK WORKING PAPER mortality would result in a decrease in the already low life expectancy at birth by 3.19 years, reaching 62.16 years. This striking scenario illustrates the importance of Millennium Development Goal 6 (combat- ing HIV/AIDS, malaria, and other diseases) to the countries of the ECA region. The case of HIV/AIDS, however, also points to some problems with the HIV related indicators used to measure progress towards this Millennium Development Goal. These indicators are the HIV prevalence among 15- to 24-year-old pregnant women, the contraceptive prevalence rate, and the number of children orphaned by HIV/AIDS. Once again, it is apparent that the choice of indicator is driven by the availability of data, in particular the wide- spread availability of seroprevalence surveys among women attending prenatal clinics in many devel- oping countries. However, while this not only provides no information about rates of infection among men in the population, it is also increasingly clear that it provides a biased measure of in- fection in the female population in the area concerned (Zaba, Boerma, and White, 2000; Crampin et al., 2003; Fylkesnes et al., 1998). Furthermore, unlike in many developing countries, sentinel sur- veillance at prenatal clinics remains uncommon in the ECA region and there is considerable uncer- tainty about the applicability of a method designed for countries where the predominant mode of transmission is heterosexual in others where it is still primarily through intravenous drug use. Other possible approaches involve tracking numbers of newly diagnosed cases of AIDS but this is also highly problematic as the pattern of service delivery will, to a considerable extent, determine who is diagnosed and at what stage of the disease. In particular, it provides no information on whether cases are representative of the general population, of high risk groups, or of a bridge pop- ulation linking the two, or any combination thereof. A change in the pattern of service delivery might affect reported rates without having any impact on the underlying dynamics of the disease. Instead, a more useful set of measures would involve tracking HIV prevalence among specified subgroups of high-risk core transmitters, bridge populations, and the general population, but obtaining such data would require substantial investment in epidemiological resources in this region, where they are at present extremely scarce. Moving Forward A policy aimed at reducing mortality levels from cardiovascular disease and external causes of death to EU levels emerges as having the greatest potential to impact on life expectancy at birth, outweighing potential gains that might be achieved when reducing infant and child mortality. In summary, it can therefore be concluded that the study made the case for complementing the Millennium Develop- ment Goals in the ECA region with another set of goals or indicators that take into account adult mortality, while at the same time showing the benefits of reducing infant and child mortality in Central Asia and the Caucasus. Millennium Development Goal 6 is to reverse the spread of HIV/AIDS and to reverse the incidence of malaria and "other major diseases." However, the only other disease specified in the related indicators in addition to HIV/AIDS and malaria is tuberculosis. Nevertheless, the use of the term "other major diseases" in the Goal makes it possible to include other major diseases that are more relevant in a specific country or regional context. This might leave countries room within the framework of the Millennium Development Goals to identify and tackle the diseases that cause the most damage to their populations and this offers an opportunity for dialogue between govern- ments and international agencies to operationalise appropriate targets. Perhaps as importantly, this study demonstrated the great variety among the countries of the ECA region. As the potential gains in life expectancy indicate, levels of infant, child, and maternal mortality, as well as of mortality from cardiovascular diseases and external causes, vary widely among the different countries of the region. At one end of the extreme is Slovenia, which would achieve almost no gain in life expectancy in the first three scenarios and only 1.36 years in Scenario 4. At the other end are Tajikistan, gaining 12.29 years in Scenario 2, and Turkmenistan, gaining 9.70 years in Scenario 4. Obviously, these countries need to adopt different strategies to improve the health of MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 41 their populations. This illustrates that appropriate policy interventions need to be developed at a country level and should not slavishly follow global development goals. It would clearly be mistaken to establish common goals for all countries of the ECA region. A similar lesson can be drawn for the level of external assistance in the health sector that the countries of the ECA region require. The level of development assistance in the health sector needs to be informed by an understanding of the burden of disease and premature mortality. It seems reasonable to argue that countries with high levels of under-5 