Report No. 23811-BR Brazil Maternal and Child Health February 26, 2002 Brazil Country Management Unit Human Development Sector Management Unit Latin America and the Caribbean Region Document of the World Bank GOVERNMENT FISCAL YEAR January 1 - December 31 CURRENCY EQUIVALENTS Currency Unit = Brazilian Real (BRL) US$1.00 = BRL 1.79 (December 1999) ACRONYMS AND ABBREVIATIONS AIH Authorization for Hospital Admission (Autoriza,cdo de Internacao Hospitalar) AIDS Acquired Immune Deficiency Syndrome ANC Antenatal Care ANVISA National Agency for Health Surveillance (Agencia Nacional de Vigildncia Sanitdria) BMI Body Mass Index DHS Demographic and Health Survey DPT Diphtheria, Pertussis, and Tetanus DRG Diagnosis Related Group FUNASA National Health Foundation (Fundaq4o Nacional de Sauide) GDP Gross Domestic Product HIV Human Immunodeficiency Virus IBGE Brazilian Institute for Geography and Statistics (Instituto Brasileiro de Geografta e Estatistica) IMCI Integrated Management of Childhood Illness IMR Infant Mortality Rate LSMS Living Standards Measurement Survey MEC Ministry of Education and Culture (Ministerio de Educacdo e Cultura) MS Ministry of Health (Ministerio da Sau'de) MMR Maternal Mortality Ratio NGO Non-Governmental Organization OECD Organization for Economic Cooperation and Development Vice President: David de Ferranti Country Director: Vinod Thomas Sector Director: Ana-Maria Arriagada Sector Manager: Charles Griffin Task Manager: Varun Gauri ORT Oral Rehydration Therapy PACS Community Health Agents Program (Programa de Agentes Comunitarios de Sazude) PAHO Pan American Health Organization PAISM Integrated Program for Women's Health (Programa de Aten,do Integral a Saude da Mulher) PCCN Program to Combat Nutritional Deficiencies (Programa de Combate as Carencias Nutricionais) PPP Purchasing Power Parity PSF Family Health Program (Programa de Sauide da Familia) SPS Policy Secretariat of the Ministry of Health (Secretaria de Politicas de Saude) SUS Single Unified Health System (Sistema Unico da Saide) STD Sexually Transmitted Disease UNICEF United Nations Children's Fund UNDP United Nations Development Program WHO World Health Organization ACKOWLEDGEMENTS The report was jointly prepared by a World Bank team consisting of Varun Gauri (Economist, DECRG, task manager), Sandra Rosenhouse (Population and Health Specialist, LCSHH), Flavia Bustreo (Public Health Specialist, HDNHE), and Heman Montenegro (Health Specialist, LCSHH). Commissioned papers written by Cesar Victora, Ana Luiza d'Avila Vianna, Elizabeth Barros, Leny Trad, Marilyn Nations, Anibal Faiundes, and Mario Roberto dal Poz contributed substantially to the study. The World Bank team worked together in Brazil in January, 2000, and Ricardo Fescina and Zuleica Albuquerque of PAHO-Brasilia accompanied the team in Goias during that January research trip. Mariam Claeson (Principal Public Health Specialist, HDNHE) and Jerker Liljestrand (Principal Health Specialist, HDNHE) offered guidance and technical expertise. Maria Eugenia Bonilla-Chacin contributed research assistance to the study, and Tourya Tourougui and Sarah Menezes provided administrative support. The report was produced largely under the supervision of Gobind T. Nankani (former Director, Brazil Country Management Unit - LCC5C), Patricio Millan (former Sector Leader, LCC5C), Xavier Coll (former Sector Director - LCSHD) and Charles Griffin (Sector Manager, LCSHH). The outside reviewers were Andre Medici (IDB) and Narayan Sastry (RAND). The collaboration of many other Government officials, other individuals, and institutions is gratefully acknowledged, as are comments received the during the workshop in Brasilia, on June 6, 2001, where the report's recommendations were discussed. Many comments from that workshop were reflected in the report. Nevertlheless, the views expressed in this report are exclusively those of the World Bank. They should not be attributed to any of the individuals or institutions acknowledged above. The report was mostly prepared during 2000 and is based on data available in that year. More recent data and important recent policy changes are not reflected in this report. TABLE OF CONTENTS 1. THE HEALTH STATUS OF MOTHERS AND CHILDREN IN BRAZIL .................................................I Why M aternal and Child Health? .............................................................................. I How do Brazil and its Regions Compare? ..............................................................................2 Infant mortality: composition, levels and trends ..............................................................................9 Child mortality: composition, levels, and trends ............................................................................. 13 Maternal mortality: composition, levels, and trends . ............................................................................. 15 Summary .............................................................................. 16 2. PRIORITIZING HEALTH INTERVENTIONS FOR IMPROVING MATERNAL AND CHILD HEALTH IN BRAZIL ............................................................................. 19 Interventions Targeted at Maternal and Perinatal Health ............................................................................. 19 Interventions to Promote Infant and Child Health . ............................................................................. 22 National and Regional Priorities .............................................................................. 25 3. HEALTH CARE AND HEALTH PROGRAMS FOR MATERNAL AND CHILD HEALTH IN BRAZIL ....... ...................................................................... 37 The Sistema Unico da Saude and Sector Issues ............................................................................. 37 Government Programs in Maternal and Child Health . ............................................................................. 41 The Family Health Program ............................................................................. . 45 Summary ............................................................................... 49 4. POLICY RECOMMENDATIONS ............................................................................. 53 Goals ............................................................................. 53 Basic Actions to Achieve the Goals ............................................................................. 53 Measures to Improve the Quality and Effectiveness of Care for Mothers and Children ............ ....................... 54 Evaluation ............................................................................. 55 APPENDIX A: A LIFE-CYCLE APPROACH TO MATERNAL AND CHILD HEALTH IN BRAZIL . 57 The Lifecycle Approach . . .57 Routine Interventions during the Reproductive Years . . .58 Interventions during Pregnancy: Antenatal Care . . .60 Pregnancy Termination and Post-abortion Care . . .63 Interventions during Delivery and Post-partum period . . .64 Interventions after Delivery ...66 Essential Newborn Care . . .66 Breastfeeding Promotion . . .67 Immunization...68 Management of Illnessss.69 Nutrition Interventions . .70 Care for Development .71 Accident Prevention . . . . .72 APPENDIX B: TARGETING AND IMPACT OF THE FAMILY HEALTH PROGRAM .73 Targeting.73 Impact.76 REFERENCES.79 Executive Summary I At Brazil's level of income per capita, mothering and simply being a child are riskier activities than they should be. Chapter 1 argues that a major reason for this is the concentration of illness and death among the poor: infant mortality rates jump 80-100% from the next to lowest income quintile to the lowest. Several interventions, such as strategies to promote productivity, increase economic opportunities, and enhance maternal education among the poor, could help reduce these inequalities. This report focuses on the health care system. The health care system is particularly important for enhancing maternal and perinatal health, and certain health care interventions, such as immunizations, oral rehydration therapy, and case management, can reduce infant and child mortality rates due to infectious and communicable diseases. In addition, because more than half of infant mortality in all regions of Brazil is a consequence of complications before or during the first seven days of life, and because maternal mortality is determined largely by the failure to diagnose and treat complications in pregnancy, adequate health care, or its absence, will be increasingly important in determining future trends in maternal and child health. 2 Chapter 2 uses estimates of avoidable child deaths to prioritize health interventions in Brazil. Health priorities for the North and Northeast, in which Brazilian citizens are worse off according to every available social indicator and where infectious diseases remain a significant challenge, are expanding access to and improving the quality of interventions to promote the management and treatment of illnesses, family planning, and nutrition education. For all regions of the country, health priorities are enhancing the quality of antenatal, delivery, postpartum, and neonatal care, and their interrelationships, as well as breastfeeding promotion and the introduction of the Hib vaccine. 3 The health care system in Brazil has established the important principle of universality of coverage, but it faces several key challenges. Its expenditure patterns favor the regions that are already better off and healthier, it is extremely medicalized, it needs to take account of diverse practices and cultures in Brazil, and its referral system works poorly. Chapter 3 argues that federal government programs in women's health and children's health could be expanded to address the priorities derived above. That chapter also emphasizes the importance of utilizing nutrition education, introducing reproductive health in school health programs, and designating sites for antenatal care, delivery, and postpartum care for pregnant women well before they enter labor. The Family Health Program aims to create more sophisticated and permanent primary health care for enrolled populations in a fixed geographic areas, emphasizing preventive care, inter-sectoral collaboration, and community-based information and education regarding health. For the Family Health Program to promote maternal and child health effectively, the federal, state, and municipal governments need to make a policy decision to prioritize maternal and child health, the program needs to address human resource constraints and address the question of how team members are contracted, and referrals from and counter-referrals to Family Health teams need to be made more effective. 4 Chapter 4 makes the following policy recommendations regarding sector goals, basic actions to achieve the goals, measures to improve the quality and effectiveness of care, and evaluation: Goals * It is feasible to continue reducing the infant mortality rate in Brazil at 5% a year, reaching a target goal of 24 in the year 2005. Basic Actions to Achieve the Goals * Focus efforts on the North and Northeast, rural areas, and the poor, and, in those areas, prioritize inter-sectoral programs that reduce poverty and increase education. In addition, health initiatives on the management of childhood illness, family planning to improve birth spacing, breastfeeding promotion, and improved antenatal and delivery care are priorities. Those interventions could prevent as many as 40% of deaths among children under five in the North and Northeast. * In the country as a whole, prioritize improvements in the quality of antenatal, delivery, neonatal, and postpartum care; the promotion of breastfeeding; the introduction of the Hib vaccine; and the case management of infectious diseases. Those health initiatives could prevent one-third of all deaths among children under five. * Gradually equalize SUS expenditures per person and per birth across the Brazilian regions. Measures to Improve the Quality and Effectiveness of Care for Mothers and Children * Quickly expand programs for training obstetric nurses, and revise legal regulations so that well-trained nurses and well-trained conununity health agents can perform more complex tasks and procedures. * Improve the functioning of the referral system in the SUS, including the use of community health agents and Family Health teams to pre-designate sites for antenatal, delivery, postpartum, and neonatal care for every pregnant woman and newborn. * Utilize NGOs and community-based organizations to provide educational materials and activities for behavior change in nutrition, reproductive health, and the recognition and prevention of infectious diseases. * Incorporate curricula on reproductive health issues into school health programs, to the extent allowed by Brazilian law, and utilize NGOs and community-based organizations for developing materials and reaching youth not attending school. * Building on federal programs in women's health, child health, and family health, quickly expand initiatives for improving family planning programs and the Integrated Management of Childhood Illness in the North and Northeast, and for improving the quality of antenatal, delivery, neonatal, and postpartum care, including breastfeeding promotion and the Hib vaccine, in all regions of the country. Evaluation * Conduct a regular household survey of mortality rates, causes of death, and morbidity in order to evaluate progress and adjust programs. 1. THE HEALTH STATUS OF MOTHERS AND CHILDREN IN BRAZIL Why Maternal and Child Health? 1.1 For advocates of development, the reasons for improving the health of mothers and children in developing countries are obvious. Poor health prevents people from living rich, fully realized lives. "The elimination of illiteracy, ill-health, and other avoidable deprivations are valuable for their own sake - they are 'the tasks' that we face. . . It [is] a mistake to see the development of education, health care, and other basic achievements only or primarily as expansions of 'human resources' - the accumulation of 'human capital' - as if people were just the means of production and not its ultimate end. The bettering of human life does not have to be justified by showing that a person with a better life is also a better producer" (Sen 1997). 1.2 Still, because it is arguably possible to spend too much on health care, or to spend it in an inefficient manner, or to sacrifice other opportunities for its sake, policy makers are interested in the additional benefits that health care investments provide. It would seem that there ought to be a connection between healthier mothers and healthier children, and that healthier children ought to be more likely to do better in school, and to stay in school and avoid dangerous factory or field work and prostitution, and ought therefore to be less poor over the course of their lives. Although empirical work has identified a significant relationship between health and nutrition in childhood and lifetime cognitive and motor skills (Hauser 1998, Martorell 1998), it has been more difficult to establish that health mitigates poverty and enhances labor productivity. The reasons for this difficulty include measurement problems related to the fact that health is multi-dimensional, changes over time, and is unreliably reported, and conceptual problems related to the dynamic relationship between health and income, including the facts that labor can be substituted for within households, that health falls in importance as the physical intensity of labor declines, and that health is in general both a cause and effect of labor productivity (Strauss and Thomas 1998). Recent studies have demonstrated some success in identifying an effect. For instance, at the individual level there is a correlation between adult height and income even among uneducated Brazilian men and women (Thomas and Strauss 1997). At the macro level writers have argued that nutritional gains account for a large part of economic growth in Europe over the past two centuries (Fogel 1992), that malaria and endemic diseases depress economic growth in Africa (Gallup and Sachs 1998), and that declining mortality and fertility rates were associated with the unprecedented economic growth in East and Southeast Asia from the 1960s to the I 990s (Asian Development Bank 1997). In other words, although the magnitude of the effects remain unclear, there is growing evidence that health enhances productivity and growth, and reduces poverty. I How do Brazil and its Regions Compare? 1.3 Although mothers and children in Brazil are healthier now than they were two decades ago, Brazilians and the international community continue to worry that mothers and children suffer substantially more afflictions than they should given the country's income. Infant mortality, for instance, is more than three times higher in Brazil than in Poland, Malaysia, and Chile, countries with comparable GDP per capita, and more than one and one-half times higher than in Mexico. Similarly, the rates of maternal mortality range from three to ten times higher in Brazil than in those comparator countries, and twice as high as in Mexico. Because of figures like these, UNDP's 1999 Human Development Report ranks Brazil seventy-ninth in its index of human development, just behind Kazakhstan, the Philippines, and Saudi Arabia, and sixteen places below its ranking in per capita GDP. 1.4 A more explicit statistical test of health status in selected countries relative to their income makes this clear.' The graph below plots the logarithm of infant mortality rates for 31 countries as a function of the logarithm of national per capita income, adjusted for purchasing power parity, and a constant, and draws a line where fitted values would lie in a simple linear relationship. The model has a reasonable degree of explanatory power (the adjusted R is 34%, and the F-test finds that there is less than a 0.1% possibility that the relationship between GDP and infant mortality is the result of chance).2 The elasticity of infant mortality with respect to GDP is -0.85. Although the regression was fitted using 31 countries, the graph indicates where the regions of Brazil, were they countries, would lie, and plots the other countries with an "X". t Infant mortality is the chosen outcome measure in this test though child mortality would serve just as well: the two variables are highly correlated. Maternal mortality ratios can be unreliable because the event is rare and under- reported. Nevertheless, casual comparisons of estimates of regional maternal mortality ratios in Brazil (Table 7) to those in other countries demonstrate that maternal mortality in every region appears higher than in countries with comparable levels of income. The magnitude of perinatal mortality, which includes deaths during the first week of life and during late pregnancy and is usually made up of roughly equal portions of both, cannot be determined through the DHS and existing data. 2 The explanatory power of this model is sensitive to the particular set of countries included in the sample. Other attempts to explore this subject (World Bank 2000) have produced a higher adjusted R2, but at the expense of excluding certain countries from the sample that would seem comparable, such as Thailand and Poland. In the present analysis, countries were selected by taking the fifteen countries both directly above and below Brazil in PPP per capita income, excluding African countries and island nations. The figures for purchasing power parity income for Brazil's regions were calculated assuming that their ratio to regional income is the same as that for the country's purchasing power parity GDP to its unadjusted GDP, assuming, in other words, that internal prices are constant. This is a reasonable assumption for these purposes: although the prices of nontradeables such as housing appear higher in more urban settings, adjusting for the quality of dwelling stock, particularly at levels near the poverty line, can reduce urban-rural cost-of-living differences substantially (Ravallion and van de Walle 1991). 2 Figure 1: Infant Mortality and GNP N EAST 4 3.5 -BRZIL \C'VEST SUEAST SOUTH 3 o E ' 2 2.5 o 2 t .5 7.5 8 8.5 9 9.5 10 Log of GDP per capita (PPP), 1997 1.5 All regions of Brazil, not only the poorer North and Northeast, have higher rates of infant mortality than they should, given their incomes. In other words, it appears that every region helps make Brazil an "outlier": although the North and Northeast have higher rates of infant mortality and push up the national infant mortality rate more than the other regions, there appears to be a nationwide problem in Brazil at every level of development. An analysis confirms this impression. When the countries and regions of Brazil are ranked by their actual or attributed residuals, in descending order, four of the top ten observations are Brazil and its regions.3 Notably, the Southeast ranks above even the Northeast in "excess" mortality, given its income. Table 1: Infant Mortality as a Function of GDP Per Capita: Countries and Regions in order of Largest Positive Studentized Residuals Country or Region Per capita GDP (PPP), Infant Mortality Rate, Predicted Infant 1997 nominal US$ 1997 Mortality Rate 1.Mexico 8,110 31 13 2. Southeast Brazil 10,175 25 11 3. Northeast Brazil 3,413 59 27 4. Brazil 6,350 34 16 5.Thailand 6,490 33 16 6. Argentina 10,100 22 11 7. South Brazil 8,760 23 12 8. Peru 4,580 40 22 9. Guatemala 4,060 43 29 10. Indonesia 3,390 47 28 Source: Author's calculation based on data from World Bank, IBGE. 3None of the regions or countries is formally an outlier in the sense that none of the studentized residuals is greater than 2. Note that the Northeast remains a large positive "outlier," ranking fourth in positive studentized residuals, when the sample of comparator countries is replaced by a set of 33 countries with similar PPP GDP per capita (between $2400-$4600 in 1997), including African countries and islands. 3 1.6 The existing literature has demonstrated that, in addition to the logarithm of per capita GNP, a number of other variables are associated with cross-sectional variance in infant mortality (Mosely and Chen 1984, World Bank 2000). These variables include income inequality, female education, ethnic fragmentation, and predominant religion.4 In addition, access to safe water, which reduces the incidence of diarrhea and other parasitic diseases, though not always significant in international cross-sectional studies, has been shown to be associated with lower infant mortality rates in a number of studies within Brazil and in other countries (Esrey and others 1985, World Bank 1998, Rose 1997). Interestingly, there is some evidence that in Brazil part of the impact of safe water occurs indirectly, through the public information that accompanies water hookups (Sastry 1996). The literature also demonstrates that access to particular kinds of health care can reduce infant mortality rates (though aggregate measures such as public sector health spending generally do not appear significant in international cross- sectional comparisons). In Brazil, for example, there is evidence that the use of oral rehydration therapy played a larger role than income, education, and access to water in the sharp decline in infant deaths due to diarrhea in the 1980s (Victora 1996). Studies have also demonstrated that immunization and better case management for respiratory infections, including antibiotics, are associated with lower infant mortality rates even after controlling for socioeconomic variables. Similar claims can be made in Brazil for family planning (principally through increasing spacing among births), antenatal care, skilled attendance at birth, and other specific health interventions.5 Finally, it is possible, and perhaps probable, that it is the distribution of income, education, access to health care, access to water, and the other factors, and not merely regional or country averages, that determines infant mortality rates. In other words, if one had data on the access to these inputs by sub-groups (including income decile) in the countries, one might be able to demonstrate that it is the unusually low levels of income, education, health care, access to safe water, and other private and public inputs among certain groups that explain why Brazil has higher rates than its average income or education would suggest. 1.7 Of the variables mentioned in the preceding paragraph, only adult literacy both significantly explained infant mortality in the sample of comparator countries and was also available at the regional level in Brazil. Continuing the analysis with the sample of countries above, then, the graph below depicts fitted and observed values in a model estimating infant mortality as a function of adult literacy and a constant. This model has an adjusted R2 of 44%, and the F-test again finds that there is less than 0.1% probability that this relationship is due to chance. Again, the predicted values are fitted using the country sample set, but the figure shows where the regions of Brazil would lie were they countries. 4 Filmer and Pritchett (1999) use these variables to explain 95% of the cross-sectional variance in infant mortality in a sample of 98 countries. 5 See Victora (2000) for a comprehensive review of the Brazilian literature on the effects of health interventions on infant, child, and maternal mortality. All studies cited in that paper control for socioeconomic variables. 4 Figure 2: Infant Mortality and Illiteracy 80 - N EAST O /x~~~~~~~~~~~ 't~ 40 - North x D x x~~~~~~Nt BRAZI- E SE 9 20- <2 xX X S 0 1 0 20 30 40 Adult Illiteracy Rate, 1997 1.8 From this graph, it appears that in the Southeast, South, and Center West regions of the country low levels of educational attainment account for infant mortality rates better than income does. Seemingly, those regions have high infant mortality rates given their income at least in part because their educational levels are relatively low. It should be noted, however, that observed infant mortality rates for those regions remain above the predicted rates. The graph also suggests that in the North and Northeast, however, infant mortality is higher than it should be given the levels of educational attainment in those regions. An analysis of actual and attributed residuals, similar to that conducted for the logarithm of GDP, above, shows that the North and Northeast occupy two of the top five places, with the Northeast having by far the largest positive residual. 1.9 The next step is to use both education and per capita income as explanatory variables. When the model is re-specified to include them both, table 2 below shows that it again appears that Brazil's regions have higher rates of infant mortality than they should. Compared to the estimates based on income alone, the predicted infant mortality rates are now higher for all regions and for the country as a whole, demonstrating that educational levels account for at least part of the unexplained variance in the first regression; but the predicted rates remain consistently lower than the observed rates for all of the regions. The Northeast has unusually high observed rates (it alone has an attributed studentized residual greater than 2), particularly in comparison with Honduras, which has a similar illiteracy rate and lower GDP per capita. One country of particular interest is the negative "outlier" Malaysia, which has both less income and more illiteracy than South and Southeast Brazil, and yet has an infant mortality rate significantly below the actual and even the predicted rates for the South and Southeast. 5 Table 2: Infant Mortality Rates as a Function of Adult Illiteracy and GDP Per Capita, in Selected Countries and Regions Country or Illiterate Adults, Per capita GDP Infant Mortality Predicted Infant Region Percent, 1997 (PPP), 1997 Rate, 1997 Mortality Rate nominal US$ Costa Rica 5.0 6,510 12 18 Paraguay 7.5 3,860 23 25 Venezuela 8.0 8,660 21 18 Southeast Brazil 8.7 10,175 25 17 South Brazil 8.9 8,760 23 18 Colombia 9.0 6,570 24 21 Portugal 9.1 14,180 6 14 Ecuador 9.3 4,700 33 24 Mexico 10.0 8,110 31 20 North Brazil 11.6 4,505 36 27 Center-West 11.6 6,850 25 23 Brazil Malaysia 14.4 7,730 11 24 Indonesia 15.0 3,390 47 32 Brazil 16.1 6,350 34 27 Northeast Brazil 28.7 3,410 59 43 Honduras 29.4 2,260 47 45 Source: Author's calculation based on data from World Bank, IBGE. 1.10 The data above demonstrate that infant mortality rates are highly unequal in Brazil's regions, with substantially higher rates in the poorer North and Northeast. A disaggregation of the data by income quintile, shown in Table 3 below on the basis of data extracted from the Demographic and Health Survey (DHS) of 1996 and the Living Standards Measurement Survey (LSMS) of 1997, further highlights these inequalities. In estimates based on those two household surveys, the infant mortality rates of the lowest income quintile are three to four times the level of rates in the highest quintile; and the jump in infant mortality from the second to the lowest income quintile is between 80-100%. Wagstaff (2000) finds that these inequalities in infant and child mortality in Brazil are far higher than in any of the other eight countries in his sample, which includes Pakistan, South Africa, and Nicaragua. It does appear, then, that the concentration of deaths among the poor is a large part of the reason why national rates are unusually high in Brazil, an explanation that gains even more credence when one realizes that the income quintiles in the table below below are for mothers, not children, and that the lowest income quintile of mothers in Brazil accounted for 34% of all births in the LSMS sample. Calculations on the LSMS estimates suggest that if the jump in infant mortality from the fourth to the lowest income quintile were reduced to 50% (which is still relatively high), the national infant mortality rate would fall approximately 15%. Figures like these suggest that Brazil should concentrate its efforts on reducing the enormous gap in infant mortality between its poorest citizens and the rest of the country. Reaching the poor, who are often excluded from government services in education and health care as well as from private sector economic opportunities, will 6 be a challenge. Still, it is fortunate that a variety of interventions - income, health care, education, access to safe water - can both independently and jointly reduce infant and child mortality. Many of these same interventions will also, up to a point, help to reduce maternal mortality, which presumably is also higher among poor women. To improve maternal and child health overall, Brazil should focus its efforts in all sectors on the poorest segment of the population. Table 3: Indirect Infant Mortality Rate Estimates in Brazil, by Income Quintile of Mothers Income quintile LSMS estimates (1987-1992) DHS estimates (1986-1996) Lowest 72.7 83.2 Second 37.0 46.7 Third 32.7 32.9 Fourth 17.0 24.7 Highest 15.3 28.6 Sources: Wagstaff (2000) for the LSMS; Gwatkin and others (2000) for the DHS. Note: The LSMS collected data only in the Northeast and Southeast regions of the country. The DHS data shown above are only for urban areas. The income quintiles for the DHS sample were constructed on the basis of information regarding household assets. 1.11 This report, while recognizing the importance of interventions in all sectors for maternal and child health, particularly poverty reduction and education for the poor, focuses on the health care system itself and existing and proposed government programs in Brazil to improve the health of mothers and children. There are several reasons for this focus on the health care system. First, there is evidence that particular health care interventions, such as immunization and oral rehydration therapy, played a significant role in reducing infant mortality rates and improving other measures of child health during the last decade. Health care information and promotion for the poor might be especially helpful in promoting maternal and child health because there is some evidence that the mechanism through which mothers' schooling enhances child health is by increasing access to knowledge about healthy behaviors rather than by improving literacy and numeracy per se, or by increasing earning power (Glewwe 1999). A second reason for focusing on the health care system is that access to health care is critical for maternal health. Complications during delivery are almost impossible to predict; and as a result avoiding maternal mortality depends on quickly diagnosing and treating obstetric emergencies when they arise. Historical records demonstrate that maternal mortality rates are largely a function of the quality of health care, not of the impact of education, nutrition, or income. For instance, maternal mortality ratios remained unchanged in England and Wales from 1840 to 1900 7 despite significant gains in income, education, and access to safe water. They started to decline in the early 1900s, when skilled attendance at birth became widely available, falling from 500- 600 maternal deaths per 100,000 live births in 1850 to around 400 after the turn of the century, and then fell further after technology to treat obstetric complications arrived starting in the mid 1930s, reaching 87 deaths per 100,000 live births in 1950. Rates in Sweden also fell when skilled birth attendance became available for all women in that country, falling from 500-600 in 1850 to 250 by 1900 (De Brouwere and others 1998, Maine and Rosenfield 1999). 1.12 Third, whatever the historical causes of improvements in maternal and child health in Brazil, an examination of the composition of under five mortality, presented in the next section, makes clear that future improvements will increasingly rely on the health care system. That is because perinatal mortality, death in late pregnancy or as a result of complications in the first 7 days of life, is now responsible for over half of infant deaths in every region of Brazil, and almost half (48.5%) of all deaths among children under 5.6 Although access to safe water can increase maternal health prior to delivery and reduce complications, and although education can promote health-seeking behavior among pregnant women, most perinatal deaths are the direct result of poor quality health care. This point is also made clear by examining the incidence of low birth weight among infants in Brazil. Low birth weight incidence in Brazil is about 8%-9% in all regions of the country, even in the Northeast.7 This is not much higher than average rates in OECD countries, about 6%, yet the infant mortality rates in Brazil's regions are from 3-6 times higher than in OECD countries, as table 4 below demonstrates. This suggests that infants die in Brazil not primarily because they are born prematurely or because maternal nutrition interferes with gestation, both of which are associated with low birth weight, but most often because the care they receive before, during, and after delivery is inadequate. Table 4: Incidence of Infant Low Birth Weight and Infant Mortality in Selected Countries Country Low birth weight incidence, Infant mortality rate, 1990-97, percent 1996 Chile 5 12 France 5 5 Canada 6 6 United Kingdom 7 6 Cuba 7 9 Argentina 7 22 United States 7 7 Brazil 8 36 Malaysia 8 11 Source: WHO and UNICEF. 6 In addition, it is likely that there are as many still births, which are uncounted in infant and child mortality rates, as deaths in the first seven days of life. ' These regional calculations, using data from 1996 DHS, makes the cautious assumption that low birth weight incidence was 15% for the set of mothers who were unable to provide information on low birth weight (Victora 2000). Data on low birth weight by income quintile in Brazil are not yet available. 8 1.13 A final reason for focusing on health care is that the Government of Brazil is examining policies to improve maternal and child health by improving the functioning of the health care system. The Cardoso administration's document of objectives for the health sector during the next several years (Plano Plurianual) emphasizes programs for maternal and child health, and it contends: "Reducing infant mortality remains a priority of the government. . . community health agents and family health teams will be fundamental in this work, given that they to a large extent provide care to the neediest communties, where infant mortality rates are highest." (Plano Plurianual) This study contributes to that effort by studying examining programs and the options available to the Government. Infant mortality: composition, levels and trends 1.14 This section and the next two draw heavily on a background paper prepared for this study (Victora 2000). That paper uses results from several censuses and surveys to develop indirect estimates of infant, child, and maternal mortality for Brazil and its regions. It then uses data from registered deaths to determine the causes of mortality among women and children. Although a comprehensive account of maternal and child health in Brazil would also include data on morbidity, morbidity statistics, particularly those disaggregated by region and cause, are notoriously hard to come by and often unreliable. It is the contention of this paper that, as a first cut, focusing on the causes of mortality provides a reasonable framework for selecting many, but not all, of the most important health care interventions, which are outlined in Chapter 2. 1.15 An additional caveat to the data below is that they are taken from registered deaths. It is possible that certain causes of death, such as diarrhea and acute respiratory infections, are higher among populations whose deaths are more likely to go unreported. Data taken from registered deaths would then underestimate proportionate mortality from those causes. (It is equally plausible that deaths due to perinatal causes are also higher among the poor and are as likely or more likely to go unreported). Several quality checks suggest that the data on causes of death are reasonable. In addition, the prominence of perinatal mortality in all regions of the country is consistent with trends around the world, as figure 3 below demonstrates. The background paper conducts sensitivity analyses for under-reporting of diarrhea and pneumonia. Still, this caveat should be kept in mind as the data below are presented. 8 This problem also makes clear the importance of good data for planning interventions to improve maternal and child health. The data on mortality, the causes of death, and morbidity in Brazil can be improved, and it would make sense of for the Government to conduct regular household surveys to assess current trends and evaluate the impact of its interventions. 8 Quality checks include the fact that death rates due to malformations should be constant across regions and income levels, and the data show that they are. In addition, the absolute values of the correlation coefficients between regional infant mortality rates and the regional causes of death, presented in the last column of table 7, are high and have the correct sign: the portion of child deaths due to perinatal mortality increases as infant mortality rates decline, for instance. Even in a country as poor as Bangladesh, data from the recent DHS suggest that neonatal mortality accounts for over 60% of deaths in the first year of life (Macro International, DHS Bangladesh, forthcoming). 9 Figure 3: Still births and Deaths per 1,000 Live Births in Developing Countries 100 - 80 \ 60- CO ~~40 -U 2-12 mfonths 0V 40 r-- |- l l 2*4 weeks O) 20 -_i* 1 st week (D ~~20 - 20 -co2 n 40 1983 1995 Source: WHO (1999). 1.16 Table 5 documents that the national infant mortality rate for Brazil was 37 deaths per 1,000 live births in 1995-7, meaning that at that time about 129,000 infants died each year. The table also shows that infant mortality rates fell about 40% from 1985-7 to 1995-7. Data from registered death certificates confirm the reduction in national infant mortality, declining from 128,745 annual deaths in 1985-7, to 75,990 a decade later, a drop of 41.0%. All regions of the country exhibited declines, ranging from 33.2% in the Northeast to 42.1% in the Southeast. Large regional differences persisted, however, with 1995-7 infant mortality rates in the Northeast still almost three times higher than in the South and Southeast. 1.17 The improvement over the past decade in Brazil appears to make up somewhat for the fact that the rate of decline in Brazil has been slower than for most countries in the hemisphere over the past 40 years. According to PAHO, Brazil's infant mortality rate fell 60-69% from 1950-5 to 1990-5. Of the countries studied, 23 showed faster declines than Brazil, and only nine had slower rates of decline (PAHO 1999). More recently, however, Brazil has been catching up: according to World Bank data, infant mortality rates from 1987-1997 declined 38% in Brazil, 10 33% in Latin America as a whole, and only 8 of 34 countries had steeper declines than Brazil over that period.9 1.18 The leading cause of registered infant deaths in Brazil are perinatal conditions, which accounted for 46.5% of infant mortality in 1985-7 and 56.8% in 1995-7. Proportionate mortality due to perinatal conditions rose in all regions, although the estimated national infant mortality rate due to perinatal causes fell from 29.0 to 21.3 per thousand. The rates are highest in the North/Northeast and lowest in the South/Southeast. 1.19 The second leading cause of registered infant deaths in 1995-7 were malformations, which overtook diarrhea and respiratory infections during the decade. These deaths are remarkably hard to prevent so that their relative weight tends to increase when mortality falls: for Brazil as a whole, the increase was from 7.1% to 11.2% of all infant deaths in the decade under study. The national infant mortality rate due to malformations remained stable, changing only from 4.4 and 4.2 per thousand. The apparent increase in the Northeast might well be the result of better data collection. The fact that malformations show little variability from one region to another is a good indicator of data consistency. 9 The convergence in infant mortality rates cannot be attributed exclusively to declining marginal returns: the Latin American country with second steepest decline from 1987-1997 was Cuba, where rates fell 46%, but which also started from the lowest base rate, 13.3 deaths per 1,000 live births in 1987. 11 Table 5: Infant Mortality Statistics for the Brazilian Regions, 1985-7, 1995-7 Region Year North Northeast Southeast South Center- Brazil Correla West -tiona Indirect IMR estimate 1985-7 58.9 90.6 44.6 37.7 40.8 62.4 1.00 1999-7 36.2 60.5 25.9 22.8 25.8 37.5 1.00 % reduction 38.6% 33.2% 42.1% 39.5% 36.8% 39.8% Percent of infant deaths according to cause Perinatal causesb 1985-7 42.1% 42.9% 48.9% 46.4% 48.7% 46.5% -0.72 ,61.4%L53.9°/eL _58.8% 13.0% 56.7%_To 56.8__ . -19j Malformationsb 1985-7 4.9% 3.6% 8.1% 11.4% 8.6% 7.1% -0.88 1995-7 8.5% 7.30 122.2 16.2%o, 13.4% 11.2%. -0.85 Respiratory infectionsb 1985-7 11.5% 11.4% 15.7% 14.2% 12.9% 13.9% -0.69 1995-i 8.9 0/- !9.7%-06 10.70% 1 .1% -8.7% l 10.2-%o -0 .27 Diarrheab 1985-7 30.1% 27.0% 11.8% 12.5% 12.9% 17.3% 0.78 1995 7 9.3% / 15.4% /_ 4.7%__,5.0% 6.4% 8.1 __/j 0.99 Other infectionsb 1985-7 5.3% 6.1% 5.3% 6.0% 7.1% 5.7% -0.12 1995-7, 5.6%. 6.0/o% .5.6 %o J 5.0, o 6.1 /o 0.45 Ill-defined causesc 1985-7 23.9% 45.5% 6.0% 11.0% 11.6% 23.0% 0.97 1995-7 _ J6.0%j 25.5% ; 5.0%o 3 6.3io 12.6% z 0.97; Estimated cause-specific IMR Perinatal causes 1985-7 24.8 38.9 21.8 17.5 19.9 29.0 _ 0.99 1995-7 22.2 r 32.6 F i5.2. 12.1 _ 14.6 ' 2l.3 i0.99 Malformations 1985-7 2.9 3.3 3.6 4.3 3.5 4.4 -0.54 1995 i 3.1 4.4 , 3.2 3.7 3.4 __ 4.2 _- 0.74 - Respiratory infections 1985-7 6.7 10.3 7.0 5.4 5.3 8.7 0.95 1995-:7' 3.2 S .9 .8 2.2..8 j 0.98 Diarrhea 1985-7 17.7 24.4 5.3 4.7 5.3 10.8 0.96 1995-7 _ 3.3 - 9.3 2_ 12 - Li _ 1.7 _ 3.0 1.00 Other infections 1985-7 3.1 5.5 2.4 2.3 2.9 3.5 0.97 1995-7 2.0 3.6 1 1.4 1.1 1.6 2.1 1.00 Ill-defined causes 1985-7 14.1 41.2 2.7 4.1 4.7 _ 14.3 0.98 11995-7' S.8 15.4 1.3 1.4 ___ 1.5 _ 4.7 1.00 Source: Victora (2000), indirect estimates based on surveys, censuses, and vital registration records. aPearson's correlation coefficient between the regional indirect IMR estimates and other mortality measures for the same three-year period. bExpressed as percentages of all deaths with a valid cause (excluding ill-defined causes). Because some valid causes of death are not listed (accidents, etc), these percentages will not sum to 100%. cExpressed as a percentage of all infant deaths. dObtained by multiplying the indirect IMR by the cause-specific percentages of deaths. 12 1.20 The third and fourth leading causes of infant deaths in the country were acute respiratory infections, largely pneumonia, and diarrhea, respectively. Proportionate mortality for both causes declined, with diarrhea falling from 17.3% of all deaths in 1985-7 to only 8.1% a decade later. The diarrhea specific infant mortality rate also fell by about two thirds. For both causes, the Northeast continued to have the highest rates. 1.21 Proportionate mortality due to other infections was stable, moving only from 5.6% in 1985-7 to 5.7% in 1995-7, but cause-specific infant mortality rates fell from 3.5 to 2.1 per thousand. Again, the infant mortality rate was particularly high in the Northeast. By far the most common cause in this group for 1996 was septicemia (59.4%), followed by meningitis (24.9%). It should be noted that septicemia is a complication of another infection, usually diarrhea or pneumonia, and - in the presence of good medical care and death certification - these deaths should have been attributed to the underlying disease. 1.22 Finally, table 5 shows the proportion of deaths due to ill-defined causes and the corresponding infant mortality rates. High proportions are indicative of poor quality medical services. Proportionate mortality fell from 23.0% in 1985-7 to 12.6% ten years later. Regional differences were marked and highly correlated to the regional indirect infant mortality rates. 1.23 Summarizing, infant mortality rates are higher in the North and Northeast regions of Brazil than elsewhere. With the exception of diarrhea, the main causes of infant mortality are more or less consistent across all regions: more than half are due to perinatal causes and about 10% are the result of respiratory infections. Infant mortality due to diarrhea fell considerably over the past decade in all regions, but it remains much higher in the North and Northeast. Child mortality: composition, levels, and trends 1.24 Table 6 shows that deaths among children between the ages of 1 and 4 accounted for 16% of the national under five mortality rate. Again, there are strong regional inequalities: the Northeastern rates are three times higher than those in the South/Southeast. The 1996 DHS sample did not include rural areas from the North region. 1.25 Time trends for the past decade are only available from registered death certificates because the only recent survey is the 1996 DHS, which provides mortality levels for around 1991. Data from registered death certificates shows a significant national decline in child mortality, from 23,761 annual deaths in 1985-7, to 12,966 a decade later, a 45.4% reduction. PAHO's analysis of child mortality rates from 1950-5 to 1990-5 finds that rates fell in Brazil more than 80%. Of the countries with available data, seven had faster declines than Brazil, and 16 were slower. 13 Table 6: Child (1-4 years) Mortality Statistics for the Brazilian Regions, 1985-7 and 1995-7: Indirect Estimates based on 1996 National Survey, and Vital Registration Statistics. Region Year North Northeast Southeast South Center- Brazil West Indirect IMR^ 1991 43 74 38 25 39 48 Indirect CMR (1-4 years)a 1991 10 16 6 5 8 9 Indirect underfive mortality 1991 52 89 43 29 46 57 ratea Ratio 1-4 to underfive^ 1991 19% 18% 13% 17% 17% 16% Percent of 1-4 year over all 1985-7 6.3% 5.2% 1.8% 2.0% 3.0% 3.0% registered deaths hs 91-5-7 Y2 =Fj 4 9% Percent of 1-4 year deaths according to cause Respiratory infectionsb 1985-7 22.8% 26.5% 26.0% 18.8% 17.8% 24.4% 1 1995-7 , 23.8%/ '27A. i 723.6% 6 , 21.3%i 17.9% D 23.8%' Other infectionsb 1985-7 18.9% 16.4% 13.8% 13.5% 14.6% 15.1% 19957 252% 22.55 22.8%_|, 21.3° 21.5% r 22.6% Injuriesb 1985-7 14.4% 12.5% 19.6% 22.9% 28.7% 17.9% 1995i 22.8-' 183°i e 22. i1: 238, % 28.6% e_ 21.9,; Diarrheab 1985-7 27.1% 21.9% 7.9% 10.5% 9.0% 14.6% L 1995-i i3.0%Mo 12.9% M _ l4.3% 5.6% 6.7% _ 78i%p Ill-defined causes" 1985-7 33.1% 52.2% 13.9% 15.7% 17.7% 33.3% '199-7 ---2-'27.1% 35%' 10i% 1 8.1p ,8.,3%° _ 18.6% _- Source: Victora (2000). aIndirect estimates based on National Demographic and Health Survey 1996 and vital registration statistics, estimates for the Northern region are restricted to urban areas. bExpressed as percentages of all deaths with a valid cause (excluding ill-defined causes). Because some valid causes of death are not listed, these percentages will not sum to 100%. CExpressed as a percentage of all infant deaths. 1.26 The leading cause of death was respiratory infection, mostly pneumonia, which accounted for one in four deaths both in 1985-7 and ten years later. Infections other than pneumonia or diarrhea were the second most common cause, increasing from 15.1% in 1985-7 to 22.6% of all deaths in 1995-7. The leading causes of death in this group, for 1996, were meningitis (38.5%), septicemia (34.2%) and HIV infections (7.2%). 1.27 Accidental injuries were the third most common cause of registered deaths. Proportionate mortality due to accidents increased in all regions because the other causes of death were falling. Car accidents (30.0%) and drowning (23.8%) were the most common accidents in 1996. As in infant mortality, deaths due to diarrhea fell sharply in the decade, going from 14.6% to 7.8% of all deaths. Important declines were observed in all regions, but disparities still persisted with the proportion in the Northeast being three times higher than in the Southeast. Finally, deaths due to 14 ill-defined causes decreased from one third of all deaths to 18.6% during the decade. Such deaths are much more frequent in the Northeast and North. 