and adult mortality, and conse- quently with larger potential for improvement, should be allocated substantially more resources than countries with already low levels of mortality. However, such advocacy must include consider- ations of the costs and effectiveness of various interventions for controlling the causes of under-5 and adult mortality. The incremental resource requirements and economic benefits of various interventions for controlling the causes of under-5 and adult mortality in ECA Region are beyond the scope of this study, but they need to be examined in each subregional and/or country context. Going beyond the ECA region, the study allows two more general conclusions. The first con- cerns adult mortality. The importance of non-communicable diseases and injuries is more obvious in the countries of the ECA region because vital registration systems, while imperfect, are much bet- ter than in other countries at comparable levels of development, but their importance is not con- fined to this region. It is often forgotten that non-communicable diseases and injuries account for more than half of all lost years of healthy life in developing countries (UNDP, 2003). Mortality from these causes of death could be used as additional indicators of progress towards the Millen- nium Development Goals, where data are available. This view has now been endorsed by senior WHO experts, calling for the inclusion of adult mortality and morbidity in the Millennium Development Goals (Kowal and Lopez, 2003). The second conclusion concerns life expectancy. Life expectancy is not only the most general summary measure of mortality in a population, but also a measure of social development. Living a long and healthy life is a basic condition of human and social development. An increase in life expectancy should therefore be a global development goal in itself. The United Nations Develop- ment Program (UNDP) recognizes the importance of life expectancy as a measure of social devel- opment by including it as one of three indicators in the calculation of the Human Development Index (HDI). Improvements in life expectancy have been previously used as UN development goals. A United Nations General Assembly resolution in 1980 called for raising life expectancy in all countries to 60 years by 2000. In 124 of the 173 countries that fell below that level this goal has been achieved (UNDP, 2003). CHAPTER 7 RECOMMENDATIONS A s was noted previously, this study has been limited to the analysis of the potential impact of reaching Goals 4 and 5 in the health-related Millennium Development Goals. Although mortality rates from HIV/AIDS are still comparatively low, this is likely to change in the near future unless ECA countries implement effective interventions to control the epidemic (World Bank, 2003). Studies modeling the future impact of HIV in this region exist but they are limited by the lack of contextually valid data on, for example, exposure to risk in the countries involved. Countries and international agencies should continue their efforts to improve HIV surveillance and to develop models of the future impact of HIV and tuberculosis in the region, exploring various scenarios and policy responses, and linking this information to Goal 6 so as to contribute to the discussion of the appropriateness of the MDGs to the ECA region. A main theme of this study was the problem with existing data on which to judge progress towards identified targets in many countries. This requires action on several fronts. Governments and international agencies should place a high priority on strengthening systems for vital registra- tion, in particular in the countries of Central Asia and the Caucasus. It will be necessary to address the main factors that contribute to the discrepancies between official and survey data in these coun- tries. First, it is important that all countries adopt and implement fully the WHO definition of a live birth rather than the Soviet one. Although this process has now started, it is crucial to acceler- ate it. International agencies can play an important role in encouraging this shift and supporting it with training programs for relevant staff. Second, training for health care staff is also required to reduce misreporting of infant deaths. This will need to be accompanied by the abolition of incen- tives that lead staff to underreport infant mortality, such as the systems of control and sanctions that inhibit open discussion of adverse events. Instead, incentives are needed that increase the accu- racy of reporting. Third, to ensure the reporting of births and deaths by parents, registration fees should be abolished where they still exist, as in Georgia. There are also many examples of how reg- istration can be linked to access to benefits (financial and otherwise) to create positive incentives to improve vital registration. However it should also be recognized that some countries in Central Asia or the Caucasus may be reluctant to improve the accuracy of their data, as high official infant 43 44 WORLD BANK WORKING PAPER mortality rates would portray their countries in a negative light. The authors have been made aware of at least one