1.28 Summarizing, it is apparent that with the exceptions of death due to perinatal causes, which only applies to infant deaths, and accidental injuries, the principal causes of child mortality are also the major causes of infant mortality. Total child mortality is about three times higher in the Northeast than in the South/Southeast. Maternal mortality: composition, levels, and trends 1.29 Table 7 shows maternal mortality ratio estimates from a number of sources. All types of estimates have their limitations. (It should be noted that the Program in Women's Health of the Ministry of Health maintains reservations regarding all of these estimates, which rely on corrections for under-reporting, and therefore reports much lower maternal mortality ratios). Drawing on the most reliable sources, the current maternal mortality ratio for Brazil appears to be around 150-200 in the 1 990s. That estimate indicates that about 5,500 women die of maternal causes in Brazil each year. This number is about four times greater than the official number of registered deaths. Data from registered maternal deaths (not shown here) indicate a decline in maternal mortality from 1979 to 1992 but a stagnant rate since then, indicating either lack of progress or, less likely, improved registration. 1.30 The regional estimates that correct for under-reporting show considerably higher rates in the North and Northeast. The inequity in the maternal mortality rates between the North/Northeast and the South/Southeast appears to have decreased between the early 1980's and the current estimates. This cannot be attributed to the use of corrections, since these tend to inflate the maternal mortality rates more markedly in the North/Northeast. A possible explanation for this narrowing of the gap is discussed in Chapter 2. 1.31 Two other ways of looking at maternal mortality are also shown in Table 7. Maternal deaths account for 2.9% of all registered deaths of women aged 10-49 years, but this proportion is higher in the North and Northeast than in the rest of the country, about 4.8% and 4.2%, respectively. The ratio of direct to indirect obstetric causes is also indicative of the magnitude of the maternal mortality problem. Since direct causes are easier to prevent through adequate health services, high ratios indicate poor health systems performance. The ratios are over 11 in the North and Northeast, and around 3 in the South/Southeast. 1.32 The background paper calculated a detailed breakdown of the causes of maternal mortality in 1996-7 on the basis of tabulations provided by the Ministry of Health. About four in five registered maternal deaths are caused by direct obstetric causes. The leading group of direct causes are hypertensive disorders, including eclampsia/toxemia and related syndromes, that 15 account for 23.6% of registered maternal deaths. These are followed by sepsis (8.1%) and hemorrhage during childbirth and postpartum (8.0%). 1.33 Complications of abortion account for 7.4% of all maternal deaths. This could be an underestimate because doctors are reluctant to report the true cause of death in these circumstances, though, as Appendix A documents, deaths due to abortions are declining as women turn to pharmaceutically induced methods. Placental disorders, other complications of labor, embolism and abnormal uterine contractions each also cause a significant share of maternal deaths. Indirect causes, including pre-existing medical conditions - heart, lung, liver or other diseases aggravated by pregnancy - account for 21.3% of all deaths. Table 7: Maternal Mortality Ratio Estimates and Related Indicators for Brazilian Regions, 1980 to 1997, from Different Sources. Source and Region year of estimate Indicator North Northeast Southeast South Center- Brasil West Siqueira, 1980 MMR 338 229 83 92 144 154 Becker, 1984 MMR 313 155 70 81 121 120 Fonseca, 1988 MMR 164 64 58 46 65 63 Ministry of Health, MMR 36 38 48 53 34 44 1995 Ministry of Health, Hospital MMR 39 42 39 38 31 39 1996 DHS (Indirect MMR 217 sisterhood method), 1983-4 DHS (Direct sisterhood MMR 161 method),1983-96 Victora, 1995-97 MMR 186 219 117 129 145 147 % maternal causes' 4.8% 4.2% 2.2% 3.2% 2.3% 2.9% Direct/indirect ratiob 11.8 11.3 3.2 2.7 4.7 4.5 Estimated no. of deaths 635 2474 1499 594 338 5052 Source: Victora (2000). 'Percent of maternal deaths relative to all registered deaths with a known cause for women aged 10-49 years. No correction for under-reporting of maternal causes. bRatio of direct to indirect obstetric deaths, for registered deaths. No correction. Summary 1.34 The health of mothers and children in Brazil is not as good as it should be, given the country's income levels. A major reason for this is the concentration of illness and death among the poor: infant mortality rates jump 80-100% from the fourth income quintile to the lowest. Several interventions, such as strategies to promote productivity, increase economic opportunities, and enhance maternal education among the poor, could help reduce these 16 inequalities. This report focuses on the health care system. The health care system is particularly important for enhancing maternal and perinatal health, and certain health care interventions, such as immunizations, oral rehydration therapy, and case management, can reduce infant and child mortality rates due to infectious and communicable diseases. Data from registered deaths show that nearly half of all deaths among children under five occur in late pregnancy or as a result of complications in the first week of life. Apart from perinatal causes, the causes of mortality for children from ages 1-4 resemble those for infant mortality. Maternal mortality ratios, like those for infants and children, are significantly higher in the North and Northeast. 1.35 The most recent (1995-7) indirect estimate of the infant mortality rate in Brazil was 37.5 deaths per 1,000 live births. The rate has fallen about 5% a year over the past ten years. Maintaining that rate in the following years will be challenging because reducing mortality due to perinatal causes requires complex changes in health care delivery. But if Brazil continues on the social, economic, and political path it has been following, and if the country successfully implements measures to improve maternal and child health, such as those described in the next section, and focuses them on the poorest segments of the population, where they would have the largest impact, rates could continue to decline at the velocity at which they have fallen in the past ten years, about 5% a year, reaching 30.5 by the year 2000 and a target rate of 23.6 by 2005. On the assumption that without additional interventions the velocity of decline in infant mortality would slow to 3.5% per year, the measures described in the next sections would prevent approximately 20,000 infant deaths over the next five years. It would also prevent a large number of unnecessary maternal and child fatalities and reduce the morbidity burden on mothers and children significantly. 17 18 2. PRIORITIZING HEALTH INTERVENTIONS FOR IMPROVING MATERNAL AND CHILD HEALTH IN BRAZIL. 2.1 This chapter prioritizes the health sector interventions that could most improve maternal and child health in Brazil. It is largely based on a background paper (Victora 2000) that assesses declines in the relative risk of mortality that can be expected from a variety of interventions and then, on the basis of data on the causes of under five mortality, estimates deaths of children under five that could be averted by implementing the interventions. Tables 10 and 11, at the end of the chapter, summarize the estimates, the assumptions underlying them, and the likely feasibility of the proposed interventions both for Brazil and for its North and Northeast regions. Due to the paucity of studies, comparable estimates could not be constructed for maternal mortality ratios; but most experts agree that the interventions most likely to reduce perinatal mortality will also most reduce maternal mortality. The estimates of under five mortality impact are helpful but not determinative in prioritizing the interventions and for building the recommendations made in the last section of this chapter. The recommendations are based on the feasibility of the interventions, estimates of their potential mortality impact, and a recognition of the size of their potential morbidity impact (particularly for nutrition education) and social importance (particularly for family planning targeted at adolescents). Appendix A presents an expanded version of this chapter, structured around the life-cycle approach to maternal and child health, that describes best practices in the field. Interventions Targeted at Maternal and Perinatal Health 2.2 Improving Antenatal Care. Antenatal care can improve child health by detecting and treating maternal diseases, making sure mothers receive adequate nutrition, detecting and treating complications of pregnancy, providing health information on the risks of smoking and drinking and on the importance of postnatal care, and preparing for delivery. For pregnant women, antenatal care is critical for reducing deaths due to hypertensive disorders, hemorrhage, anemia, sepsis, and indirect obstetric deaths. 2.3 Data from the 1996 DHS show that while 4.9% of women in the South of Brazil failed to attend any antenatal care (ANC) sessions, 25.2% failed to do so in the Northeast. In the lowest income quintile, 32.5% of women did not attend any ANC sessions. (See Tables 8 and 9). Studies that correct for confounding variables show that mothers with fewer than five antenatal sessions in Brazil suffer about 2.5 times greater risk for neonatal mortality than mothers with more than five sessions. Combining these findings, the simulations estimate that if the percentage of mothers in Brazil with fewer than five ANC sessions could be halved (from 31.9% to 16.0%), the intervention would theoretically prevent 16.2% of perinatal-cause deaths and 7.8% of deaths among children under five, which are substantial impacts. There are very few 19 studies regarding the quality of antenatal care in Brazil, but most observers agree that quality improvements could also significantly improve maternal and child health. 2.4 Appropriate antenatal care can also help to address many of the risk factors associated with low birth weight among infants, including genital infections, uterine dysfunction, maternal heavy work, low energy intake in late pregnancy, and maternal hypertension. An additional intervention might target low birth weight directly with methods such as reducing cigarette smoking and enhancing maternal nutrition before and during pregnancy. Simulations found that reducing the incidence of low birth weight babies in Brazil from the current rate of 8.5% to 7.0% would theoretically prevent 6. 1% of under five deaths. Such a reduction is not feasible, however, because it is very hard to improve birth weight when the baseline levels are already low, as they are in Brazil even for the Northeast, where the figure is 9.0% even after correcting for under- reporting. International studies on the efficacy of smoking cessation programs, for example, find marginal, statistically insignificant impact on very low birth weight and perinatal mortality. 2.5 Another specific interventions for reducing low birth weight incidence would target the body-mass index (BMI) and the height of mothers and potential mothers. But the data show that male and female growth is improving in all regions of the country, and the percentage of mothers with low BMI does not decline monotonically with income (the percentage is actually lower in the fourth income quintile than the highest). Population based interventions would be difficult in this area both because adult height is determined early in life - interventions would therefore take a long time to have an impact - and because Brazilian food supplementation programs historically have not had good results. This analysis does eliminate the potential value of interventions that target pregnant women for micronutrient supplements such as folic acid, iron, and vitamin A, which studies have shown can improve maternal and child health. 2.6 Pregnancy Termination and Improving Post-abortion Care. There are an estimated 1.0-1.4 million abortions each year in Brazil, or about a third of the number of births. Abortion is legal in Brazil when the health of the mother is threatened, and recent legislation requires government health facilities to provide legal and safe abortion to mothers who need it. (Correa and others 1998) The number of hospitals in Brazil that offer the service increased from seven in 1997 to twenty-eight in 1999, but many states still do not have any hospitals that provide legal abortion. These changes, and especially the availability of the abortifacient Cytotec on the black market, have reduced the incidence of induced abortion complications from 1 per 5 deliveries in 1992 to 1 per 7.4 deliveries in 1997 (Faundes 2000). Still, unsafe abortion accounts for 9% of maternal deaths and 25% of the cases of infertility, and it is the fifth leading cause of hospitalization among women (RNFSDR 1999). 2.7 Data do not exist to estimate the health benefits of expanding safe and legal abortion or of improving post-abortion care, but providing such services and care clearly would benefit maternal health. In addition, doctors providing post-abortion care in Brazil generally do not address the contraceptive needs of their patients - remedying that failure would also improve maternal health at relatively low cost. 2.8 Improving Delivery Practices, Postpartum Care, and Neonatal Care. Over 90% of deliveries in Brazil take place in hospitals, but that figure falls as low as 82% in the North and 20 rises to 97% in the South. Similarly, although skilled attendants perform 87% of deliveries nationwide, regional disparities are sizable, with only 75% and 76.3% of births in the North and Northeast taking place under the guidance of skilled attendants, respectively, but over 93% in each of the other regions of the country. These regional disparities appear comparable to the disparities by income level: about 72% of mothers in the lowest income quintile gave birth with skilled attendants, compared to 89% in the second quintile and 96% in the third quintile. There are insufficient studies to estimate the impact on maternal or under five mortality if the number of births in hospitals and with skilled attendants could be increased in the North and Northeast regions, but accumulated knowledge worldwide suggests that the impact could be very large. As Chapters 1 and 2 argued, a functioning referral system that quickly diagnoses and treats complications during pregnancy is critical for maternal and perinatal health. In addition, numerous studies worldwide have demonstrated the efficacy of birth attendants trained in clean delivery practices; resuscitation of asphyxiated infants; thermal protection for newborn infants; and the early prevention, diagnosis, and treatment of infections. 2.9 The rate of cesarean section deliveries in Brazil is notoriously high: the most recent estimates place the figure at 36.4% for the nation as a whole and over 50% for the South, Southeast, and Center-West. The reasons for the high c-section rate in Brazil include the fact there are insufficient obstetric nurses in Brazil, leaving doctors, who prefer quick deliveries, to watch over births, that women have been taught that c-sections carry no additional risks, that low-income women imitate high-income in their preference for c-sections, believing they are a sign of good medical care, and that women often used c-sections to obtain tubal ligations covertly, which until recently were illegal. Patients and doctors have circumvented recent efforts on the part of the Ministry of Health to reduce c-section deliveries by creating facility-based caps on the number of reimbursable c-sections, and the rates remain stubbornly high. The WHO recommends that no more than 15% of births be delivered through a c-section, but there are few studies in Brazil that have been able to quantify the relative risk to the health of the mother or child of unnecessary c-sections. It is noteworthy that maternal mortality figures across Brazil's regions are less strongly associated with regional socioeconomic differences than are infant mortality rates - just the opposite of what is observed internationally - and it is conceivable that high c-section rates in the South, Southeast, and Center-West account for this. 2.10 Postpartum care in Brazil usually occurs at the facility where a woman gives birth - if she indeed gives birth in a facility -and it tends to be "surgical" in orientation. In general, it does not focus on health promotion and the well being of the mother and the newborn. Information and instruction on essential newborn care, such as resuscitation, thermal protection, and the promotion of breastfeeding, are not usually emphasized. Although the Ministry of Health, along with other partners such as UNICEF and PAHO, now promotes "Baby Friendly Hospitals," and some states, such as Pernambuco, have developed programs to promote breastfeeding and "the Kangaroo method" for the warmth retention, these are the exceptions rather than the rule. 2.11 Promoting Family Planning. While modem contraceptive use overall is 69% among married women or their partners in Brazil, regional and income-related disparities persist. The rates in the North and Northeast are 10-15% lower than in the other regions, and they are some 11-13% lower in the lowest income quintile compared to the second income quintile. The choice of method, moreover, is limited. Two-thirds of females using contraception opted for 21 sterilization, and only 7% of couples were using male-based methods. The current rates of adolescent fertility are also high and increasing. From 1993-97, the share of births among girls aged 10-14 increased from 0.9% to 1.2%, and the share among girls aged 15-19 increased from 21% to 25%. In poorer states as many as 50-60% of maternity beds can be occupied by girls under 19, and almost half (48%) of all legally sanctioned abortions in Brazil are performed on girls under 20 years of age. 2.12 The Brazilian literature on the risks of reproductive behavior for child health has focused on three risk factors: teenage pregnancies, short birth intervals, and high parity. Studies in Brazil find that while infants born to very young mothers (under age 15) might be at higher risk for mortality, infants born to mothers aged 18-20 suffer no greater risk once socioeconomic factors are controlled. Studies also show no statistically significant risk for children of higher birth order once confounding effects are controlled. But there is compelling evidence on spacing between births. Short birth intervals (less than 24 months) were associated with a relative mortality risk of 1.8 after controlling for confounding effects. A simulation exercise found that if the number of births occurring in short intervals could be cut in half (29.2% to 14.6%), the number of deaths among children under 5 would fall by 8.1%. 2.13 In recent decades Brazil has experienced one of the fastest fertility declines on record, with the total fertility rate falling from 5.8 in 1970 to 2.3 in 1996. So for the population as a whole new family planning programs might not be necessary: the government simply needs to sustain current activities while expanding the choice of methods available. For the poor, however, and for young women, undesired births remain a problem. According to the 1996 DHS, the total fertility rate was 78% higher among women aged 15-49 in the lowest income quintile than in the second income quintile. Most strikingly, this difference remained a problem even among more recent mothers: among women aged 15-19, the age-specific fertility rate was 62% higher in the lowest income quintile than in the second income quintile. This suggests that the importance of focusing and strengthening family planning interventions among low-income and young populations. To be successful, such interventions need to include men and women. Promoting family planning will both reduce infant mortality by improving birth spacing and will enhance the social, educational, and economic opportunities available to young women. Interventions to Promote Infant and Child Health 2.14 Improving case management of illnesses. Apart from perinatal conditions and malformations, the leading causes of death among children under five in Brazil were acute respiratory infections (largely pneumonia) and diarrhea. Effective treatments are available for both: virtually every pneumonia death can be prevented with prompt and appropriate antibiotic treatment, and most diarrhea can be treated at home with oral rehydration therapy (ORT). Pneumonia treatment requires that caretakers bring symptomatic children to health facilities, and that health care workers at the facilities identify the children who have life-threatening pneumonia and treat them with suitable antibiotics. Effective case management, in other words, requires both increasing awareness among caregivers and increasing the accessibility and effectiveness of diagnosis and medication. The self-reported incidence of acute respiratory infection was nearly constant across income quintiles, but the likelihood of obtaining treatment conditional on being sick increased with income. A simulation found that if 80% of children who 22 have pneumonia were taken to health care facilities, not an unreasonable assumption for Brazil, under five mortality would decline 3.9%. The use of oral rehydration therapy during diarrheal episodes has been a success story in Brazilian public health, with usage growing from about 24% in the Northeast in 1990 to nearly three-quarters in every region except the South in 1996. ORT use conditional upon having diarrhea was fairly constant across the income quintiles although, unsurprisingly, diarrheal incidence was highest for the lowest quintile, about 30% higher than for the second quintile. Treatment for severe diarrhea requires professional treatment at a health facility. A simulation found that if 80% of children with severe diarrhea were brought to a health facility, 5. 1% of under five deaths in Brazil could be prevented. 2.15 Other diseases contributing to under five mortality include meningitis, septicemia, HIV/AIDS, malaria, viral hepatitis, and congenital syphilis. A simulation making a variety of assumptions regarding this category of "other diseases" concluded that appropriate case management could prevent 20% of deaths attributable to them, saving 1.6% of deaths among children under five. 2.16 Interventions for HIV/AIDS should receive special attention in Brazil because its incidence is rising. In addition, repeated episodes of malaria among children can be damaging even if they are not fatal, leading to anemia that affects school performance and productivity in adult life. Brazil's malaria program has been successful in recent years, and interventions that incorporate malaria diagnosis and treatment into routine health care in endemic areas should be continued. 2.17 Promoting Breastfeeding. Breastfeeding protects against diarrhea, pneumonia, and other infections by providing infants with maternal antibodies and antimicrobial substances. It also ensures adequate nutrition early in life and promotes more adequate birth spacing among mothers. Because of these pervasive benefits, international recommendations call for exclusive breastfeeding during the first 4-6 months of life and continued breastfeeding through 24 months and beyond. Brazil falls well short of those recommendations, with median rates for exclusive breastfeeding of about 1 month (0.7 months in the Northeast) and a median duration for any breastfeeding of 7 months. Only 17% of Brazilian children are breastfed up to 20-23 months, compared with 53% in developing countries as a whole. Using figures from international meta- analyses that found relative risks of mortality among children who are not breastfed of 2.5 for pneumonia and other infections, 6.1 for diarrhea for children less than six months, 1.9 for children 6-1 1 months, a simulation found that at every age from birth to 11 months, if half of Brazilian children who are not currently breastfed were to become so, deaths among children under five would fall 4.0%. This is an achievable goal in Brazil, as the previous success of breastfeeding campaigns, some based on television soap opera promotions, in some parts of the country have demonstrated. 2.18 Extending Immunizations. Complete immunization with the six basic childhood vaccines covered 72.5% of all Brazilian children in 1996. Its recent successes are illustrated by the decline in deaths due to measles from 1433 in 1979 to 17 in 1997. Regional rates for a complete series varied from as low as 60.7% and 63.3% in the Northeast and North, respectively, to 87.1% in the South. Complete immunization among the lowest income quintile was 56.6%, compared to 74.0% among the second income quintile. Deaths from these six vaccine 23 preventable diseases are low. These figures illustrate that Brazil needs to maintain the success of the current program while trying to expand coverage among the poor and in the Northeast. Brazil recently introduced the Hib and hepatitis vaccines into its basic immunization program. These will reduce deaths due to pneumonia and meningitis. Simulations found that if coverage rates for the Hib vaccine reach 80%, under five mortality would decline 2.6%. An additional simulation found that if Brazil decides to introduce the rotavirus vaccine and reaches coverage of 80%, 7.3% of deaths due to diarrhea would be avoided, equivalent to 0.6% of deaths among children under five. Given the existing infrastructure for vaccine delivery, these are highly feasible targets. 2.19 Improving Nutrition. National nutrition surveys indicate that underweight prevalence (low weight for age) in Brazil fell 60% between 1975 and 1989 and a further 20% from 1989 to 1996. One study suggests that the reasons for the improvements include modest increases in family income and an expansion of health, education, and water supply services, all of which were facilitated by falling family size and urbanization. Still, there remain sharp regional variations in measures of nutritional intake. Stunting (low height for age) and underweight prevalence were two to three times higher in the North and Northeast than in the other regions. Data disaggregated by income level finds that children in the lowest income quintile were more than twice as likely to be underweight, six times as likely to be severely underweight, and nearly three times as likely to be stunted than children in the second income quintile. 2.20 Simulations found that the mortality impact of improved nutrition among children in Brazil would be small: assuming that the difference between current rates of underweight prevalence and that observed in a well-nourished population (2.3%) could be halved, under five deaths would decline 0.6% (0.5% through the effect on pneumonia plus 0.1% through the effect on diarrhea). Still, because improved nutrition can have significant and positive effects on cognitive development, motor skills, and learning, effective interventions that improve nutritional status, particularly among poor children, are warranted. 2.21 Precisely what constitutes effective nutrition interventions is controversial in Brazil. There is evidence that supplementary feeding programs have expended a great deal of resources over the years without much impact on nutritional status. Nutrition education programs, on the other hand, have been more successful both internationally and in preliminary studies of the nutrition component of WHO's Integrated Management of Childhood Illness in Brazil. Growth monitoring and promotion are also controversial. While these programs have been effective in some settings internationally, there is no solid evidence from Brazil that they have improved nutritional status. 2.22 Micro-nutrient supplements, including iodine, iron, vitamin A, and zinc can reduce the risks of mortality and morbidity among children. The best evidence from Brazil demonstrates a 20% reduction in severe diarrhea associated with the intake of vitamin A supplements. A simulation assumed that half of deaths due to diarrhea in the Northeast are in the vitamin A- deficient, semi-arid areas and that 80% of such children were to take vitamin A every 4-6 months. It found a reduction of 8.6% of deaths due to diarrhea, or about 0.7% of all deaths among children under five. 24 2.23 Improving Water and Sanitation. Both water quality and quantity are associated with fewer childhood deaths due to diarrhea. While the number of homes with treated water supply in Brazil was 74.2% for the country as a whole in 1996, it was 59.7% and 56.2% in the North and Northeast, respectively, precisely those areas with higher rates of death due to diarrhea. One study from Brazil found that the relative risk for diarrhea mortality is 3.9 times greater in homes without piped water. The study found no increased relative risk for homes without sanitation facilities once it adjusted for water supply and socioeconomic status. A simulation using that relative risk figure found that if half of the houses that are currently not hooked up to treated water were to become so, 21.4% of diarrheal deaths could be avoided, or 1.7% of all deaths among children under five (including 3.8% of all under five deaths in the Northeast). 2.24 Improvements in water supply can also prevent deaths from a variety of other diseases, such as hookworm and schistosomiasis, but these are not significant causes of death in Brazil. Though water hookups are costly, their health benefits can be pervasive, and their impact should not be assessed on their contribution to under five mortality reduction alone. 2.25 As Chapter 1 pointed out, there is evidence that some of the impact from water hookups on child mortality is the resdlt of better community information and awareness regarding hygienic practices. That is an argument for promoting hygienic practices among Brazilian families directly. There is evidence that the promotion of hygienic practices can reduce diarrhea morbidity, but there were insufficient data to conduct a simulation of their impact on under five mortality. 2.26 Preventing Injuries. Car accidents and drowning are the most common causes of accidental death among children in Brazil. Other reported accidents among women and children could be related to domestic violence, but the available data are sparse and likely to be low estimates due to under-reporting. Preventing injuries is complex, requiring joint efforts from legislators, law enforcement, educators, and the society at large. A simulation found that if 30% of injury deaths could be prevented, an ambitious goal, overall deaths among children under five would decline 1.5%. National and Regional Priorities 2.27 The data presented above make clear that mortality rates and other indicators of maternal and childhood health are consistently higher in the North and Northeast regions, which are characterized by a larger share of poverty and greater rural populations than the rest of the country. It is also evident from the data that the health of the poorest segment of the population is significantly worse than the health of even the nearest income group. It is also true, with a few exceptions, that the current level coverage of health interventions are lower among the poor, in the North and Northeast, and in rural areas (which are largely but not entirely overlapping sub- populations). Table 11 shows that the estimated impact for the interventions mentioned above are larger for the North and Northeast regions than for the country as a whole, which is unsurprising since risk factor prevalences are higher there. The table demonstrates that implementing WHO's package of interventions known as the Integrated Management of Childhood Illness (including the case management of illness, breastfeeding promotion, nutrition counseling, and providing 25 vitamin A supplementation) could theoretically prevent as many as 20% of under five deaths.'t Because fertility rates, particularly adolescent fertility rates, continue to be high among the poor, family planning interventions should also be a priority. Improved birth spacing could potentially save 10% of under five deaths in the North and Northeast. In addition, inter-sectoral initiatives that include education, water, and poverty reduction, such as the Government's National Program for Reducing Infant Mortality, which targets poor municipalities largely in the Northeast, Bolsa Escola, which rewards families who keep their children in school, as well as successful public works and income-generating activities, should be maintained and expanded. 2.28 The other regions of the country have successfully reduced the incidence of diarrhea, pneumonia, and measles. To maintain these achievements, interventions in immunization, family planning, and diarrhea management should be sustained. Because perinatal mortality has grown in proportional importance and because maternal mortality ratios have been stagnant for the past decade, further improvements in maternal and child health will largely come from improvements in antenatal, delivery, neonatal, and postpartum care. The estimates show that reducing by half the percentage of women who receive less than five antenatal care sessions could reduce prevent about 8% of under five deaths. Quality improvements in these areas will also have substantial impact on maternal mortality and perinatal deaths, which constitute nearly half of all deaths of children under five. Improvements in antenatal, delivery, neonatal, and postpartum care would also have a very large impact in the North and Northeast, where the cause-specific perinatal mortality rates were two to three times higher than in other parts of the country and where perinatal mortality has declined much more slowly over the past decade than deaths due to diarrhea and pneumonia. Other priorities for the country as a whole are breastfeeding promotion and the introduction of the Hib vaccine. 2.29 Summarizing, these estimates show that in the country as a whole interventions in the case management of illness, breastfeeding promotion, family planning to improve birth spacing, improvements in the coverage of antenatal care, and the introduction of the Hib vaccine could jointly avert as many as 33% of under five deaths. Improvements in the quality of delivery, postpartum, and neonatal care would also have a substantial impact on mortality and morbidity rates among mothers and children. It is plausible, therefore, that implementing all of these measures over five years could cut the infant mortality rate by at least one-fourth, from 30.5 in the year 2000 to 23.6 in the year 2005, a decline of 5% a year over the next five years. Some states, including Rio Grande do Sul and Brasilia, already have estimated rates near or below 20, so the target is feasible for Brazil. 10 It should be recalled that the estimates of potential lives saved for the interventions are not strictly additive, but they are useful for developing priorities. 26 Table 8: Distribution of Indicators Relevant to Possible Interventions against Underfive Mortality, by Region, Brazil, circa 1996 Year Region and North Northeast Southeast South Center- Brazil source West ANTENATAL CARE Did not attend antenatal care 1996a 17.1% 25.2% 6.1% 4.9% 7.0% 13.2% Antenatal care starting first trimester 1996a 55.7% 51.9% 74.2% 79.7% 71.7% 66.0% Median number of antenatal attendances 1996a 6.3 6.4 8.1 8.3 7.6 7.4 Received 2 or more doses of tetanus toxoid 1996a 51.0% 49.6% 38.2% 47.1% 53.7% 45.3% Less than 5 antenatal attendancesb 1996a 44.5% 49.3% 20.8% 18.0% 24.7% 31.9% DELIVERY PRACTICES Hospital deliveries 1996a 81.9% 83.4% 97.0% 97.4% 97.1% 91.5% Deliveries by doctors 1996a 55.1% 57.4% 92.7% 87.6% 92.0% 77.6% Deliveries by skilled attendants 1996a 75.0% 76.3% 96.1% 93.1% 96.4% 87.7% C-section rate 1996a 25.5% 20.4% 47.2% 44.6% 49.1% 36.4% MATERNAL NUTRITION Mean matemal height (cm) 1996a 154.4 154.7 157.5 157.9 156.6 156.3 Mean maternal body mass index (kg/M2) 1996a 23.0 23.4 24.5 24.8 23.6 24.0 Maternal BMI < 18.5 kg/m2 1996 5.9% 7.1% 6.6% 2.7% 8.1% 6.3% BIRTHWEIGHT Reported prevalence of birthweight <2500 g 1996a 7.4% 7.4% 8.9% 7.6% 9.1% 8.1% Birthweight information not available 1996a 14.9% 21.2% 5.5% 2.4% 5.7% 5.7% Hospital birthweights <2500 g 1997d 6.4% 7.0% 8.7% 7.9% 7.1% 7.8% Corrected low birthweight estimatee 1996a 8.5% 9.0% 9.2% 7.8% 9.4% 8.5% FAMILY PLANNING Ever pregnant adolescents (15-19 years) 1996a 23.5% 20.6% 16.2% 16.2% 17.0% 18.1% Birth interval less than 24 mos 1996a 33.3% 37.7% 25.6% 18.1% 22.4% 29.2% Contraceptive use rate among married women 1996a 72.3% 68.2% 79.5% 80.3% 84.5% 76.7% Total fertilityrate 1996a 2.7 3.1 2.2 2.3 2.3 2.5 ARI CASE MANAGEMENT Utilization of medical care during ARI 1996a 14.1% 15.0% 20.4% 20.2% 19.9% 18.2% CROWDING Mean number of persons per bedroom 1996c 2.4 2.1 2.1 1.9 2.0 2.1 CHILD NUTRITIONAL STATUS Stunting prevalence 1996a 16.2% 17.9% 5.3% 5.1% 8.2% 10.5% Wasting prevalence 1996a 1.2% 2.8% 2.4% 0.9% 2.9% 2.3% Underweight prevalence 1996a 7.7% 8.3% 4.8% 2.0% 3.0% 5.7% PROMOTING BREASTFEEDING Median breastfeeding duration (mo) 1996a 10.3 7.0 6.3 7.1 9.8 7.0 Median exclusive breastfeeding duration (mo) 1996a 1.4 0.7 1.3 1.7 0.7 1.1 IMMUNIZATIONS Fully vaccinated (children 12-23 mo) 1996a 63.3% 60.7% 77.5% 87.1% 76.2% 72.5% DIARRHEA CASE MANAGEMENT Received ORT during diarrhea 1996a 75.4% 74.2% 73.2% 67.9% 80.8% 73.4% Utilization of medical care during diarrhea 1996a 33.3% 27.4% 30.7% 49.6% 36.6% 32.0% VITAMIN A SUPPLEMENTATION Received vitamin A during last 6 mos 1996a 21.5% 40.5% 11.3% 11.1% 8.4% 21.6% 27 WATER AND SANITATION Treated water supply 1996c 59.7% 56.2% 86.5% 77.0% 65.5% 74.2% Adequate sanitation facilities 1996c 48.6% 37.7% 82.9% 66.6% 26.3% 63.6% aSource: National Demographic and Health Survey 1996 (North region: urban areas only). bCalculated from the original figures by adding all mothers with no attendances, those with 1-3 attendances and one third of those with 4-6 attendances. cSource: National Household Sample Survey 1996 (North region: urban areas only). dNational Birth Registration System (SINASC), Ministry of Health. 'Based on 1996 DHS results but assuming that LBW prevalence was 15% when mother was not able to provide this information. 28 Table 9: Maternal and Child Health by Income Quintile, Brazil, DHS 1996 Indicator Summary Definition Income Quintiles Lowest Second Middle Fourth Highest Total HNP Status Indicators Infant Mortality Rate Deaths under age 12 months 83.2 46.7 32.9 24.7 28.6 48.1 per thousand births Under Five Mortality Rate Deaths under 5 years per 98.9 56.0 39.2 26.7 33.3 56.7 thousand births Children Stunted (%) Below -2 sd z-score, height 23.2 8.7 5 3.9 2.3 10.5 for age, children under 5 years Children Underweight (°) Below -2 sd z-score, weight 11.5 :5.1 2.8 1.9 3 5.7 for age, children under 5 years Children Underweight (%) Below -3 sd z-score, weight 1.2 0.2 0.5 0.7 0.3 0.6 for age, children under 5 years Low Mother's BMI (%) Body Mass Index < 18.5 8.8 4.6 6.8 5.3 5.4 6.3 Total Fertility Rate Births per woman age 15-49 4.8 2.7 2.1 1.9 1.7 2.5 Age-Specific Fertility Rate, Births per 1000 women age 15-19 176.0 109.0 70.0 57.0 28.0 86.0 15-19 HINP Service Indicators Immunization coverage Children age 12-23 months, by (%): vaccination card or mother's report -- Measles 77.9 86.8 96.0 91.3 90.2 87.2 -- DPT3 65.6 83.9 90.4 91.3 81.9 80.8 -- All 56.6 74.0 84.9 83.1 73.8 72.5 --None 6.4 2.6 0.4 1.3 3.0 3.1 Treatment of Diarrhea --Prevalence % Ill in the preceding 2 weeks 18.3 12.9 12.7 9.3 7.4 13.1 -- ORT use ORS, RHF, or increased liquids 73.1 73.6 76.9 74.3 (65.8) 73.4 -- Seen Medically Brought to a health facility if ill 27.5 26.7 38.9 32.8 (51.8) 32.0 --% Seen in a Public Facility Among those medically treated 23.4 23.8 30.8 14.6 (25.3) 23.9 Treatment of Acute Respiratory Infection (%): --Prevalence % Ill in the preceding 2 weeks 25.3 26.7 23.0 20.6 20.5 23.7 -- Seen Medically Brought to a health facility if ill 33.4 47.4 47.6 52.6 65.1 46.1 --% Seen in a Public Facility Among those medically treated 31.1 42.2 32.0 36.2 27.0 34.3 .29 Antenatal Care Visits (%): -- to a Medically Trained Doctor, nurse, or nurse-midwife 67.5 87.7 93.4 96.9 98.1 85.6 Person --to a Doctor 58.1 83.8 91.6 95.7 98.1 81.4 -- to a Nurse or Trained Nurses and nurse-midwives 9.4 4.0 1.8 1.1 0.0 4.2 Midwife -- 2+ visits 64.2 83.9 90.5 95.0 97.2 82.8 Delivery Attendance (%): -- by a Medically Trained Doctor, nurse, or nurse-midwife 71.6 88.7 95.7 97.7 98.6 87.7 Person -- by a Doctor 52.5 77.5 89.2 94.3 97.6 77.6 --by aNurse or Trained Nurses and nurse-midwives 19.1 11.2 6.5 3.4 1.0 10.0 Midwife -- % in a Public Facility 75.9 88.3 86.6 72.2 61.3 77.9 -- % in a Private Facility 2.1 6.6 11.0 26.5 37.8 13.6 % at Home 20.0 3.6 0.9 0.3 0.0 7.0 Use of Modern Currently married persons using Contraception (°): a modem method --Females 55.8 68.9 73.6 73.8 76.8 70.3 -- Males 52.8 66 72.6 70.3 77.8 68.6 Knowledge of HIV/AIDS Knows sexual transmission Prevention (%): routes of HIV/AIDS -- Females 65.9 82.6 90.3 91.3 93.7 85.8 --Males 72.4 85.7 90.4 93.5 93.9 87.7 Source: Gwatkin and others (2000). 30 Table 10: Potential Impact and Feasibility of Interventions against under five Mortality in Brazil Intervention Assumptions Under-5 deaths likely to Current levels and trends in risk Feasibility of intervention be prevented (°) factor or intervention Improving antenatal Reducing by half the current The mean number of sessions is Medium Would require public education as care proportion (31.9%) of 7.8% adequate but many high-risk women well as improving accessibility. women with less than 5 have too few sessions sessions Improving the quality of Little information is available but Medium-high. Would require training antenatal care Potentially large average quality appears to be poor health workers and providing drugs and equipmenL Improving delivery Increasing the proportion of 86% of all births are already assisted by a Medium-low. Involves accessibility problems care deliveries by a trained Low to medium doctor or nurse, and this rate has been in rural areas as well as training of staff. professional increasing steadily. Improving the quality of The high perinatal mortality of babies Medium-high. Would require training and delivery care in hospitals Potentially very large with appropriate birth weight suggests providing equipmenL that quality is poor Avoiding unnecessary 36% of all deliveries are by a c-section, Low. Social, cultural and economic factors are cesarean sections Low and this is still rising. But there is no responsible for the high rate. strong effect on infant mortality. Improving birth weight Current rate of 8.5% could be Current levels already low. Time trends Low. Interventions against low birth weight lowered to 7.0% 7.1% stable in the South and Southeast; no have limited efficacy unless adult malnutrition (developed countries rate is information for other regions is highly prevalent. currently 6.0%) Improving matemal Improving maternal body Secular trend in growth is present in all Low. Nearly all interventions are ineffective nutrition mass index and improving Low regions and women tend to be larger. and may lead to obesity. maternal height. Improving height requires intervention to start in childhood. Reducing maternal Reducing smoking from the 27% of women smoke and this rate is not Medium-high according to the intemational smoking current rate of 27%. Low being reduced. No clear association with experience. infant mortality. Improving Ensuring FP counseling and Low Gradual increase High. postabortion care provision at all n postabortion care I 31 Intervention Assumptions Under-5 deaths likely Current levels and trends in risk Feasibility of intervention to be prevented (%) factor or intervention Promoting Preventing teenage 14% of women aged 15-19 are or High, but there may be limited needfor family pregnancies Low or none have been pregnant; the trend is extra investment since Brazil is planning rising. But there is no clear experiencing one of the most dramatic association with infant mortality. fertility declines in the world. Contraceptive use rates are high and Reducing by 50% the Proportion of birth intervals <24 still rising. proportion of short birth 8.1% months feUlfrom 46% to 29% from intervals 1986-96. Reducing the proportion of Family sizes are dropping rapidly. high-parity women Low or none There is no clear association with infant mortality. Improved Ensuring adequate case- 3.9% (pneumonia) The number of health workers Medium-high. Requires health worker case- management and 80% +5.1% (diarrhea) (doctors, nurses, and community training and drug supply. management care-seeking ratefor +1.6% (other infections) health workers) is increasing severe episodes Total = 10.6% steadily in BraziL Reducing Reducing mean number of The reduction in fertility is likely to Low for irnproving housing. crowding persons per bedroom from 1.6% lead to less crowding. High for reducing family size but as the current level (2.1) to noted above there is already a strong 1.5 trend in this direction. Improving Reducing underweight 0.5% (pneumonia) Nutritional status of Brazilian Low for supplementary feeding nutritional prevalence from the +0.1 (diarrhea) children is improving in all regions. programs; no evidence of impact of status current level (5.7%) to Total = 0.6% growth monitoring. 4.0% Promoting Reducing by halthe 1.1% (pneumonia) + Breastfeeding duration is High, as breastfeeding counseling breastfeeding fraction of infants, at any 1.6% (diarrhea) + 1.3% increasing in Brazil but is still well activities have had a clear effect in given age in months, who (other infections) short of international severalparts of the country. are not breastfed Total = 4.0% recommendations. Introducing Reaching 80% coverage 2.0% (pneumonia) Current coverage is practically zero. Very high. The national program of Hib vaccine with Hib vaccine +0.6% (meningitis) Immunizations is very successful in Total = 2.6% terms of existing vaccines. 32 Intervention Assumptions Under-5 deaths likely Current levels and trends in risk Feasibility of intervention to be prevented (%) factor or intervention Preventing Improving antenatal care Current quality of antenatal care Medium-high for improving ANC: malformations quality to prevent and treat Low appears to be low. requires training health staff and risk factors for providing access to laboratory facilities. malformations But malformations have multiple etiologies, each contributing to a small number of deaths. Vitamin A Reaching 80% coverage of 40% of children in the Northeast High. Supplements can be distributed on supplemen- regular vitamin A 0.7% have received supplements at some vaccination days. tation supplementation in high- time in their lives. risk areas in the Northeast Introducing Reaching 80% coverage of The vaccine has not been introduced Very high. The Brazilian experience with rotavirus rotavirus vaccine 0.6% in Brazil immunizations is highly positive. vaccine Improving Reducing by half the About one in each four homes does Low-medium. Investments in improving water and proportion of homes 1.7% not have treated water, but this water and sanitation bring multiple sanitation without treated water proportion is being reduced. benefits but are extremely costly. Preventing Reducing by 30% the No information is available at Low-medium. Involves multiple sectors injuries number of deaths due to 1.5% regional level on risk factors for of society including legislation, law injuries injuries. enforcement, engineering and health. Source: Victora (2000). Note: Recommended priority interventions are in bold italics. 33 Table 11: Potential Impact of Interventions against under five Mortality in the North and Northeast Regions (see Table 10 for feasibility issues). Intervention Assumptions Under-5 deaths likely to Comments be prevented (%) Improving Reducing by half the current antenatal care proportion (48.2%) of women 10.0% with less than 5 sessions Improving the quality of Potentially large antenatal care improving Increasing the proportion of Potentially large Large potential impact delivery care deliveries by a skilled professional due to the high proportion of home deliveries Improving the quality of delivery Potentially very large care in hospitals Avoiding unnecessary cesarean Low sections Improving birth Current rate of 8.9% could be This is a rather weight lowered to 7.0% 8.9% ambitious goal given the (developed countries rate is effectiveness of LBW currently 6.0%) prevention Improving Improving maternal body mass Low maternal nutrition index and improving maternal height. Reducing Reducing smoking from the Low maternal smoking current rate of 27%. Promotingfamily Preventing teenage pregnancies planning Low or none Reducing by 50% the proportion 9.7% of short birth intervals Reducing the proportion of high- Low parity women Improved case- Ensuring adequate case- 3.9% (pneumonia) management management and 80% care- +8.8% (diarrhea) seeking ratefor severe episodes +1 7% (other infections) Total = 14.4% Reducing Reducing mean number of persons crowding per bedroom from the current level 1.8% (2.2) to 1.5 34 Intervention Assumptions Under-5 deaths likely to Comments be prevented (%) Promoting Reducing by half the proportion 1.1% (pneumonia) breastfeeding of infants, at any given age in +2.8% (diarrhea) months, who are not breastfed +1.4% (other infections) Total = 5.3% Introducing Hib Reaching 80% coverage with Hib 1.9% (pneumonia) vaccine vaccine +0.6% (meningitis) Total = 2.5% Preventing Improving antenatal care quality to malformations prevent and treat risk factors for Low malformations Vitamin A Reaching 80% coverage of regular supplementation vitamin A supplementation in 1.5% high-risk areas in the Northeast Introducing Reaching 80% coverage of rotavirus vaccine rotavirus vaccine 1.0% Improving water Reducing by half the proportion of and sanitation homes without treated water 3.8% Preventing Reducing by 30% the number of injuries deaths due to injuries 1.1% Source: Victora (2000). Note: Recommended priority interventions are in bold italics. 35 36 3. HEALTH CARE AND HEALTH PROGRAMS FOR MATERNAL AND CHILD HEALTH IN BRAZIL The Sistema [nico da Saiide and Sector Issues 3.1 In accordance with the Constitution of 1988, the Government of Brazil offers universal health care free of charge. Health care professionals and facilities affiliated with the government's health care system (Sistema Unico da Satide - the SUS) provide services ranging in complexity and cost from simple palliative care and immunizations to organ transplants and anti-retroviral therapy for HIV/AIDS. Although the system, the product of a long struggle against the military regime and its sectoral policies, was intended to be the "unique" or "only" system of health care in the country, the financing and governance of health care has grown diverse and complicated in the decade or so since it was established. 