country where this is the case. This will again be an area in which international agen- cies can play a crucial role. While national registration is still incomplete, it is possible to improve the quality of survey data. One way is by increasing the sample sizes that will consequently narrow the confidence inter- vals and provide more accurate estimates, as well as by increasing the number of surveys. This is most acute in the countries of Central Asia and the Caucasus, but also concerns other countries in the ECA region, where concerns about the quality of data have been raised, but where hardly any surveys have been undertaken so far. This study demonstrates that there is currently no consensus on the true scale of infant and child mortality rates in Central Asia and the Caucasus. Estimates differ not only between national governments and international organizations, but also differ among international organizations. This ultimately threatens the credibility of the estimates. It is important to have in the public domain the sources and methods used for derivation of estimates for key health indicators being used by international agencies, with a view to achieving consensus on the best estimates for each country. On a national level, vital registration data remain in general the official baseline against which countries are tracking their progress. It is therefore necessary to involve national governments in the dialogue on what should be considered to be the most accurate estimates, thus securing national ownership of estimates generated by international organizations and strengthening the case for improvements in national vital registration. As this study has shown, life expectancies in Central Asia and the Caucasus are likely to be dramatically lower than official data suggest. It is essential that the governments of these countries recognize this problem. This would be a first step of addressing the serious matter of improving the health of their populations. Improving the regis- tration of infant and child mortality is not a goal in itself but it will help to increase national recog- nition of the main problems these countries face and to direct policy responses, both by national governments and international agencies, to where they are needed most. Obviously, health targets and policy interventions need to be adapted to the individual country context. Perhaps the most important message of this study is that simply following global develop- ment targets will not be the best way to enhance the health of the populations in the ECA region. The reason for this is that the burden of disease and mortality varies widely between countries and generally differs markedly from the patterns observed in developing countries. The eight countries of Central Asia and the Caucasus exhibit very high under-5 and maternal mortality rates. It is those countries that would benefit most from achieving the MDGs. In those countries where infant, under-5, and maternal mortality remain high there is a continuing need to direct resources to policies that will reduce them. This is likely to include strengthening basic social services and primary health care although, given the need to reflect local context, it will also be important not simply to transfer uncritically policies from other parts of the world. Instead, inter- national agencies, bilateral donors, and global foundations should support primary research into the causes of adverse health outcomes among these populations. In addition to high rates of childhood mortality, some of the countries of Central Asia and the Caucasus are at the same time struggling with staggering rates of adult mortality. One of the main findings that emerged from this study is that the health-related MDGs need to be adapted to the ECA region, so that they include adult mortality. This is particularly true of the countries of the former Soviet Union, but also of other countries of the ECA region. A potential average gain of 7.75 years in life expectancy at birth, when reducing mortality from cardiovascular diseases and external causes of death to EU levels, simply cannot be ignored. There are various ways in which adult mortality could be included in the MDGs. One way of doing so would be to include additional health indicators under Target 8 of Goal 6. At the most general level, life expectancy at birth could be used as an additional indicator. More specifically, major causes of adult mortality could be used as indicators of development. Although there are MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 45 problems with the accuracy of mortality data in some countries of the region, these inaccuracies are unlikely to be any greater than those for traditional MDG indicators, and may even be less. Two additional indicators important for the ECA region are those used in this study, namely mortality rates from cardiovascular disease and external causes of death. However, it is not necessary to wait for mortality levels to change. Instead, it is possible to inter- vene to address the main risk factors for cardiovascular disease, tracking changes that will appear before mortality declines. Cardiovascular disease