3.2 Management responsibility for primary care has been transferred to the nation's five thousand five hundred municipalities. In theory, most municipal health secretariats now conduct immunization programs, implement community health outreach and education (including family planning and antenatal care), collect data on local morbidity and mortality, and provide basic care and medications. Depending on the human and financial resources available, municipalities can also provide dental care, psycho-social services, and testing in medical laboratories. In exchange for assuming management of primary care, municipalities receive monthly transfers (called the piso ambulatorial basico - the PAB) from the federal government of about R$10-18 per inhabitant. Municipalities receive additional monthly transfers from the federal government if they provide a few additional initiatives: the basic medications program (Farmaceutica Basica), the family health program (PSF) and the community health agents program (PACS), the program to combat nutritional deficiencies (PCCN), payments for mental health medications, and epidemiological surveillance. Brazil's twenty-seven state governments are, to varying degrees, involved in training and supervising municipal health secretariats and in coordinating activity among municipalities. The states often act as liaisons for information and resources flowing from the Ministry of Health in Brasilia to the municipalities. 3.3 The management and financing of hospital and higher level care is more complicated. Private providers, whether SUS-affiliated facilities (filantr6picas) or private hospitals that sign occasional agreements (convenios) with the SUS, play a large role: some 70% of hospital beds in the SUS are in private or philanthropic facilities. Because they cannot charge SUS patients, these private facilities and their doctors make claims for reimbursement to federal, state, or municipal governments, depending on the degree of autonomy granted to the local governments, and are paid according to a uniform national fee schedule (the tables associated with the AutorizaCdo da InternaCdo Hospitalar - AIH). States and municipalities granted autonomy receive funds from the federal government for paying private providers in their area, and they must pay the local 37 private providers at rates at least as high as those specified in the AIH tables. The federal government has transferred ownership of many public hospitals to the states and municipalities, but it has no plans to transfer several key facilities, including the teaching hospitals under the control of the Ministry of Education, to localities, nor do several state governments plan to transfer many of the remaining facilities under their control to municipalities. 3.4 Federal payments for hospital and higher level services to public, philanthropic, and private facilities and providers are based on uniform national fee schedules that blend prospective, diagnosis-related payments and elements of fee-for-service reimbursement. While the basic payment schedule was conceived to be prospective in order to transfer a degree of financial risk to the providers, it has been difficult to avoid compensating facilities on a fee-for- service basis for technologically intensive and specialized procedures. There now exists a special category of payments for "hospital and ambulatory care of medium and high complexity" that includes reimbursement for procedures such as emergency care, high risk pregnancy treatment, intensive care units, drugs for transplantation, breast and prostate surgery, as well as reimbursement to facilities and localities for treating patients outside of their municipal and state jurisdiction (cdmara de compensa,do) and, in 1998, for unusually low payments during the period 1994-98 (fator de recomposa,co). It has also been difficult to refrain from raising the fees paid for costly procedures more quickly than other procedures, which erodes the incentives of prospective payment; and some part of the increasing complexity of the case-mix observed in Brazil might be due to re-classification or "DRG creep." (Carter and others 1990) Because fiscal constraints, combined with inefficiencies, have forced the federal government to set AIH reimbursement prices an average of 50% below the estimated real cost of providing services, the government rations care by setting annual caps in all SUS-affiliated facilities for procedures and financing, resulting in global caps for each municipality and state. These caps (tetosfinanceiros), which almost all facilities reach every year, function essentially as annual budgets for each facility and are determined annually in a negotiated exercise (Programa Pactuado Integrado - PPI ). 3.5 The sources of federal financing for health care consist of social security contributions, corporate taxes, a tax on financial transactions, and other general taxes. Municipalities and states contribute 15% and 18% of their own resources to total health spending in the public sector, respectively (Medici 1998). In many cases, however, particularly in the 2,700 municipalities of less than 17,000 inhabitants, the amounts fall significantly below those averages. Recent legislation amended the Constitution to require municipalities and states to spend fixed percentages of their own resources on health care. 3.6 Privately financed health care includes a wide variety of indemnity insurers, in-house company health plans, managed care organizations, group practice associations, and direct out- of-pocket consumer payments. About 26% of Brazilians enjoy some third-party payment plan for health care expenses (Rose 1997). A recent survey found that although the SUS is accessible to all Brazilians regardless of income, 20% of the population did not utilize its services at all, preferring privately financed care outside the SUS network. The fraction of the population not utilizing the SUS rose to 48% among respondents with university degrees. Only 52% of the population considered itself exclusive or frequent users of the SUS, the remainder relying on either the private sector or their own resources for the majority of their medical care (IBOPE, cited in Medici 1998). 38 3.7 For the purposes of the present study aimed at improving maternal and child health in Brazil, five key sector issues deserve most attention: 3.8 Regional Inequalities in Health Expenditures. Numerous analyses have documented how low levels of health expenditure in the North and Northeast, relative to the other regions, not only fail to compensate for regional inequalities in health status but actually compound them. Table 12 documents that SUS expenditures for prenatal and antenatal care in the North and Northeast regions are 10-15% lower, per birth, than in the other regions, and that there are two to three times as many doctors, per inhabitant, in the Southeast as in the North and Northeast. The reason for this is that the formulas for the regional allocation of SUS resources tend to follow historical consumption patterns, and the South and Southeast have long had more health facilities and health professionals. The result of these regional inequalities is the persistence of pockets where modern health care is simply inaccessible: an IBOPE survey found that 1 1% of individuals with no education in Brazil, some 2-3 million people, never used the SUS (Medici 1998). That means that about 50,000 Brazilian women with no formal education and who have had no contact with the public health system give birth each year, assuming that fertility rates in that group resemble those of the nation as a whole. Similarly, about 30% of mothers in the lowest income quintile give birth without skilled assistance: that amounts to about 11% of all births in Brazil, or more than 350,000 births a year. Redressing these inequalities cannot be done overnight, but the means to do so are not complex. The government could choose to gradually increase the spending caps for facilities in poor states and municipalities, or it could de-link SUS transfers from expected hospital expenses. Setting a floor for minimum spending on primary health care, which the government did several years ago with the establishment of the PAB, was a first step in this direction. Table 12: Regional Distribution of Health Resources in Brazil Region SUS mean expenditures Inhabitants per Inhabitants per Doctors per on delivery and ante-natal doctor, 1998 registered nurse, registered nurse, care,perbirth,R$ 1998 1998 1998 South 239.80 724 1,973 2.80 Southeast 231.98 478 1,921 3.83 Center-west 231.20 771 3,035 3.41 Northeast 209.23 1,202 2,714 2.31 North 193.03 1,564 3,269 2.10 Source: Author's calculations, Barros (2000), and Dal Poz (2000). 3.9 The Medicalization of Health care. There are about three doctors for every nurse in Brazil. Even in the North and Northeast, where the health conditions and resource scarcity might argue for less medicalized care, the doctor to nurse ratio varies from 3.7 in Alagoas to 1.6 Paraiba. While the national ratio of 3.2 is not unusually high by Latin American standards, it is instructive to compare the ratio to that in Malaysia, with 0.3 doctors per nurse, Thailand, with 0.2 doctors per nurse, and Costa Rica, with 1.2 doctors per nurse. In maternal and child health, the consequence of insufficient nurses is that birth attendants, rather than nurses, assist doctors during deliveries, and doctors do not have enough time to spend on delivery and postpartum 39 care. This work load for doctors could be responsible in part for the high c-section rate in Brazil. Medical training is also highly specialized, with a shortage of generalists. As a result, much more care than is necessary is delivered by doctors in hospitals, and not enough time is spent on family planning counseling, breastfeeding promotion, and health education on topics like immunization and the recognition of severe pneumonia. Brazil also does not appear to spend enough on public health: rough calculations show that public health constitutes about 1% of total spending in Brazil on health care, compared to 3% in the United States, even though the payoffs to public health spending are probably larger in Brazil (World Bank 1997). The consequence is inefficiency and avoidable hospitalizations: even though as many as 10% of Brazilian children do not have access to hospitals, the hospitalization rate for children under 5 in Brazil was 10.7% in 1988, compared to 7.4% in the United States in 1996.11 Addressing these issues requires training more nurses, re-writing professional regulations to allow qualified nurses to perform a broader array of medical procedures, reallocating spending toward public health, and resisting efforts to carve out ever more special reimbursement rates for technologically complex procedures. The Government has taken one important step in this direction with its PROFAE project, financed in part by the IDB, which aims to train and professionalize the 225,000 unskilled nurses' aides in the country. 3.10 Diverse Cultures and Practices. The present study conducted two focus group evaluations regarding primary health care with low-income women in Ceara and Bahia. Among the strongest policy recommendations to emerge from them were to increase health care providers' and administrators' knowledge of local practices and cultures. Understanding why teenagers become pregnant, why parents feed their children what they do, why patients choose not to seek health care, and understanding them in the conceptual world and vocabulary of local individuals, is critical for promoting behavior change. Educational materials on leprosy, diarrhea, AIDS, and other public health concerns were also too sophisticated for many focus group participants. Brazil has the resources to overcome problems like this. A large number of NGOs and community-based organization have successfully responded to the Governments' request for educational materials and direct work with communities regarding HIV/AIDS prevention. The Government has developed collaborations with NGOs in the area of maternal and child health, and these could easily be expanded. 3.11 Medical Referrals. The referral and counter-referral system in most of the SUS is vague and functions badly. The absence of clear procedures complicates getting access to care and makes doctors only vaguely accountable for their services. One community health care agent described the problem: At their reference health post, there's no obstetrician. . .no person to collect material for Pap smears... the people complain a lot. They have to go to Maracanaui Hospital or Elias or JPA... they think it's a hassle to get an appointment. . . so much bureaucracy. . .you have to speak with the chief. . . for the chief to send a document proving that you live here and that the post doesn't do Pap smears... it's all so vague (Nations 2000) " Calculations based on data from Datasus/MS, IBGE, Barros (2000), and NCHS 1999. 40 This is a critical problem for maternal and child health and particularly for delivery care. Medical records on antenatal care are not linked to treatment at the time of delivery and to postpartum care. It also compromises the capacity of the system to diagnose and then rapidly treat obstetric emergencies, which is essential for maternal health. Ideally, birth attendants and others in health posts should be able to recognize complications in pregnancy and refer mothers to higher level facilities; and municipalities should have emergency transport vehicles and transport financing available. 3.12 Several factors contribute to the weakness of the referral system in Brazil. First, political rivalries, bureaucratic divisions, and undefined roles in the decentralized system lead to a lack of coordination. The several programs and agencies at the federal level, discussed below, have established parallel reporting and information structures at the state and municipal levels; states and municipalities have separate coordinators and databases for the various programs; and municipalities led by different political parties at times do not collaborate. Second, the authority of public sector managers over SUS-affiliated philanthropic and private hospitals, which constitute the bulk of facilities in the system, including 68% of the obstetric beds, is not well delineated. Third, the payment system rewards high volume work on the part of public and private philanthropic providers with little time for follow-up. Public sector health facilities can use AIH payments to give bonuses of up to 30% to their staff if they meet "productivity targets," and physicians in philanthropic hospitals, who generally work independently of each other in rotating shifts, are credited separately from the payments to the hospitals. Finally, the staff turnover and part-time, itinerant status of doctors in SUS facilities complicates communication. Salaries for doctors are so far below private sector wages as to leave public sector managers little choice but to allow doctors to work part-time. Again, solutions to these problems cannot be created overnight. But hiring more full time doctors in the SUJS; compensating health professionals as teams composed of generalists, nurses, and specialists; and developing rules and procedures for referrals, patient transfers, and co-operative arrangements among facilities and municipalities are steps that could address some of these issues. Government Programs in Maternal and Child Health 3.13 The Ministry of Health, in addition to services provided through the SUS, coordinates several programs for maternal and child health at the federal level. The predecessor of these programs was the Programa de AtenVdo Integral da Saude (PAISM) of the mid 1980s, which sought to provide health care services for women at a time when the health care system in Brazil was limited to formal sector workers, then largely men and their dependents. The PAISM constituted the first national policy in Brazil to incorporate family planning, and it aimed to change health care provision so that it focused on the needs of women at all stages of life. The work practices and institutions of health care delivery in Brazil, developed over decades, were not easily altered, however; and PAISM, rather than creating a new "model" of health care services, became a program advocating a few additional interventions, particularly contraception, that the health care system had not previously offered. 3.14 A similar challenge confronts existing government programs in the area of maternal and child health. While several of them provide training and educational materials for the SUS, their delivery systems, information systems, and reporting structures often parallel the national health 41 care network. As a result the programs cannot address and are not designed to resolve the underlying sectoral issues described above. In addition, the division of responsibilities can create difficulties in coordination. For instance, condoms are often delivered by posts and officials connected to the AIDS program, not by decentralized SUS health centers, to which people go for the majority of their health concerns. Some states do not receive vitamin A supplements at the same time that they receive federally purchased vaccines, and as a result the vitamin supplements lose their "ride" to interior municipalities and are delivered late or not at all. The parallel reporting structures of the programs also leads municipalities and states to appoint coordinators for each of them, and the divisions at the federal level are then replicated locally. 3.15 Table 13 below summarizes available information regarding the twelve programs in the Ministries of Health and Education most directly related to maternal and child health. (Note that these figures include only program expenditures specifically itemized in the official budget, largely for staff at the federal level. Additional expenditures on these programs from municipal and state sources, and through other Ministry of Health programs, such as general financing for primary and higher levels of care, are not included). Together, they accounted for R$2,444 million in the 1999 budget. The six largest programs by resources allocated, accounting for 87% of the total, were the School Lunch Program ($903 million), the AIDS/STD program (R$593.5 million, including R$487 million for drugs), the Family Health/Community Health Agents program (R$379 million), the National Immunization Program (R$269.6 million), the Food and Nutrition Program (R$158 million), and the Program of Health Care for the Disabled (R$128 million). All of these were managed in the Secretaria de Politicas de Saude (SPS), including the Family Health Program (which was recently moved to the SPS so that it could integrate more effectively with other programs in maternal and child health), with the exception of the School Lunch Program (Ministry of Education) and the National Immunization Program (National Health Foundation of the Ministry of Health). 3.16 The principal tasks of the two federal programs most directly connected to the subject of this study, Women's Health and Child Health and Maternal Breastfeeding, are defining norms, collecting data, conducting studies, consulting with state and municipal health officials, and conducting training. These are relatively small initiatives whose total budget for 1999 was R$5.6 million. For comparison, overall SUS transfers to providers for antenatal care and births were R$583 million in 1998, and transfers for conditions originating in the perinatal period were R$105 million in the same year. An additional R$605 million were spent on hospitalizations of children under age 14 in 1998, and R$6.6 million was spent in 1999 (through October) for high risk pregnant women in the new category of special payments for services of medium and high complexity. Adding up these numbers, the two federal programs in women's health and child health constitute less than 0.5% of hospital payments for medical care provided for pregnant women and children. The federal government spends considerably more resources on related programs, of course, programs such as immunization, school lunches, and AIDS, which are described in the table below. 3.17 The remainder of this chapter analyzes the programs most relevant to the themes of this study and takes an extended look at the Family Health Program. 42 3.18 Food and Nutrition. The federal government's Program to Combat Nutritional Deprivation (PCCN) certifies whether municipalities can acquire and deliver milk and soy supplements to pregnant women and children between the ages of 6-23 months. If so, the municipalities qualify for a supplement to the transfers for primary care, the PAB. With an annual budget of R$158 million, this food supplement program is much smaller than many previous nutrition programs in Brazil that past evaluations had found to be ineffective (Musgrove 1990). Still, it is not clear how beneficial these milk and soy supplements are, and some state- level evaluations suggest that they do not reduce the numbers of women and children at nutritional risk because the supplements merely displace household expenditures. The PCCN also provides vitamin A and iodine supplements to areas where deficiency is endemic. Nutrition education is not the top priority of the PCCN, a strategic decision that deserves re-evaluation in light of the international evidence demonstrating that nutrition education is more effective than food supplementation alone. The Ministry also finances 70% of the budget of Pastoral da Crianca, an NGO affiliated with the Catholic Church that delivers a multi-mix nutrition supplement to pregnant women and children and conducts child growth monitoring in poor communities. Its health interventions' impact on health status requires formal evaluation. Because the NGO works with the poorest communities in Brazil and works broadly - promoting self-esteem, persuading family members to foreswear domestic violence, encouraging schooling and literacy - its activities might usefully be expanded. 3.19 Women's Health and Child Health. The Ministry's Program for Women's Health, with a budget of R$1.7 million, works with state and municipal health secretariats, professional associations, and others to establish clinical norns for women's health issues. It has encouraged payment for anesthetics during delivery and worked with hospitals, professionals, patients, and others to reduce c-section rates, awarding prizes for "humanized births." It also has helped to create "maternal mortality committees," which investigate maternal deaths in several states and municipalities, promoted antenatal care, expanded screening for cervical cancer, encouraged local initiatives to reduce violence against women, and supervised the training of 237 new obstetric nurses at 11 different universities. The Program for Child Health and Maternal Breastfeeding, with a budget of R$3.9 million, similarly sets norms and trains professionals in neonatal care for low birth weight infants, accredits "baby friendly" hospitals (with PAHO and UNICEF), registers birth, combats anemia, trains professionals in case management of childhood illness, and promotes breastfeeding. Both of these are relatively small programs, with a total of twenty-five professionals between them. For comparison, the STD/AIDS program has a budget of R$106.5 million for national coordination and 120 professionals. Given that the burden of illness on mothers and children remains substantial in Brazil, both programs (and/or parallel programs at the state level) could usefully be expanded, particularly in areas such as family planning in poor communities, building coordinated referral systems for antenatal, delivery, neonatal, and postpartum care, contracting NGOs to produce greater numbers of appropriate educational materials and promote behavior change in culturally distinct communities, designating facilities for safe and legal abortions, and expanding training in the integrated management of childhood illness in the Northeast. Expansion in many of these areas will pay for itself in the form of reduced hospitalization of severely ill children and mothers. 3.20 School Health and Reproductive Health. The school health program spent R$16.1 million in 1999. Its emphasis was on the diagnosis and treatment of visual and auditory disabilities (World Bank/PAHO Partnership for School Health in Latin America and the 43 Caribbean 2000). For maternal and child health, the program could usefully develop peer group discussions of reproductive health issues among school-age adolescents. It will be important to create discussion groups of young men and women, to focus them in poorer communities where fertility rates are highest, to coordinate discussions with primary health care providers, and, to extent allowable under Brazilian law, to make contraceptives available to adolescents. The Program for Adolescent Health (R$1.2 million budget, 9 professionals) is planning a campaign against teenage pregnancy. While useful, it will be important for the messages of the campaign to be integrated into the daily work of health care providers, school health programs, and other initiatives, so that its achievements are not forgotten once the campaign ends. 3.21 Program to Reduce Child Mortality (PRMI). The executive committee of Comunidade Solidaria, a partnership of government and civil society, is located in the office of the President of the Republic. The basic agenda of Comunidade Solidaria includes the program to reduce child mortality, which aims to coordinate 19 programs (seven in the Ministry of Health) in five different ministries and strengthen their impact in the 1,373 poorest municipalities in the country. The PRMI does not provide additional resources to the program or the municipalities, but it assists the municipalities in implementing the programs and exempts them from contributing counterpart funds when necessary. 3.22 State-Level Initiatives. Several states have initiated programs to structure delivery care for pregnant women and postpartum care of mothers and children. Some of these programs, launched in Sao Paulo, Goias, Parana, among other states, aim to "humanize birth" by accrediting hospitals that provide quality maternity care and that allow newborns to stay with their mothers at birth, permit mothers to feed their infants at their own schedules, and encourage a women to visit before they deliver. Some states, including Pernambuco, have initiatives to train obstetric nurses to perform normal deliveries and place them in "Birthing Homes" (Casas de Parto) in Recife and interior municipalities around the state. Other states are designating tertiary care centers for pregnancy complications, using agentes to identify and monitor women at risk, and creating "waiting homes" near hospitals for women about to deliver. Some cities, such as Goia.nia, have programs to identify every pregnant woman and educate her about her rights, often with the help of PSF teams, designate a health care provider as her prenatal care giver and a hospital as her delivery site several months prior to delivery, and lift the quantity restrictions for births in all hospitals in the area. Several state are also beginning to build municipal coalitions in health services both to help municipalities with limited capacity and to make referrals work more smoothly. 3.23 These attempts to improve and coordinate delivery care in Brazil are too recent to have been evaluated systematically. It is nevertheless possible to comment on the strategies that might appear most promising. Training agentes to identify women at risk for pregnancy complications probably will not significantly reduce perinatal and maternal mortality because those complications are extremely difficult to predict. Skilled attendance at birth, with proper back-up, is a critical component of maternal mortality reduction programs, so training obstetric nurses is an important step in the right direction. The use of new tertiary care centers will depend critically on transportation to the facilities and the effectiveness of the referral system; otherwise women who develop complications will not benefit from the specialty care. The designation of a facility and a doctor as the site for delivery, combined with the lifting of quantity restrictions for births in facilities, could help make doctors more accountable for their services and could ease the 44 difficulties women face when trying to find a hospital to take them in, particularly when they face complications. These latter approaches, as well as efforts to build municipal coalitions, should be watched carefully and expanded if the evidence suggests that they are successful in reducing maternal and perinatal mortality and morbidity. 3.24 Making Providers More Accountable. The accountability of providers is part of a broader problem in the political economy of health care in Brazil. There are several experiments throughout the country to contract out more care, grant public hospitals more autonomy, increase the authority and responsibility of physicians in public and private hospitals, and introduce quality monitoring systems in public and private facilities. There is also a proposal to increase the accountability of doctors for monitoring pregnant women by creating a DRG category that combines delivery with prenatal care: doctors would be paid more for delivery if they have also done prenatal care for a given patient. All of these experiments should be encouraged and then evaluated carefully. The Family Health Program 3.25 The Family Health Program (PSF) aims to restructure the delivery of primary health care in the SUS. It builds on and will eventually subsume the Community Health Agents Program (PACS). In the PSF, family health teams, consisting minimally of a doctor, a nurse, a nurse auxiliary, and four to six community health agents, provide basic care to a defined set of 1,000 - 1,200 families. The teams both provide primary health care and lead efforts at prevention and behavior change. In the area of maternal and child health, the PSF teams will in theory be the preferred vehicles for the delivery of many of the key interventions identified in the previous chapter, such as case management of childhood illness, antenatal and postpartum care, family planning, and breastfeeding promotion, because they have sustained contact with families in their communities.12 The health care professionals work full-time in their public duties and accordingly receive higher salaries than most public sector doctors and nurses. The community health agents live in the communities in which they work, often have not completed secondary education, and receive salaries close to minimunm wage. Where and when it is fully realized, the PSF will in theory be the principal point of entry into the SUS, making the system less "hospital- centric." In summary, the PSF creates more sophisticated and permnanent primary health care for an enrolled population in a fixed geographic area, selects the family unit as the focus of primary care, identifies risk factors and the individuals that might be exposed to them, and emphasizes preventive care, inter-sectoral collaboration, and community-based infornation and education regarding health. 3.26 Although municipalities generally manage PSF teams, hire personnel, and provide non- personnel inputs such as facilities and medical equipment, the federal government provides 12 Community health agents in the PACS are directed to carry out these tasks: registering all families in their area and visiting them at least once per month, weighing and measuring all children under two years of age, verifying vaccination, identifying children of school-going age not attending school, teaching families how to prevent diarrhea and use oral rehydration, provide information about preventing acute respiratory infections, promoting breastfeeding, identifying pregnant women and encouraging them to receive antenatal care and tetanus vaccinations, encouraging the prevention and diagnosis of cervical and breast cancer, providing information about family planning and sexually transmitted diseases (including AIDS), assisting ill family members who have received medical treatment, and encouraging solutions to environmental problems such as garbage collection and sewage disposal. 45 partial financing, essentially matching grants, for municipalities that implement the program. In their current structure, the federal grants for the PSF range from R$28,000 to R$54,000 per PSF team per year, reaching the upper limit when the population coverage in a municipality exceeds 70%, which gives incentives for smaller municipalities to achieve total coverage of their population quickly (Ministerio de Safude, Portaria No. 1,329, December 12, 1999). State health secretariats are supposed to supervise and train municipalities in the theory and practice of the PSF program, particularly the smaller municipalities in the interior with limited management capacity, and provide additional inputs, such as medical equipment, pharmaceuticals, and educational training, to municipal PSF teams. Several states offer direct financial incentives or support to municipal PSF programs. For instance, Acre finances municipal PSF teams until the municipalities are qualified to hire them on their own; Ceara hires agentes for all municipalities in the state; Mato Grosso hires and pays for doctors and nurses in the state's PSF teams; Tocantins contracts PSF nurses; and Goias offers a matching grant of R$2,000 per team per month to its municipalities (Ministerio de Sauide 2000). 3.27 In December, 1999, there were 4,945 PSF teams working in 1,870 municipalities in each of Brazil's 27 states. Those teams, assuming a coverage rate of 3,450 people per team, provided care to 17.1 million people, or about 10.4% of the Brazilian population. First adopted in 1994, the PSF has expanded quickly recently: about 67% of PSF teams have been operational for less than two years, and 70% of the municipalities that have PSF teams have implemented them in 1998 or 1999. Under current Government plans, the PSF will provide coverage to 50% of the Brazilian population by the end of the year 2002, which means it will have to quintuple in size in less than two years. In December, 1999, an additional 87,000 community health agents provided health outreach services to an additional 47 million Brazilians through the community health agents program (PACS), in which teams of health agents conduct health outreach under nurse supervision. As the PSF expands, the community health agents in the PACS will be integrated into PSF tearns. 3.28 Appendix B analyzes the targeting of the PSF and presents some preliminary information on its impact. It finds evidence suggesting that the PSF might already be having a discernible impact on maternal and child health. For the PSF to evolve as planned, however, and for it to have a substantial impact on maternal and child health, several key issues will have to be revolved. 3.29 Referrals. As yet, the PSF teams do not have the power to refer patients to facilities, nor do they generally follow up on their patients' care when they travel to hospitals. PSF teams rarely receive counter-referrals from higher level facilities. This relationship between facilities, and between primary and higher levels of care, is especially important for maternal health. Ascribing to PSF teams the power to refer patients will also increase the importance of primary care relative to hospitals and higher level care in Brazil. For those reasons, developing a more formal referral system is an important challenge for Brazilian authorities. It could be accomplished either through more explicit rules and norms governing the transfer of patients and information, and placing such referral power in the hands of PSF teams, or by granting some budgetary power and control over AIH admissions to PSF physicians, in the manner of GP fundholders in the British system. Whatever the option taken, building a functioning referral system will be critical to the success of the PSF. 46 3.30 Determining Priorities. Every federal government health campaign and program adds a new task for the agentes, with the result that their health promotion messages tend to change with every new initiative. For example, diabetes and hypertension are currently on the lips of PSF teams everywhere as the result of a national decision to address these risk factors. Though these are indeed risk factors in all regions of the country, the emphasis on them tends to supplant local planning and decision making, so that in the Northeast the importance of maternal and child health, including antenatal care, receives less emphasis. That can dilute the effectiveness of PSF teams' messages as well as provide them with a great deal of work. In the focus groups with health care providers, one nurse in Bahia said: People here have a lot to do, for example, in the coming week we are going to meet with a group of asthmatics, in the next two weeks we have to spend a day with the high blood pressure group, the diabetics, so you can't get it all done. If we're going to reach our goal of seeing one thousand families, we'll have to stop helping those with high blood pressure, the diabetics, stop doing family planning, and so on. And we can't, on top of all this, count how many people have high blood pressure, have diabetes, have mental problems. That is a whole program on top of everything else (Trad and Bastos 2000). 3.31 To work most effectively in a country as diverse as Brazil, the PSF teams should develop priorities based on the needs of their communities. That flexibility might even extend to the structure of the teams: in areas with high perinatal and maternal mortality, for example, it might be more important for PSF teams to include a well-trained obstetric nurse than a physician, particularly if a good generalist physician is hard to recruit. In general, it should be noted that the PSF is not primarily a maternal and child health initiative, and PSF teams will not focus on the set of interventions identified in Chapter 2, even in communities with high maternal and child mortality rates, unless explicit policy decisions are taken to do so. 3.32 Qualified Health Professionals. Municipalities trying to build primary health care through the PSF report a shortage of physicians trained in primary health care. In the words of an official in the Ministry, what the PSF is trying to do is "change a flat tire on a car while it is still running." In one municipality in Goias, a casual inquiry yielded a not uncommon situation: PSF doctors there were originally trained in gynecology, cardiology, anesthesiology, and plastic surgery. Medical schools do not train many generalists, and students do not demand such training, because labor markets do not reward general medicine and because the specialization garners relatively low status in the profession. In the decade and a half from when the National Commission in Medical Residencies established a specialty in "general and community medicine" to 1996, only 346 diplomas had been registered (Campos 1997). Many of those hired found jobs in emergency rooms in peripheral areas and other less desirable jobs because there was little demand for their expertise. In some cases, most notably Tocantins, states have made special arrangements with the government of C'uba to import physician generalists to Brazil. About 200 Cuban physicians are working in the program in Brazil. In nursing, the impact of the demand created by the PACS program is already visible: the number of nursing jobs in the economy is now expanding faster than the number of jobs for doctors, and 20 of the 137 nursing programs in Brazil as of 1998 had been created in the last two years (dal Poz 2000). 47 3.33 The government is using two approaches to confront the human resources problem in the PSF. First, the states and municipalities train new personnel in the PSF teams. Usually the full team, except the agentes, receives an introductory training course together. Agentes receive one- week introductory courses and sporadic in-service training of shorter duration (half a day to two days), conducted either by the PSF nurse or a contracted trainer. No formal evaluations of these in-service training programs exists. Site visits to four states suggested that the knowledge and skills of PSF personnel varies widely. In some cases agentes had completed only two or three years of primary school; in others they had completed secondary school, were attending nursing school or related training, and could vaccinate children, take blood pressure, and give clear explanations of health topics. Second, the federal government has, partly with funds from Reforsus, established ten permanent training centers around the country (polos de capacita,do) and twenty other special projects and courses for PSF nurses and agentes. The expenditures on these programs total $R9 million. Third, the Ministry of Health is also proposing a required civil service requirement of six months to one year for new graduates of medical and nursing schools, which total about 7,200 doctors and 4,700 nurses a year. The new graduates would work in or even lead PSF teams around the country. In the medium to long term, the success of the PSF will depend on the integration of generalist training into the curricula of medical and nursing schools. 3.34 Contracting. According to the government sponsored evaluation of the program that received information from 997 municipalities (Ministerio de Sauide 2000), most PSF personnel are hired under temporary contracts (49% of doctors, 44% of nurses), or are they paid on a piecework basis (27% of doctors, 24% of nurses). These arrangements are made either directly by the municipalities or through NGOs, cooperatives, or other autonomous entities that then sub- contract the health professionals. Approximately 16% of PSF doctors and nurses are hired as full public servants, and an additional 11-12% are hired by municipalities but work under the private sector labor codes. The remaining personnel are contracted under a variety of other arrangements. 3.35 Because most of the personnel who work under temporary contracts and who are paid on a piece work basis do not qualify for benefits available to other Brazilian workers, particularly paid holidays and wages for the "thirteenth" month, and because the hiring processes for PSF personnel are not as transparent as in other areas of public employment (only 12-14% were hired through concursos pzublicos, or advertised public hiring), a number of political and legal challenges have hounded the PSF program. The Ministerio PuTblico Federal, the Tribunal de Contas da Unido, unions, and others have questioned the contracting procedures and claimed for PSF personnel the benefits enjoyed by other workers in Brazil. They have also argued that these contracting methods promote staff turnover and impact negatively on the quality of services, the continuity of care, and user satisfaction. For some officials and other analysts, however, the PSF is the beginning of a new form of public sector employment in Brazil, one that allows for greater flexibility in personnel policies. "Slowly, the management of public resources and the provision of public services are beginning to become distinct tasks. In the SUS, this tendency is seen in the multiplication of forms of contracting out through private companies and cooperatives, in various situations, such as hospitals and the Programa de Saiude da Familia, demonstrating that the public sector is becoming less of a direct employer and more a contractor of labor." (Dal Poz 2000) Nevertheless, until such time as this kind of contracting is more widely accepted, political and legal challenges might slow the expansion of the PSF. 48 Summary 3.36 The health care system does not effectively deliver many of the interventions mentioned in Chapter 2, including quality prenatal care and effective delivery care, not primarily because spending in the sector is too low but because resources are not well allocated and utilized. Regional inequalities in spending and the allocation of health resources are regressive. The system could deliver services more cost-effectively by relying less on physicians to perform services that other health professionals could provide. Despite decentralization, health sector goals in most municipalities continue to be derived from undifferentiated national priorities. The referral system, critical for maternal and perinatal health, is vaguely defined and functions poorly; and the quality of care in many facilities needs improvement. Federal government programs for maternal and child health often run parallel to rather than integrate with SUS facilities, where most of the recommended interventions from Chapter 2 will have to be delivered. The last chapter makes policy recommendations to improve maternal and child health. 49 Table 13: Federal Programs in Maternal and Child Health Program and Agency Description 1999 Budget and Population Covered in 1999 Resources Food and Nutrition programs For populations at risk, provides R$158 million Municipalities qualified: 4,026 SPS - MS milk and soy milk to children ages 6-23 months, pregnant women, the 9 technical professionals Beneficiaries: 777,103 total elderly, some children between 24- 518,069 children 59 months, children of HIV+ women under 6 months. There are additional programs in iodine, vitamin A. Family Health and Community Provides greater access to basic R$379 million Family Health: 17 million Health Agents Programs health care and aims to restructure (PSF and PACS) the model of care in the SUS 16 professionals Community Health Agents: 64 SPS - MS million Program of Health Care for the Coordinates activities implemented R$ 128 million Not available Disabled by local governments to support, (for purchasing orthotics, SPS - MS include, and rehabilitate individuals prostheses, and medical with mental, motor, auditory, and care for the disabled) visual disabilities. It helps purchase equipment for the disabled. 3 professionals Women's Health Establishes norms and conducts R$ 1.7 million Not available SPS - MS training to improve care during pregnancy, delivery, and 10 professionals postpartum; promotes efforts to reduce teenage pregnancy and unwanted pregnancies Cancer Prevention, Control, and Promotes efforts to prevent, detect, R$ 6 million spent 6 million women tested for Treatment and treat forms of cancer prevalent through Oct 1999 cervical cancer; testing SPS - MS in the country information available for only 2.2 million; 11,988 women tested positive; 7,805 given treatment Reduction of Infant Mortality Coordinates existing programs No dedicated budget - Comunidade SolidAria works in SPS - MS linked to Comunidade SolidAria uses funds from other 1,100 small municipalities programs Child Health and Maternal Establishes convenios with R$3.9 million Not available Breastfeeding localities for initiatives to improve SPS - MS child health; leads campaigns to 15 professionals register births, reduce iron deficiencies, accredit hospitals as "Baby Friendly," establishes breastmilk banks in hospitals Prevention and Control of Works with NGOs and localities to R$106.5 million for 78,000 AIDS patients Sexually Transmitted Diseases prevent HIV and help patients; national coordination and AIDS acquires and distributes AIDS drugs Convenios with 8 SPS - MS and condoms; trains labs to R$487 million for AIDS municipalities, 27 states, and 3 diagnose and control viral loads; drugs municipal consortia trains health care professionals to care for AIDS patients 120 professionals 50 million condoms distributed National Immunization Program Purchases vaccines and imuno- R$ 269.6 million 1997 vaccine coverage: 75.5% FUNASA - MS biological materials and distributes DPT, 78.5% measles, 77.5% them to localities; promotes polio vaccination 50 School Health Provides exams and basic care to R$ 16.1 million 2.9 million school children in FNDE - MEC children in public schools, grades 1-4 (50% of total particularly in the areas of vision enrollees) were tested .__ _ _ _ _ _ _ _ _ _ and hearing ability School Lunch Establishes conv8nios with states R$903 million 33 million students FNDE - MEC and municipalities for school lunches and transfers resources to them Adolescent Health Works with states and R$1.2 million Not available SPS - MS municipalities to diagnose and treat adolescents in areas of reproductive 9 professionals health and violence I Source: Information on programs from Barros (2000). Expenditure figures are taken from budget line items and do not encompass all government spending on the associated health care categories. 