is associated with complex risk factors, including poverty, smoking, alcohol, and poor nutrition, and although the precise contribution of different factors in this region differs from that in western populations (for example, the very low levels of dietary anti-oxidants and the consequences of episodic heavy drinking), changes in the prevalence of key risk factors could be used as additional indicators under Target 8 of MDG 6. So far, however, available data on the prevalence of relevant risk factors is very limited. Again, the gathering of more accurate data would help to direct policy responses to where they are most needed. Appropriate policy interventions to reduce the high levels of adult mortality must include the strengthening of public health services, with more emphasis on health promotion and disease pre- vention. Concerted public health efforts are in particular required to counteract the tobacco epi- demic which is spreading in the ECA region. The spread of tuberculosis and HIV/AIDS are other strong reminders of the need for such improvements in public health programs, including health promotion activities. The high rates of external causes of death, in particular among infants and children, call for concerted efforts to prevent injuries, especially during childhood. International assistance to the health sector could play an important role in improving health in the region. In general, international assistance should be focused on the countries in the ECA region with the largest burden of disease and premature mortality. As this study has shown, this concerns not only infant, child, and maternal mortality, but also adult mortality. In this sense, the current study highlights where the need for development assistance in the health sector is greatest. A stronger emphasis on countries with the highest burden of mortality and disease could also be an incentive to improve vital registration systems. APPENDIX Impact on Years of Potential Life Lost In an additional calculation, for each of the four scenarios and based on national registration data, Years of Potential Life Lost (YPLL) were calculated for each of the specified causes of death. YPLL were calculated as the number of premature life years lost up to the age of 64 per 10,000 persons. This measure offers a basis for considering the costs of different causes of mortality, when com- bined with economic measures of human capital. The disadvantage of the measure of YPLL is that it will not consider competing causes of death, which may lead to an overestimation of the impact and thus cost of specific causes of death. Table 20 shows the results of Scenario 1 for Years of Potential Life Lost on the age group 0­64 per 10,000 persons. As in the case of life expectancy, YPLL gains were maximized when reducing infant mortality, followed by the gains in reducing child mortality. Gains in YPLL from reducing maternal mortality, on the other hand, were minimal. The average gain in YPLL taking ECA as a whole was 5,364 years. Maximum gains were achieved in Central Asia and the Caucasus (6,876), followed by the remaining countries of the for- mer Soviet Union (6,126), the Eastern European EU candidate countries (3,172), and Southeast- ern Europe (1,595). As in the case of life expectancy, maximum gains on a country level were achieved in the Central Asian countries Turkmenistan (13,096), Kyrgyzstan (9,611), and Kazakhstan (8,271). The results for Scenario 2 are shown in Table 21. The overall average gain for the ECA region would be 6,500 years. The subregion of Central Asia and the Caucasus would achieve maximum gains (10,824), followed by the remaining countries of the former Soviet Union (6,557), the East- ern European EU candidate countries (3,741), and Southeastern Europe (3,379). As in the case of life expectancy, on a country level, maximum gains would be achieved in the Central Asian coun- tries Turkmenistan (20,205), Kyrgyzstan (13,469), and Uzbekistan (12,165). Gains in YPLL for Scenario 3 are shown in Table 22. The average gain for the ECA region would be 4,241 YPLL (age 0­64) per 10,000. Maximum gains were achieved for Central Asia and the Caucasus (7,059), followed by the Eastern European EU candidate countries (4,272), the 47 48 WORLD BANK WORKING PAPER TABLE 20. IMPACT ON YPLL (AGE 0­64 PER 10,000) OF REACHING THE MILLENNIUM DEVELOPMENT GOALS 4 AND 5 Overall gain in Gain due to reducing YPLL after Infant Maternal reaching MDGs mortality 1­4 mortality mortality Eastern European EU Candidate Countries Czech Republic 331.40 251.20 79.28 0.92 Slovenia 1033.53 1029.63 0 3.90 Poland 1870.00 1794.59 73.33 2.07 Hungary 2185.03 2093.23 91.80 0 Slovakia 3101.72 2960.25 140.52 0.95 Estonia 3159.10 3032.38 126.55 0.17 Lithuania 3261.75 3012.24 247.20 2.30 Latvia 4123.32 4005.39 110.10 7.83 Bulgaria 6140.32 5893.32 241.59 5.42 Romania 6177.58 6044.02 128.04 5.52 Subregional average 3172.13 3058.91 110.37 2.85 Southeastern Europe Albania 593.82 202.01 384.84 6.98 Macedonia, FYR 1255.43 964.06 283.80 7.56 Croatia 2463.16 2376.95 86.21 0 Subregional average 1594.77 1364.95 225.93 3.89 Central Asia and Caucasus Azerbaijan 4079.49 3239.72 833.49 6.29 Tajikistan 4282.69 4282.32 