51 52 4. POLICY RECOMMENDATIONS 4.1 The analysis in the preceding chapters makes clear that Brazil can take several measures to improve maternal and child health in the country. Four categories of policy recommendations follow: goals, basic actions to achieve the goals, measures to improve the quality and effectiveness of care, and evaluation. Goals Aggressive targets for maternal and child health in Brazil are both desirable and feasible. The easiest indicator to utilize is the infant mortality rate. As a result of economic growth, urbanization, and past investments in education, health care, water, and other social programs, the infant mortality rate has fallen 5% per year over the last decade. It is feasible to maintain that velocity of decline, reaching an infant mortality target rate of 24 by the year 2005. On the assumption that without additional interventions the velocity of decline in infant mortality would slow to 3.5% per year, the measures described in the next sections would prevent approximately 20,000 infant deaths over the next five years. It would also prevent a large number of unnecessary maternal and child fatalities and reduce the morbidity burden on mothers and children significantly. Basic Actions to Achieve the Goals * Because health indicators are so much worse for the poorest part of the population, it makes sense to focus efforts on the North and Northeast, rural areas, and on the poor, which are overlapping population segments. Priorities for these areas are inter-sectoral programs that reduce poverty and increase education. In addition, health initiatives on the management of childhood illness, family planning to improve birth spacing, breastfeeding promotion, and improved antenatal and delivery care are priorities. Those interventions could prevent as many as 40% of deaths among children under five in the North and Northeast. * Because perinatal mortality is the largest component of infant mortality in every region of the country and because maternal mortality has been stagnant for a decade, priorities for all regions are quality improvements in antenatal, delivery, neonatal, and postpartum care and breastfeeding promotion. A further priority for all regions is the introduction of the Hib vaccine. Those health initiatives, along with better case management of infectious diseases, could prevent one-third of all deaths among children under five. * SUS expenditures continue to favor the regions that are better off and healthier. Gradually reversing the allocation formulas, so that average expenditures per person and per birth are at 53 least as high in the North and Northeast as in the other regions of the country, would advance maternal and child health overall. Measures to Improve the Quality and Effectiveness of Care for Mothers and Children * There are insufficient nurses and nurse auxiliaries in the country. As a result, doctors perform procedures that well-trained nurses could perform as effectively, and nurses perform tasks that well-trained community health agents could manage. Doctors have insufficient skilled nursing assistance for antenatal care and for deliveries. That might be one reason that cesarean-section rate remains so high. Existing initiatives address this issue in part by training nurses aides and certifying a few hundred obstetric nurses per year. The obstetric nurse program could be rapidly expanded. In addition, it would be consistent with those initiatives to gradually revise legal regulations so that well-trained nurses and well-trained community health agents could take on some of the tasks and procedures currently reserved for physicians and nurses, respectively. * The referral system functions poorly in the SUS. This is a particularly acute problem for antenatal, delivery, and postpartum care. Consequently, the development of functioning referral systems at the municipal and state levels is a priority. Of particular value to maternal and child health are systems in which pre-designated providers offer antenatal, delivery, and postpartum care to every pregnant woman and newborn. For the referral system more broadly, a variety of methods might be utilized, including the creation of health care micro- regions or co-operative arrangements across municipalities, paying health care providers in teams, developing more precise rules and norms for the transfer of patients and information, gradually replacing part-time positions in the SUS with full-time jobs, and granting budgetary and referring power to health professionals in Family Health teams. * The activities of Family Health teams and municipal health secretariats are not driven often enough by local data and community priorities. A promising and feasible way to address this problem is to expand contracts with NGOs and community-based organizations for the development of educational materials and for community-level work in nutrition education, family planning, reproductive health, and the prevention and treatment of infectious diseases. * School health programs currently do not prioritize issues related to maternal and child health. In areas where fertility rates among teenagers are high, it would make sense for school health programs to work with peer groups of adolescents on reproductive health issues, to the extent allowable under Brazilian law, and to collaborate with NGOs and community-based organizations to reach youth not attending school. If successful, this initiative could be expanded. * Building on federal programs in women's health, child health, and family health, it would be enormously beneficial for maternal and child health to quickly expand initiatives for improving family planning programs and the Integrated Management of Childhood Illness in 54 the North and Northeast, and for improving the quality of antenatal, delivery, neonatal, and postpartum care, including breastfeeding promotion, in all regions of the country. Vehicles for quality improvement include new norms, protocols, and training, as well as new incentives and governance arrangements to make providers throughout the SUS more accountable for their care. Evaluation * Data on the causes of death in Brazil are based on a facility-based reporting system and liable to under-reporting. More timely measures of maternal and child health would help policy makers evaluate initiatives and adjust programs quickly. Regularly conducting household surveys that collect information on mortality rates, the causes of death, and morbidity across all age groups, sexes, regions, and income levels would facilitate evaluation and planning. 55 APPENDIX A: A LIFE-CYCLE APPROACH TO MATERNAL AND CHILD HEALTH IN BRAZIL A. 1 This chapter assesses health sector interventions to improve maternal and child health and recommends relevant ones for Brazil. It first describes the lifecycle approach to maternal-child health, which is the conceptual framework that organizes the following sections. Subsequently, a section on each of the interventions describes the practice, relates it to the Brazilian context, and briefly comments on the benefits that Brazil might gain from implementing it. The Lifecycle Approach A.2 The lifecycle approach is a conceptual framework for achieving improved health, nutrition and population outcomes which was developed in the context of the World Bank poverty reduction strategy. It looks at stages during life, the risks associated with each stage, the potential interventions at each stage, and the desired outcomes. The hypothesis upon which the life-cycle approach is based is that clear identification of the main risks to health at different life stages can lead to the appropriate selection of interventions. This is particularly important in situations where resources are limited because it permits priority setting regarding the strategies for delivering specific health interventions. The figure below depicts the main risks at the different life stages. For example, it indicates that the greatest risks at the neonatal stage are infection, poor breastfeeding and neonatal death. While pregnancy risks refer specifically to the mother, risks during other ages refer to both sexes. In the case of the reproductive ages, focusing on both sexes is the key to reducing risk and increase the effectiveness of interventions. A.3 The life cycle framework highlights the cumulative nature of health interventions - benefits in one age group are partially dependent on interventions for that group at an earlier point in time. Prioritizing interventions at several points across the life cycle is necessary for sustaining improvements in health outcomes. In addition, research demonstrates that all the benefits of some interventions are not seen immediately - they benefit successive generations. The remainder of this chapter is structured around stages of in the lifecycle and the interventions most appropriate for them. 57 Figure Al: Risks in the Health, Nutrition, and Population Lifecycle Main risks of pregnancy and early life Pregnancy (mother) Neonatal period D4nsembs (Onfection aXEclanpsia Opoorbreastleeding SDUnsae. abortion QNbonaW death aEctopk prgnancy 5#Matenal death Pregnancy (child) aAnaena / altUGR infancy sffjati,n/ o\ \ aPoornuritbon 5Fo aal death *Poor growth and dweveopmet Birth (mother) \\Sfmu.nl illness Bwwiry ( r Eariy neonatal \ ant death conplicatdons period (chld \ SfHaemonhage SSepsjs Daternal death 4sphys,a Birth (child) cfallure ro Initiate SLow bith weight braastfeeding DStllbirth alHypothennia f'tehnn birth Post-partum (maternal) *Wnth trauma of as6epsis death SliermOnhage *Congenltal syphilis Slatera death Main risks of childhood, adolescence and reproductive period Reproductive period SSTDs SEarly/unwentelV unsafe sex aSEarty/unwanted pregnancy \ 20 rean / / yea Adoiescence Y Adtolnescon \ Childhood Op.n.b.hila P o o (SPoor. nurition growth fPoor and development development SFraquent Illness tWrug uselabuse "School-age" Wjr SDViolence Poor nutition, growth and @ Abuse and neglect SDDeath development Weath lrnJury SfHelminth infections ajAbuse and neglect Routine Interventions during the Reproductive Years * Family planning counseling and services * Screening, counseling, and treatment for sexually transmitted diseases (STDs) * Cervical and breast cancer screening and treatment A.4 Family Planning. Access to family planning counseling and services for both men and women reduces unwanted births and induced abortion, and promotes birth spacing. Because adolescents girls, who are particularly vulnerable, do not come into contact with the health system until they are pregnant or become ill, family planning programs need to reach them in other contexts, such as schools. Similarly, men can be reached in schools and universities, clubs, and at the workplace. 58 A.5 One option for adolescents who have unplanned intercourse is the use of emergency contraception, which has been shown to be safe and effective (Ellertson and others 1995). While emergency contraception should not be used routinely, it is a perfect backup method in cases of contraceptive failure, unplanned sex, and for rape victims. Its provision and use require little training and can be distributed through community-based distribution systems. Despite its availability, the utilization of emergency contraception in Brazil is relatively low because few physicians know how to use it (Galvao and others 1999). A.6 There is evidence that family planning programs can be improved in Brazil. While contraceptive use is high in Brazil (69% of women or their partners were using a modem method of contraception in 1996), choice of method is limited. Two-thirds of females using contraception opted for sterilization; and only 7% of couples were using male-based methods (DHS 1996). Current rates of adolescent fertility and induced abortion are also high, suggesting poor targeting by existing programs. SUS data on deliveries performed show that from 1993 and 1997 births to girls between the ages of 10 and 14 increased from 0.9% of all births to 1.2%, and corresponding data for 15-19 year olds increased from 21% to 25%/lo (RNFSDR 1999). In facilities in poor states such as Maranhao as many as 50-60% of all maternity beds can be occupied by girls under 19. The fact that 48% of all legally sanctioned abortions in Brazil are performed on girls under 20 years of age is evidence that the needs of adolescents need to be addressed. A.7 Sexually Transmitted Diseases, including HIV. Screening, counseling, and treatment for STDs should be made readily available to high risk men and women. Beyond the pain and discomfort of acute illness, STDs can lead to infertility, chronic pelvic pain, increased risk of ectopic pregnancy, and recurrent infection. There is also evidence that STDs increase both the infectiousness of and susceptibility to HIV. A.8 As of June 1999, Brazil had 155,590 reported cases of AIDS, close to two-thirds of all AIDS cases in Latin America. Recent estimates of the total number of people infected with HIV in Brazil place that figure at 536,000 (personal communication, Pedro Chequer, MS). The epidemic, which initially centered on higher income, homosexual men living in southeastern urban Brazil, slowly became characterized by heterosexual transmission in low-income populations from a broader geographical background. Brazil has a strong STD/AIDS control program that appears to have been successful in reducing rates of infection. While it has raised awareness among key high-risk groups, it has been slow to target women at risk, particularly poor women and adolescents. Moreover, its focus has been on HIV, with STD prevention receiving considerably less attention. Access to voluntary counseling and testing in general antenatal and family planning clinics is still quite limited. A.9 Cervical and Breast Cancer. Screening for these are relatively simple procedures is useless if not accompanied by timely treatment when cancer is detected. While routine Pap smears are effective in reducing the incidence of invasive cervical cancer, they are costly and not always available, particularly outside urban settings. A study conducted in India showed that visual inspection after application of diluted acetic acid technique was effective in detecting 59 cancer at an earlier, more treatable stage, and in South Africa its predictive value was shown to be as effective as Pap smears (Gaffikin and others 1997). Should routine Pap smears be available, women aged 35 to 65 should be screened every 3 years'3. To lower costs, visual inspection could be used as the primary means of screening, using cytology only for confirmation. For breast cancer, routine autoexamination and palpation by a physician/nurse may be the best and least expensive approaches. Mammography should only be used for screening in high risk cases every 2 years, focusing primarily on women aged 50 and 65. A. 10 Treatments for pre-cancerous lesions are cryotherapy and loop electrosurgical excision procedure (LEEP), which are low cost and easily applied. Cryotherapy is noninvasive, can be utilized at the health post level, and can reduce the odds of developing cancer for 5 to 10 years. A. 11 Breast and cervical cancer are the most common cancers afflicting women in Brazil. Pap smears are widely used to screen for cervical cancer, and in health posts screening is often conducted annually or even semi-annually for all women aged 15-59. Pap smears could be applied once every three years to women aged 35-65 without much loss in effectiveness.'4 A study of municipal facilities in the state of Sao Paulo found that over 50% of gynecologists' consultations were spent on Pap smears, which is a misallocation of physicians' time (Diaz and others 1999). Cervical cancer screening can effectively be done by paramedics, such as specially trained nurses. Interventions during Pregnancy: Antenatal Care * Infection Control (tetanus, malaria, parasites, STDs (syphillis, AIDS)) * Nutrition (iron, folic acid, iodine) * Detection and Treatment of Complications * Preparedness for Delivery (IEC, transport, home-based records, selection of delivery site, preparation for emergencies) A. 12 Antenatal care (ANC) has long been considered an essential component of maternity care in part because it was viewed as a way to identify women "at risk" of serious complications and refer them to appropriate levels of care. It is well known, however, that most obstetric complications occur among women with no risk factors (Rooks and others 1990). A study conducted in Zaire found that only 29% of obstructed labor cases had a previously identified obstetric history, and 90% of women with "risky" obstetric histories did not develop obstructed labor (Maine 1991). Thus, rather than risk assessment, ANC should focus on early detection and 13 The difference in percentage reduction in the rate of cervical cancer in screening every two years and every three years is one percent, i.e. negligible (Ludwig 1995), and thus where resources are scarce screening should be done every three years. See also World Bank (1991). 14 In Brazil, the norms of the Cervical Cancer Prevention Program call for Pap smears every 3 years, following two years with negative results, in all women aged 25-60, a broader age group than what is recommended. However, the Manual para a Organizacdo da A tencdo Bdsica of the MOH targets Pap smears to women 15-59, and although the frequency of screening is not specified in the manual, many health posts visited indicated annual and even semi- annual screening. Invasive cancer takes 8-10 years to develop and thus screening need not be conducted so early nor so frequently. 60 treatment of eclampsia, hemorrhage, spontaneous abortion, premature childbirth and the retardation of intra-uterine growth. The quality of the care provided in each visit is critical and a total of four quality visits is recommended. ]deally, ANC should be intimately linked with delivery care, which is not the norm in public services in Brazil, though the links are being established in some states (see Chapter 3). A. 13 ANC should contain, at a minimum, infection control, nutrition interventions, immunization with tetanus toxoid, and the detection and treatment of complications of pregnancy. It should also include information and counseling on the effects of harmful substances and local/indigenous practices observed during pregnancy, breastfeeding promotion and infant care, and various activities designed to prepare the mother, the providers and the community for the delivery, making the necessary arrangements to deal with an emergency should the need arise. What part of the ANC package can be provided in the community and what should be provided at a health facility depends on the capability of community health agents (CHAs). Routine ANC can be effectively delivered by a nurse practitioner, with proper back-up when complications are detected. In Brazil, however, it is often physicians who provide ANC. Although there is a near consensus that the quality of ANC should be improved in Brazil, as yet there are no studies that allow for estimates of the expected impact of improving quality. A.14 Infection Screening and Management. Infection control should always include tetanus toxoid immunization, screening for STDs and HIV, malaria prophylaxis where endemic, and deparasitation. While the mechanism is not fiully understood, women have increased susceptibility to malaria during pregnancy, which can lead to severe anemia in pregnancy and fetal wastage if not treated. Women living in areas where malaria is endemic are more likely to be asymptomatic and thus should either automatically receive antimalarials during pregnancy, or immediately receive antimalarials in case of unexplained fever. Syphilis can lead to increased fetal wastage, low birth weight, and congenital infections. As many as one-quarter to half of pregnancies in infected women result in fetal deaths, low birth weight or prematurity occurs in 25-65% of the cases, and congenital infection in 33-66% of infants. Moreover, there is evidence that STDs increase susceptibility to HIV infection. Syphilis and HIV can have devastating consequences for both mother and child. Screening and treatment for both of these needs to be part of basic ANC in areas where prevalence is high. Low cost detection of STDs can be made among symptomatic women (who make up as many as 40 percent of women) using syndromic case management, and can be done at health posts and higher levels of care AZT given to infected pregnant women can reduce vertical HIV transmission. Even when given only in connection with birth, antiretrovirals have a significant preventive effect on vertical transmission of HIV. A. 15 At present, routine ANC provided in the primary level of care in Brazil does not always include screening for STDs, deparasitation, and least of all, malaria prophylaxis. For STDs and HIV, antenatal care depends on access to laboratories. A.16 Maternal Nutrition. Low pre-pregnancy weight and low weight gain during pregnancy, a consequence of maternal malnutrition, are associated with fetal wastage and low birth weight (Susser & Stein, 1994, Ceesay and others 1997). What is less clear is the effect of supplementary 61 feeding programs on fetal wastage and perinatal deaths, possibly because supplemental feeding for pregnant women can substitute for existing food intake. A study in Gambia found an extra 900 calories per day reduced low birth weight by 35%, stillbirths by 55%, and perinatal deaths by 49% (Ceesay and ohers 1997). However, a meta-analysis (Kramer 1993) showed increased energy intakes led to only modest increases in maternal weight gain and fetal growth and statistically insignificant reduced risk of pre-term births and fetal wastage. A variety of micro- nutrient supplements during pregnancy, including iron and folate supplements, and Vitamin A, have been shown to have positive impact on maternal and child health (Galloway 2000, West and others 1999). Trained community workers can be very effective in distributing these supplements and encouraging their use. A. 17 There is a secular trend in increasing height, a measure of chronic malnutrition, among Brazilian women in all regions of the country (Victora 2000). In addition, low birth weight, which is the risk factor that would improve most from enhanced maternal nutrition, is already fairly low in Brazil, about 8%, even in the Northeast. Although there are parts of the rural Northeast and North where maternal nutrition is a problem, screening for these women on the basis of anthropometric measures has both low sensitivity and low specificity. As a result, programs that aim to improve maternal nutrition are unlikely to have a marked impact (Victora 2000). The effectiveness of the Brazilian government's food supplementation programs have been questioned in past evaluations (Musgrove 1990). A.18 Detection and Treatment of Complications. Prompt detection, management and referral of pregnancy complications are key components of effective maternal care. Eclampsia is an important cause of both maternal and perinatal death. Early detection of pre-eclampsia and hypertension can be done in the community through blood pressure and symptom assessment. Severe pre-eclampsia and eclampsia should be managed at a hospital. Detection of urinary tract and other infections is also necessary. Pre-partum hemorrhage can occur in the third trimester of pregnancy, due to placenta previa or abruptio placenta. Although not as common as post-partum hemorrhage, health workers should be prepared to refer women with heavy bleeding immediately without conducting vaginal examination because on average, death can occur within 12 hours (Maine 1991). For women with severe bleeding intravenous (IV) fluids and blood transfusions should be available. The capacity to detect and treat early complications varies widely in Brazil. A.19 Preparedness for Delivery. This includes educational activities within the community and with the expectant mother and her family, as well as activities designed to ensure timely transportation to a higher level of care should an emergency arise. It is generally estimated that at least 15% of women in childbirth develop complications. Studies conducted in Indonesia, Guatemala and Brazil have noted that community birth attendants can effectively detect and refer early signs of pregnancy and delivery complications (Alisjahbana and others 1995; John Snow Inc. 1997; Janowitz and others 1985). Similar studies conducted in Nigeria and Ethiopia found that improved referral by nurse mid-wives, combined with maternity waiting homes available and near a referral facility for women who show early signs of complications, led to declines in hospital-based matemal mortality. Improved detection makes little difference unless transport to the referral facility is readily available. Health posts and centers have a key role in educating 62 expectant women about avoiding harmful substances such as tobacco, alcohol and drugs, and about local cultural practices on care during pregnancy and infant feeding that may be detrimental to the fetus' and infant's health. A.20 Due to the weak links between ANC and delivery care in the SUS, preparedness for delivery can be difficult to achieve in Brazil. Because referrals function so poorly, pregnant women sometimes scramble to find an available hospital bed when they enter labor. Pregnancy Termination and Post-abortion Care * Safe pregnancy termination services when legally allowed * Emergency treatment of abortion complications * Family planning counseling and services A.21 Unsafe abortion is one of the major causes of maternal mortality worldwide. Women who have experienced complications from incomplete abortion who reach medical care are generally given a dilation and curettage (D and C), often with little or no anesthesia. D and C's are not only painful but can lead to infection in the absence of proper hygiene. Treatment of complications of incomplete abortion, and pregnancy termination, can be performed with manual vacuum aspiration (MVA) which, if performed during the first trimester with local anesthesia and antibiotics, can reduce the health risks to the woman, lower hospital costs, and speed recovery. Women should be given access to family planning counseling and services while still at the clinic to reduce the probability of another undesired pregnancy. Studies in Ghana have shown that with proper training, MVA can be successfully performed by non-physicians to treat incomplete abortion (Billings and others 1999). A.22 In Brazil, unsafe abortion accounts for 9% of maternal deaths and 25% of the cases of infertility. It is also the fifth leading cause of hospitalization among women (RNFSDR 1999). Given estimates of as many as 1.0 to 1.4 million abortions per year in Brazil, or about half the number of births, the government moved to extend safe abortion services to women in cases of rape or risk to life in 1997 (Correa and others 1998). Although only two hospitals offered the service in 1997, by the end of 1999, 28 hospitals were performing legal abortions, mostly with MVA (Faundes 2000). Still, however, a number of Brazilian states do not have a single public hospital that provides legal abortion. The provision of these services, together with the illegal use of cytotec'5, has led to a marked reduction in mortality and serious complications from induced abortion. The incidence of induced abortion complications has declined from one out of five deliveries in 1992 to 1 per 7.4 deliveries in 1997 (Faundes 2000). 