0 0.37 Georgia 4836.24 4806.23 25.97 4.04 Uzbekistan 6253.33 5304.44 946.94 1.95 Armenia 6594.79 6344.47 242.83 7.48 Kazakhstan 8270.72 7687.57 562.81 20.34 Kyrgyzstan 9610.91 8288.96 1312.16 9.80 Turkmenistan 13095.78 10831.57 2258.51 5.70 Subregional average 6876.34 6080.58 788.36 7.40 Remaining countries of the former Soviet Union Belarus 3687.47 3422.28 263.33 1.86 Ukraine 5346.82 4982.57 361.86 2.38 Russian Federation 6523.56 6097.52 415.08 10.96 Moldova 7570.38 7231.99 328.68 9.70 Subregional average 6125.69 5723.62 393.61 8.46 Overall ECA average 5364.24 4972.25 385.32 6.66 Note: Analysis based on national registration data; data refer to latest available years; no recent data available for Bosnia and Herzegovina, Serbia and Montenegro and Turkey. remaining countries of the former Soviet Union (3,383), and Southeastern Europe (1,631). Maxi- mum gains on a country level would be achieved in Turkmenistan (16,795), Kyrgyzstan (9,802), and Romania (9,708). The results of Scenario 4 are shown in Table 23. On average, ECA countries would gain 4,873 YPLL (age 0­64) per 10,000. Gains would be maximal in the remaining countries of the former Soviet Union (7,328), followed by Central Asia and the Caucasus (2,790), the Eastern MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 49 TABLE 21. IMPACT ON YPLL (AGE 0­64 PER 10,000) OF REDUCING INFANT, CHILD, AND MATERNAL MORTALITY TO EU LEVELS Overall gain in Gain due to reducing YPLL after Infant Maternal reaching EU levels mortality 1­4 mortality mortality Eastern European EU candidate countries Czech Republic 0 0 0 0 Slovenia 3.04 0 0 3.04 Poland 2104.18 2068.31 34.81 1.05 Hungary 2160.33 2125.71 34.62 0 Lithuania 2272.74 2022.10 248.28 2.36 Slovakia 2465.84 2383.06 82.10 0.68 Estonia 2744.13 2533.98 204.65 5.50 Latvia 3952.64 3790.64 154.00 8.00 Bulgaria 6182.57 5913.74 263.54 5.29 Romania 9025.92 8655.45 358.50 11.97 Subregional average 3741.20 3604.46 132.93 3.80 Southeastern Europe Croatia 1638.99 1608.20 30.79 0 Albania 4539.70 3871.94 658.74 9.03 Macedonia, FYR 5357.15 5221.27 130.01 5.87 Subregional average 3378.99 3117.52 257.27 4.20 Central Asia and Caucasus Georgia 5251.42 5166.32 81.09 4.01 Azerbaijan 6738.86 5061.13 1672.32 5.41 Tajikistan 6978.99 6827.93 150.35 0.71 Armenia 7547.70 7149.42 387.96 10.32 Kazakhstan 11119.89 10291.95 802.59 25.36 Uzbekistan 12164.72 10289.34 1873.20 2.18 Kyrgyzstan 13468.82 11686.82 1768.92 13.08 Turkmenistan 20205.26 17037.04 3161.66 6.55 Subregional average 10824.28 9452.27 1363.12 8.90 Remaining Countries of the Former Soviet Union Belarus 3118.69 2845.37 271.48 1.84 Ukraine 5013.12 4612.59 397.76 2.77 Russian Federation 7264.23 6791.71 458.75 13.77 Moldova 8798.60 8334.03 453.25 11.32 Subregional average 6556.63 6110.96 435.14 10.54 Overall ECA average 6499.66 5973.57 517.80 8.29 Note: Analysis based on national registration data; data refer to latest available years; no recent data available for Bosnia and Herzegovina, Serbia and Montenegro and Turkey. European EU candidate countries (1,753), and Southeastern Europe (1,135). On a country level, gains would be maximal in the Russian Federation (8,417), Kazakhstan (5,785), and Belarus (5,542). Comparing the impact of the four scenarios on YPLL (Table 24 and Figure 13), it can be seen that average overall gains in the ECA region were maximized when reducing infant, child, and mater- nal mortality to EU levels, reaching 6,500 YPLL per 10,000 (age 0­64). The regional average would 50 WORLD BANK WORKING PAPER TABLE 22. IMPACT ON YPLL (AGE 0­64 PER 10,000) OF REACHING THE LOWEST INFANT, CHILD, AND MATERNAL MORTALITY IN THE SUBREGION Gain due to reducing Overall gain Infant Maternal in YPLL mortality 1­4 mortality mortality Eastern European EU candidate countries Czech Republic 96.13 0 96.13 0 Slovenia 196.64 193.02 0 3.62 Latvia 848.29 843.90 0 4.39 Poland 2787.88 2659.26 126.60 2.02 Hungary 2856.00 2856.00 125.89 0.80 Lithuania 2982.82 2636.07 343.65 3.10 Slovakia 3151.71 2977.21 173.67 0.83 Estonia 3448.50 3131.74 306.10 10.66 Bulgaria 6842.39 6485.84 350.38 6.17 Romania 9708.25 9244.43 450.91 12.91 Subregional average 4272.17 4062.22 217.77 4.54 Southeastern Europe Croatia 0 0 0 0 Albania 2896.59 2266.78 619.15 10.66 Macedonia, FYR 3214.37 3116.17 90.37 7.84 Subregional average 1631.42 1404.55 221.72 5.15 Central Asia and Caucasus Georgia 61.17 0 58.39 2.78 Azerbaijan 3036.35 1386.34 1649.05 0.97 Armenia 3868.08 3502.30 365.79 0 Tajikistan 5412.52 5184.01 228.10 0.41 Kazakhstan 7485.63 6690.09 780.20 15.33 Uzbekistan 7900.60 6063.77 1835.90 0.94 Kyrgyzstan 9801.68 8047.42 1746.43 7.83 Turkmenistan 16795.45 13652.85 3138.82 3.78 Subregional average 7058.65 5709.89 1344.12 4.66 Remaining countries of the former Soviet Union Belarus 0 0 0 0 Ukraine 2004.33 1833.29 169.56 1.47 Russian Federation 4035.55 3802.80 222.82 9.93 Moldova 5446.12 5244.13 193.22 8.78 Subregional average 3383.30 3176.97 198.90 7.42 Overall ECA average 4241.01 3827.00 411.20 6.10 Note: Analysis based on national registration data; data refer to latest available years; no recent data available for Bosnia and Herzegovina, Serbia and Montenegro and Turkey. be 5,364 years for reaching the Millennium Development Goals for infant, child, and maternal mor- tality, and 4,873 when reducing mortality from cardiovascular disease and external causes of deaths to EU levels. The overall average would be lowest with 4,241 years when reaching the lowest sub- regional levels of infant, child, and maternal mortality in Scenario 3. On a subregional and country level, the situation is more complex. The Eastern European EU candidate countries would gain maximum YPLL in Scenario 3, Southeastern Europe and MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 51 TABLE 23. IMPACT ON YPLL (AGE 0­64 PER 10,000) OF REDUCING ADULT MORTALITY FROM INJURIES AND VIOLENCE AND CARDIOVASCULAR DISEASE TO EU LEVELS Decrease due to reducing deaths from Overall gain in YPLL Injuries and violence Cardiovascular disease Eastern European EU candidate countries Slovenia 870.78 764.91 105.87 Czech Republic 1009.11 541.95 467.15 Slovakia 1256.70 504.89 751.81 Poland 1314.46 710.70 603.76 Hungary 1692.41 867.98 824.43 Romania 2061.53 859.46 1202.07 Bulgaria 2174.38 330.84 1843.54 Lithuania 4026.32 3164.83 861.48 Latvia 4626.80 3344.41 1282.39 Estonia 4734.24 3596.13 1138.11 Subregional average 1753.27 886.43 866.83 Southeastern Europe Macedonia, FYR 817.28 0 817.28 Croatia 1134.50 547.24 587.25 Albania 1342.45 594.88 747.57 Subregional average 1134.99 446.35 688.64 Central Asia and Caucasus Armenia 544.25 0 544.25 Tajikistan 1044.22 0 1044.22 Georgia 1127.78 0 1127.78 Azerbaijan 1375.01 0 1375.01 Uzbekistan 2090.00 387.98 1702.02 Kyrgyzstan 3576.28 2006.33 1569.95 Turkmenistan 4108.88 593.36 3515.52 Kazakhstan 5784.99 3296.56 2488.43 Subregional average 2790.46 1002.89 1787.57 Remaining countries of the former Soviet Union Moldova 3918.24 2332.19 1586.05 Ukraine 4747.57 2689.20 2058.37 Belarus 5541.68 3785.56 1756.12 Russian Federation 8417.07 5903.28 2513.79 Subregional average 7328.05 4975.24 2352.81 Overall ECA average 4873.31 3058.85 1814.46 Note: Analysis based on national registration data; data refer to latest available years; no recent data available for Bosnia and Herzegovina, Serbia and Montenegro and Turkey. the countries of Central Asia and the Caucasus would achieve the highest gains in Scenario 2, while the remaining countries of the former Soviet Union would achieve maximum gains in Scenario 4. On a country level, the subregion with the highest consistency is Central Asia and the Caucasus. All countries of this subregion would achieve maximum gains in YPLL in Scenario 2. The average gain in YPLL in Central Asia and the Caucasus in this scenario is 10,824, the highest subregional gain in YPLL in any scenario. In the other subregions, the results are more varied. Slovenia, Croatia, 52 WORLD BANK WORKING PAPER TABLE 24. IMPACT ON YPLL (AGE 0­64 PER 10,000) IN THE FOUR SCENARIOS Scenario 4: Scenario 1: Scenario 2: Scenario 3: Reaching EU levels Reaching Reaching Reaching best for CVD and MDGs 4 and 5 EU levels subregional levels external causes Eastern European EU candidate countries Bulgaria 6140.32 6182.57 6842.39 2174.38 Czech Republic 331.40 0 96.13 1009.11 Estonia 3159.10 2744.13 3448.50 4734.24 Hungary 2185.03 2160.16 2856.00 1692.41 Latvia 4123.32 3952.64 848.29 4626.80 Lithuania 3261.75 2272.74 2982.82 4026.32 Poland 1870.00 2104.18 2787.88 1314.46 Romania 6177.58 9025.92 9708.25 2061.53 Slovakia 3101.72 2465.84 3151.71 1256.70 Slovenia 1033.53 3.04 196.64 870.78 Subregional average 3172.13 3741.20 4272.17 1753.27 Southeastern Europe Albania 593.82 4539.70 2896.59 1342.45 Croatia 2462.85 1638.99 0 1134.50 Macedonia, FYR 1255.43 5357.15 3214.37 817.28 Subregional average 1594.77 3378.99 1631.42 1134.99 Central Asia and Caucasus Armenia 6594.79 7547.70 3868.08 544.25 Azerbaijan 4079.49 6738.86 3036.35 1375.01 Georgia 4836.24 5251.42 61.17 1127.78 Kazakhstan 8270.72 11119.89 7485.63 5784.99 Kyrgyzstan 9610.91 13468.82 9801.68 3576.28 Tajikistan 4282.69 6978.99 5412.52 1044.22 Turkmenistan 13095.78 20205.26 16795.45 4108.88 Uzbekistan 6253.33 12164.72 7900.60 2090.00 Subregional average 6876.34 10824.28 7058.65 2790.46 Remaining countries of the former Soviet Union Belarus 3687.47 3118.69 0 5541.68 Moldova 7570.38 8798.60 5446.12 3918.24 Russian Federation 6523.56 7264.23 4035.55 8417.07 Ukraine 5346.82 5013.12 2004.33 4747.57 Subregional average 6125.69 6556.63 3383.30 7328.05 Overall ECA average 5364.24 6499.66 4241.01 4873.31 Note: Analysis based on national registration data; data refer to latest available years; no recent data available for Bosnia and Herzegovina, Serbia and Montenegro and Turkey. and Ukraine would achieve the highest gains in Scenario 1, while Bulgaria, Hungary, Poland, Romania, and Slovakia would achieve maximum gains in Scenario 3. Surprisingly, in several countries of the former Soviet Union (Estonia, Latvia, Lithuania, and the Russian Federation), as well as in the Czech Republic, gains in YPLL were largest in Scenario 4 (reduc- ing mortality rates from cardiovascular disease and external causes of death to EU levels), in spite of the emphasis on deaths at early ages and under the age of 65 that the measure of YPLL implies. MILLENNIUM DEVELOPMENT GOALS FOR HEALTH IN EUROPE AND CENTRAL ASIA 53 FIGURE 13. OVERALL IMPACT ON YPLL ACCORDING TO THE DIFFERENT SCENARIOS (AGE 0­64 PER 10,000): REGIONAL AND SUBREGIONAL AVERAGES 3172 Eastern European EU 3741 candidate countries 4272 1753 1595 South Eastern Europe 3379 1631 1135 6876 Central Asia & 10824 Caucasus 7059 2790 6126 Remaining countries of 6557 the former Soviet Union 3383 7328 5364 Overall ECA average 6500 4241 4873 0 2000 4000 6000 8000 10000 12000 Scenario 1: Reaching MDGs Scenario 2: Reaching EU levels Scenario 3: Reaching best sub-regional levels Scenario 4: Reaching EU levels for CVD and external causes Note: Analysis based on national registration data; data refer to latest available years. The results of the calculation of YPLL illustrate how the method of assessing population health influences the perceived importance of challenges to population health. Invariably, value judg- ments are implied in the choice of method. With the notable exception of several countries of the former Soviet Union, including the Russian Federation, the average ECA gain in YPLL was largest in Scenario 2 (reaching EU levels on infant, child, and maternal mortality). This is perhaps not sur- prising, because the use of YPLL gives emphasis to death at young ages. With the limit of 65 years, reducing mortality after the age of 64 does not yield any benefits in YPLL. This obviously implies a judgment on the value of life at certain ages. Although societies usually hold the prevention and treatment of death in young ages as a higher priority than the prevention of death in older ages, there is hardly any society that would not value health gains over the age of 64. While thus not unproblematic, the measure of YPLL acknowledges that deaths at young ages represent a greater social and economic loss to a country. In assessing the appropriateness of the Millennium Development Goals to the countries of the ECA region, the gains in YPLL clearly demonstrate the benefits of reducing infant and child mor- tality. This is particularly the case for the countries of Central Asia and the Caucasus, which have the highest infant mortality rates in the region. As is the case for life expectancy at birth, gains in 54 WORLD BANK WORKING PAPER YPLL in these countries would be much higher, if World Bank data on infant mortality were used instead of the official data reported to WHO. Given the implicit priority YPLL gives to deaths at young ages, it is especially striking that in some countries Scenario 4 emerges as yielding the most gains. Although outside the scope of this project, it should be noted that the method of YPLL has been applied, using different age limits or weighting factors to change the emphasis given to deaths at different ages. There is not, however, a consensus as to whether and how this should be done. To illustrate the effect of weighting factors on YPLL, we have recalculated the YPLL for ages 0­64 attributable to achieving the Millennium Development Goal for infant mortality for two countries with different patterns of mortality, Russia and Azerbaijan. We used the weightings developed in the Global Burden of Disease, in which the weight applied to years lost at different ages is Cxe-bx, where x is age and C and b are constants with values 0.01658 and 0,04 respectively. This weighting reduced the YPLL for achieving the MDG for infant mortality for Russia from 6,098 to 495 and for Azerbaijan from 3,240 to 263. When using the same weighting for the reduction of mortality from cardiovascular disease, the YPLL increased from 2,514 to 2,820 for Russia and from 1,375 to 1,504 for Azerbaijan. These results indicate that gains in YPLL resulting from reductions of adult mortality would have been even greater if the Global Burden of Disease weighting was used. The use of YPLL, or as in the main part of this study, life expectancy at birth, for judging the appropriateness of the health-related Millennium Development Goals for the countries of the ECA region requires another qualification. The case of maternal mortality clearly reveals the limitations of both measures. Reducing maternal mortality in Scenarios 1, 2, and 3 did result in very minimal, sometimes hardly detectable, gains in life expectancy at birth or in YPLL. This shows the degree of interpretation the results require. Obviously, the study does not intend to make the case that mater- nal mortality does not matter. Instead of directly guiding policy responses, the results of the analysis need to be read with caution. REFERENCES Acheson, D. 1998. Independent Inquiry into Inequalities in Health. London: HM Stationery Office. Aleshina, N. and G. Redmond. 2003. "How High is Infant Mortality in Central and Eastern Europe and the CIS?" Innocenti Working Paper No.99. Innocenti Research Centre: Florence. Badurashvili, I., M. McKee, G. Tsuladze, F. Meslé, J. Vallin, and V. Shkolnikov. 2001. "Where there are no data: what has happened to life expectancy in Georgia since 1990?" Public Health 115:394­400. Bozicevic I., S. Oreskovic, R. Stevanovic, U. Rodin, E. Nolte, and M. McKee. 2001. "What is happening to the health of the Croatian population?" Croatian Medical Journal 42:601­5. Chiang, C. L. 1968. Introduction to stochastic processes in biostatistics. New York: John Wiley & Sons, Inc. Fylkesnes, K., Z. Ndhlovu, K. Kasumba, R. Mubanga Musonda, and M. Sichone. 1998. "Studying dynamics of the HIV epidemic: population-based data compared with sentinel surveillance in Zambia." AIDS 12:1227­34. Government of Tajikistan. 2003. Progress toward the Millennium Development Goals in the Republic of Tajikistan. Draft, April 30th. Government of Tajikistan: Dushanbe. Crampin, A., J. Glynn, B. Ngwira, F. Mwaungulu, J. Ponnighaus, D. Warndorff, and P. Fine. 2003. "Trends and measurement of HIV prevalence in northern Malawi." AIDS 17:1817­25. International Monetary Fund, OECD, United Nations, and World Bank. 2000. 2000--A Better World for All. Progress towards the international development goals. Kelly, J. A., and Y. A. Amirkhanian. 2003. "The newest epidemic: a review of HIV/AIDS in Central and Eastern Europe." International Journal of STD & AIDS 14:361­371. Kowal, P. R. and A. D. Lopez. 2003. "Child survival." The Lancet 362:915. Lock, K., E. M. Andreev, V. M. Shkolnikov, and M. McKee. 2002. "What targets for international development policies are appropriate for improving health in Russia." Health Policy and Planning 17:257­263. McKee, M., J. Healy, and J. Falkingham, eds. 2002. Health care in central Asia. European Obser- vatory on Health Care Systems Series. 55 56 WORLD BANK WORKING PAPER Miller, D. S. and A. M. Ryskulova. 2004. Evaluation of Epidemiologic Surveillance Systems in Eastern Europe and Central Asia. Manuscript in preparation. World Bank. Moshiro, C., R. Mswia, K. Alberti, D. Whiting, and N. Unwin (AMMP Project Team). 2001. "The importance of injury as a cause of death in sub-Saharan Africa: results of a community- based study in Tanzania." Public Health 115:96­102. Ruehl, C., V. Pokrovsky, and V. Vinogradov. 2002. The Economic Consequences of HIV in Russia. World Bank. Accessed at: http://www.worldbank.org.ru. UNAIDS. 2002a. Ukraine. Epidemiological Fact Sheets on HIV/AIDS and Sexually Transmitted Diseases. 2002 Update. ------. 2002b. Russian Federation. Epidemiological Fact Sheets on HIV/AIDS and Sexually Transmitted Infections. 2002 Update. UNDP. 2003. Human Development Report 2003. New York, Oxford: Oxford University Press. UNICEF. 2001. A Decade of Transition. The MONEE Project on CEE/CIS/Baltics. Florence: Innocenti Research Centre. ------. 2003. Social Monitor 2003. Innocenti Social Monitor. Florence: UNICEF Innocenti Research Centre. Walker, R., D. McLarty, H. Kitange, D. Whiting, G. Masuki, D. Mtasiwa, H. Machibya, N. Unwin, and K. Alberti. 2000. "Stroke mortality in urban and rural Tanzania. Adult Morbidity and Mortality Project." Lancet 355:1684­7. World Health Organization. 2000. The World Health Report 2000. Health Systems: improving performance. Geneva: World Health Organization. ------. 2003. Health For All (HFA) Database. Regional Office for Europe. January. World Bank. 2003a. Averting AIDS Crises in Eastern Europe and Central Asia. Washington, D.C.: World Bank. ------. 2003b. Estimating Child Mortality. Draft, February 7th. Zaba, B., T. Boerma, and R. White. 2000. Monitoring the AIDS epidemic using HIV prevalence data among young women attending antenatal clinics: prospects and problems. AIDS 14:1633­45. Millennium Development Goals for Health in Europe and Central Asia is part of the World Bank Working Paper series. These papers are published to communicate the results of the Bank's ongoing research and to stimulate public discussion. This study aims to contribute to the debate about the appropriate- ness of health-related Millennium Development Goals (MDGs) for the countries of the Europe and Central Asia (ECA) region. The study has important implications for policy choices at the regional, subregional, and country levels. Despite the fact that aver- ages mask variations that exist within countries, certain conclusions stand out: N For the ECA region as a whole, proportionately more gains in life expectancy would accrue from the control of non-communicable diseases (NCDs) than from achieving the targets in the classic MDGs. This pattern holds in all subregions. N There are proportionately more gains to be had from achieving the classic MDGs in Central Asia and the Caucasus subregions. For priority setting, policy formulations, programs, and development assistance, the key messages from this study are: N the importance of reducing morbidity and premature mortality from NCDs and external causes across the entire region; N the need for particular attention to the classis MDG indicators (infant mortality rates, under-5 mortality rates, and maternal mortality ratios) in the countries of Central Asia and the Caucasus; N the importance of establishing effective and sustainable surveil- lance, vital registration, and health information systems to pro- vide valid data for local decisions and international comparisons, and N thecontinuedneedforinteragencycollaborationinsupportofall these objectives. World Bank Working Papers are available individually or by subscrip- tion, both in print and online. ISBN 0-8213-5878-2 TMxHSKIMBy358788zv":&:>:(:> THE WORLD BANK 1818 H Street, NW Washington, DC 20433 USA Telephone: 202 473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org