15 Cytotec, originally developed for treatment of gastric and duodenal ulcers, is currently used in France in combination with mifepristone (RU486) to induce abortions up to 49 days of gestation. In Brazil it has been used as an abortifacient since 1986 when it was approved for sale in Brazilian pharmacies (for the treatment of ulcers). Public outrage regarding its widespread use led the government to restrict its sale to prescription only although it is easily available in the black market. 63 Interventions during Delivery and Post-partum period * Clean delivery * Essential obstetric care * Comprehensive essential obstetric care * Labor management via partograph * Maternal death reviews/clinic-based audits * Monitoring for infection and hemorrhage * Family planning counseling and services A.23 Maternal mortality rates fall largely as a result of institutionally-based medical interventions. While interventions at the community level, such as ANC, education, and nutrition can reduce the number of complications and facilitate their management, maternal deaths will not substantially decline unless the health system treats complications when they arrive (Maine 1999). That implies a functioning referral system, including transport and timely hospital bed availability, and effective care at higher levels. Linking all levels of care with clear protocols to manage obstetrical complications, and providing transportation and communications can help women receive appropriate care in a timely manner. A.24 A functioning referral system requires maternity waiting homes located near referral facilities for women who have histories of poor pregnancy outcomes, early signs of complications, or who live far from adequate care. It should also include training on certain life saving tasks for local level staff particularly in emergency situations. In Zaire, obstetric nurses saved lives after receiving training in how to remove the placenta manually and how to perform cesarean sections (Campbell and others 1995). A functioning health center, maternity, or local hospital requires the infrastructure, equipment, supplies and human resources to provide essential obstetric care (EOC) for normal deliveries. EOC includes clean and safe umbilical cord care to reduce infection, parenteral antibiotics for infection, parenteral oxytocin for third stage of labor to reduce post-partum hemorrhage, and parenteral anticonvulsants for eclampsia. Providers should also be able to perform manual removal of the placenta and retained products, conduct assisted vaginal delivery, repair vaginal and perineal tears, labor management via the partograph, and medical treatment of anemia. Use of the partograph can reduce prolonged labor, and in multiparous women, can reduce overall cesarean section rate (Lennox and others 1998). Higher level referral facilities must be able to provide comprehensive essential obstetric care (CEOC) to ensure adequate care for obstetric complications (Liljestrand 1999; John Snow Inc.1998). This includes everything in EOC, plus, surgical facilities to conduct cesarean sections, including access to anesthesia and blood transfusions. A.25 Where deliveries take place in the community, the delivery should be conducted by a trained midwife. Traditional birth attendants working alone can prevent only two of the five main causes of maternal death: infection, through proper hygiene during and after delivery, and post partum hemorrhage through proper management of the placenta (Starrs, 1997). Studies in Indonesia and Gambia have found that training traditional birth attendants in safe delivery was ineffective in reducing maternal mortality (Campbell and others., 1995). While not optimal, home deliveries can be made safer by linking them to the service network (Koblinsky, and 64 others. 1999) through a functioning referral system with transport organized by the community. Where professional help is unavailable, the birtlh attendant must be trained to recognize danger signs during labor and delivery. A.26 In Brazil, nurse/midwives perform only ten percent of all deliveries, with the proportion doubling in the North and Northeast (DHS 1996). Ninety-two percent of deliveries in Brazil take place in hospitals. Despite the high proportion of births that are delivered institutionally, perinatal mortality and mnaternal mortality remain high suggesting the quality of care provided is suboptimal at these facilities. Currently, many hiospital deliveries are attended by staff without midwifery qualifications. At least 40% of maternal deaths in Brazil are due to inadequate delivery practices (Victora 2000). A.27 Cesarean sections are performed at notoriously high rates in Brazil. The average for the country is 36.4% of all deliveries and reaches as much as 52% in Sao Paulo. WHO considers a range of between 5-15% to be adequate. While Cesarean sections can be lifesaving for both the mother and infant when obstetric complications arise, they can also increase maternal morbidity and pre-term births if the expected date of delivery has been miscalculated (Barros and others 1991). Although they can be performed more quickly and can be planned, they are also more costly than vaginal deliveries. Most c-sections are performed in low risk, higher income women (Barros and others 1986). It is striking that regional differences in maternal mortality rates in Brazil are narrower than those for infant mortality, the opposite of what is observed internationally since socioeconomic differentials seem to affect maternal deaths more strongly than infant deaths. It is possible that the South and Southeast regions have unusually high maternal mortality rates because of the high c-section rates there (Victora 2000). The Government attempted to refuse to reimburse c-sections above a fixed percentage of deliveries in given hospitals, but this only increased misclassification and fraud: patients and physicians continue to demand c-sections. A.28 Another problem contributing to higher than expected maternal and perinatal mortality in Brazil is the lack of integration between ANC and delivery care. Different providers offer ANC than perform deliveries, and there is often little communication among them. Deliveries are generally treated as emergencies, with physicians assisting women they have never before examined. Women cannot plan the location of their deliveries and meet their attending physicians. The SUS, moreover, sets quantitative limits on the number of deliveries that can be performed in hospitals, so some patients go from hospital to hospital during labor hoping to find one that will admit them. A.29 Postpartum care should include monitoring during the first 24 for hours for early detection of hemorrhage and infection. Postpartum hemorrhage is responsible for one-quarter of obstetric deaths, and occurs in about 10-20 percent of deliveries. Routine injection of oxytocin after delivery has been shown to reduce its occurrence and can be delivered in health posts by trained staff. For women with severe bleeding intravenous (IV) fluids and blood transfusions should be available. If severity of bleeding is not reduced once these measures have been undertaken, referral should take place immediately as death from postpartum hemorrhage can 65 occur within two hours of onset (Maine 1991). Postpartum care should also include family planning counseling and services. Interventions after Delivery * Essential newborn care * Breastfeeding promotion (described separately) * Immunization * Management of illness * Nutrition * Care for development * Accident prevention Essential Newborn Care A.30 Basic interventions performed after delivery are essential because they can reduce early neonatal mortality. These include resuscitation; thermal protection; early and exclusive breastfeeding; the prevention, early identification and treatment of infections; immunization (discussed separately); and the prevention of mother to child HIV/AIDS transmission (WHO 1994). A.31 Resuscitation. In developing countries it is estimated that there are 4-9 million cases of birth asphyxia every year, and that the crucial barrier to delivering proper resuscitation is the fact that delivery takes place at home and skilled attendants are not available. Standard resuscitation guides recommend ventilation by bag and mask and more complex interventions, such as intubation, chest compression, use of drugs and intensive care. However, a study from Sweden has shown that almost 80% of newborns who required resuscitation only needed the bag and mask intervention (Palme-Kilander 1992). In addition, an international multicenter controlled trial which enrolled asphyxiated newborn infants from Norway, Spain, Estonia, Egypt, and Philippines, showed that resuscitation with room air was as efficient as with pure oxygen (Saugstad and others 1998). A.32 Thermal Protection. Warmth and food are basic needs of a newborn. For preterm and/or low birth weight infants, in particular, special care is needed, including thermal protection. In many countries preterm babies are placed in incubators; and, due to shortages, they sometimes share incubators, which increases the risk of infection. The alternative "Kangaroo method" requires no special technology and can be applied in locations without electric power. In it, a mother places her otherwise healthy preterm infant, wearing only a diaper, between her breasts in skin to skin contact. The contact with the mother stimulates the newborn to start breastfeeding within an hour. Studies in Zimbabwe and Colombia have demonstrated the safety, effectiveness, and improved survival of this method. Kangaroo babies had survival rates 20-50% higher than babies in incubators. Another study has shown that the Kangaroo method is less costly than standard incubator care (Bergman and Jurisoo 1994, Kambarami and others 1998, Charpak and others 1994). 66 A.33 Prevention, Early Identification and Treatment of Infections. As described above, reducing the risk of perinatal infections depends on improved care for mothers, including the referral of high risk pregnancies, maternal immunization for tetanus, and care of the umbilical cord. In addition, preventing infections in hospitalized infants requires measures such as hand washing, rooming in and early contact with the mother, early discharge from hospital for well infants, and caring for infected infants in a separate group. A multicenter study in four developing countries identified signs and symptoms for predicting serious infections and are being included in case management guidelines. Feeding during illness and preventing hypoglycemia were also recognized as important parts of casement management (Mulholland 1998). A.34 Prevention of Mother to Child HIV/AIDS Transmission. There is evidence that HIV can be transmitted through breastfeeding, particularly from mothers with recent HIV infection and high viral loads. Estimates of this risk are still debated but are judged to be around 15%. Eliminating this risk completely requires replacement feeding for infants of HIV-infected women. A.35 Several observations regarding the Brazilian context are possible. First, in rural areas where deliveries take place in the home, early experience with community based neonatal care can be beneficial. A three-year study (Bang and others 1999) of 39 interventions and 47 control villages in rural India demonstrated that community-based neonatal care helped reduce neonatal and infant mortality rates nearly 50% in an illiterate and malnourished population. A.36 Although throughout Brazil the vast majority of births take place in hospitals with doctors in attendance, about 25% of deliveries in the rural areas of the North and Northeast occur without skilled attendants, more than five times the national rate of home delivery. In these areas there is good reason to believe that interventions such as those tried and tested in India could have a beneficial impact. A.37 There are several state and federal programs addressing the improvement of neonatal health in Brazil. The federal program on "Child Health and Breastfeeding," described in Chapter 3, has established technical norms for humanizing assistance to low birth weight newborns with the Metodo Mae Canguru ("the Kangaroo method"). The Instituto Materno-Infantil de Pernambucco (IMIP) has had a positive experience using this method with well preterm and low birth weight babies, and it is helping to replicate its experience in several reference hospitals across the country. However, in some hospitals the method is applied incorrectly: it is sometimes combined with incubator care, and well preterm infants are not separated from infected ones. Breastfeeding Promotion A.38 Even beyond the neonatal stage, breastfeeding has been widely recognized as having beneficial effects on child health. Breastfeeding protects children from diarrhea mortality, reduces the severity of diarrhea, and protects against pneumonia and other infections. 67 Breastfeeding especially protects against late neonatal deaths, primarily due to infection such as sepsis, pneumonia, meningitis, umbilical infection and diarrhea. In addition, breastfeeding contributes to reducing child mortality by providing optimum nutrition and it contributes to birth spacing. Maximum protection is afforded by exclusive breastfeeding. Artificial milk or even water will reduce the protective effect on mortality. Despite the high prevalence of breastfeeding in many developing countries, exclusive breastfeeding is often not high, and the duration of breastfeeding is often below optimal (WHO Collaborative Study Team 2000, American Academy of Pediatrics 1997, Victora and others 1987). A.39 Despite the significant progress achieved in the promotion of breastfeeding and government policies favoring it (Rea 1990), breastfeeding practices in Brazil have still failed to achieve the international recommended practice of exclusive breastfeeding for 4-6 months and continued breastfeeding until 24 months (DHS 1996). Perhaps in response, the Ministry of Health is now promoting the "Baby Friendly Hospital" initiative in collaboration with PAHO/WHO and UNICEF. In the initiative, which now covers 137 main hospitals across the country, hospitals designated as "baby friendly", which requires among things elaborating a written breastfeeding policy for all staff, are eligible for a 10% premium in federal payments for deliveries. As a result, the strategy focuses on rapid results without ensuring that breastfeeding standards in the hospitals and among mothers are maintained. Two other initiatives in Brazil have had positive results: training health workers to advocate exclusive breastfeeding, and training them in the integrated management of childhood illness (IMCI) strategy have increased the duration of breastfeeding. Immunization A.40 Immunization is one of the great public health success stories (Henderson 1998). The basic vaccines available to combat six major diseases in children include tuberculosis, poliomelitis, diphteria, pertussis, tetanus, and measles (WHO 1996). New vaccines have recently been introduced, including the vaccines for hepatatis B and Haemophilus influenza type b (Hib), whose effectiveness was demonstrated in studies in Gambia and Chile. (Viviani and others 1999, Levine and others 1999, Mulholland and others 1997, Mullohand and others 1999) A.41 In Brazil, the Expanded Program of Immunization, established in the seventies, includes the six basic vaccinations, plus vaccines against hepatitis B and Hib. In some states the triple viral against measles rubella and mumps is given instead of measles alone. Coverage for the all the basic vaccinations was estimated in 1996 at 72.5% of children, ranging from 60.7% in the Northeast to 87.1% in the South of the country (DHS 1996). Brazil has eradicated polio and is launching a measles immunization campaign to eliminate that disease by the end of this year. A.42 Continuing efforts to maintain and improve the coverage for the basic vaccinations is essential to keep those diseases under control. In addition, vaccines against Hib and hepatitis, and the triple viral vaccine are still in the early implementation phase, and national coverage is low. 68 Management of Illness A.43 The appropriate case management of illness has been a cornerstone of childhood mortality reduction, particularly with respect to diarrhea and pneumonia. It is well known that most diarrhea cases can be treated with oral rehydration therapy (ORT), and caretakers can learn the basic rules for its home treatment. Similarly, a meta-analysis of interventions for managing pneumonia in community settings has shown that appropriate antibiotic treatment in developing countries can reduce infant deaths by 35% and under-five mortality caused by respiratory infections by 53% (Sazawal and Black 1992). A.44 Recently, based on research that there is significant clinical overlap among several major childhood illnesses, WHO and UNICEF developed a new strategy, called the integrated management of childhood illness (IMCI) This new strategy addresses the case management of diarrhea, pneumonia, malaria, measles and malnutrition and calls for their appropriate, and when needed, combined, treatment and prevention with nutrition, immunization, disease prevention, and health promotion. It advocates counseling and health seeking behavior and has demonstrated cost savings and increased efficiency (English and others 1996, Gove and others 1997, Tulloch 1999, Boulanger and others 1999). A.45 Although in Brazil mortality from diarrhea has declined significantly, the rate remains nine times higher in the Northeast than in the South. Recent improvements are attributable to several factors, including improved case management and oral rehydration therapy, water supply, vaccine coverage, breastfeeding duration, and a decline in malnutrition prevalence. (Victora and others 1996) Pneumonia deaths have not declined as rapidly. During site visits, it appeared that health workers in many facilities in Brazil were not aware of appropriate pneumonia case management and education materials, and did not teach mothers to recognize symptoms, such as fast and difficult breathing. A.46 The IMCI approach is being introduced in a few states in Brazil, particularly in the Northeast. Preliminary results find improvements in case management in health facilities and a positive effect from nutritional counseling on the nutritional status of children (Santos 2000, Governo de Pernambuco 1998). The community component of IMCI has been limited so far. It could be closely linked with the community health agents program (described in Chapter 3) established in several states. A.47 It is estimated that in the North and Northeast the full implementation of IMCI would reduce over 20% of under five mortality through improved case management of diarrhea, pneumonia, malaria and severe infections (reducing 14.4% of all under five deaths), breastfeeding promotion (5.3%), vitamin A supplementation (1.5%), and nutrition counseling (1.25%) (Victora 2000). 69 Nutrition Interventions A.48 Nutrition interventions can enhance childhood physical growth, which has positive effects on morbidity and mortality, and can improve performance in school achievement and work capacity. Good nutrition assists the immune response to infectious diseases and promotes psychological development. Nutritional interventions after delivery include supplementary feeding of pregnant and lactating women (described above); supplementary feeding for children under five; nutrition education to improve breastfeeding and complementary feeding; correction of micronutrient deficiencies; and growth monitoring and promotion (Pellettier and others 1993, Martorell 1995). A.49 Supplementary Feeding for Children Under Five. In micro-level, community-based efficacy trials supplementary feeding has enhanced childhood physical growth in Indonesia, Jamaica, Guatemala (Husiani and others 1991, Walker and others 1991, Rivera and others 1995). Evaluations of large-scale food supplementation programs, however, are more controversial because the supplements sometimes do not reach the intended targets (Beaton and Ghassemi 1982). One meta-analysis found that of nine trials reviewed, four showed significant impact on physical growth, whereas the remaining five did not include information to verify that the supplementation actually reached the target children (Habicth and Butz 1979). A.50 Nutrition Education. Nutrition education can promote children's physical growth (WHO 1998). Evidence from a study in Bangladesh showed that nutrition education provided by community volunteers resulted in weight gains an average of 460 grams greater than gains among children in control villages (Brown and others 1992). A review of data from 12 developing countries, found that successful nutrition education programs included appropriate messages through interpersonal communication by local workers with reinforcement through mass media (Asworth and Feachem 1985). Another review of five efficacy trials and sixteen effectiveness evaluations identified factors associated with program success: nutrition education programs should address the changing needs of the infant and young child, build on current local practices, and describe not only what, but also how to feed infants (Caulfield and others 1998). A.51 Correction of Micronutrient Deficiencies. Micronutrients, including iodine, iron, vitamin A, and zinc, can reduce the risk of morbidity and mortality. A meta-analysis of eight mortality trials established that improving the vitamin A status of children aged six month to five years can reduce overall mortality rates by 23% (Beaton and others 1994). In Brazil vitamin A supplementation reduced severe episodes of diarrhea by 20% (Barreto and others 1994). A double-blind controlled trial in urban India showed that zinc supplementation resulted in a 50% reduction in respiratory morbidity in infants and preschool children (Sazawal and others 1998). As with food supplementation, however, national or regional programs cannot replicate the conditions of these controlled trials, and the targeted beneficiaries do not easily obtain the micronutrients intended for them. A.52 Growth Monitoring and Promotion. This is a controversial intervention. Experiences from Zaire and India show that the use of a growth chart does not have additional benefits over 70 other educational interventions in the rural community (Gerein and Ross 1991, George and others 1993). Women with limited literacy skills have had difficulty understanding growth monitoring charts in Brazil (McAuliffe and others 1993). When the term "growth monitoring and promotion" also includes counseling to motivate the caretaker, the intervention can be beneficial as an educational tool (Griffiths and others 1996), though it does not seem to make nutritional supplement programs more cost-effective (Ruel 1995). A.53 Nutrition supplementation programs in Brazil have had limited impact and have been costly (Victora 2000, Musgrove 1990). The federal government currently finances milk and soy milk supplements for pregnant women and malnourished children (described in Chapter 3). Although detailed evaluations of it are not available, there are reasons, on the basis of discussions with and evidence received from state coordinators, to question its impact on health status. Apparently, mothers often use the supplements for their entire family and use them to replace their household purchases. The coverage of growth monitoring activities appears to be low in Brazil. Only 11.4% of all children had their weight recorded on their growth charts in the 2 months preceding the 1996 DHS survey, and evaluations of the monitoring were not positive. The federal government also has programs to prevent iodine, iron and vitamin A deficiencies, and one to control iron deficiency anemia. A.54 There is evidence of a favorable trend in nutritional status among children in Brazil. As a result, the estimated impact of food supplementation programs on overall child mortality is low (Victora 2000). Still, given the regional variation in malnutrition prevalence - stunting, wasting and underweight prevalence are almost three times higher in the Northeast than in the South - the beneficial effects of an appropriate nutritional status in children would seem to call for continuing micronutrient programs, improving their targeting in the North and Northeast, and combining them with nutrition education. Preliminary results from the IMCI nutrition counseling module in Southern Brazil showed positive results for mothers' knowledge and behavior, children's diet, and child weight gain. Care for Development A.55 For children, early physical and psychological development is critical. Although the mechanisms, and the interactions between mind and body, are not entirely understood, early development influences educational achievement, and later in life, work performances (Gorman 1995, Meeks-Gardener and others 1995). A.56 A recent review of the benefits of investing in early childhood care and development programs, as well as programs to improve childhood physical growth, suggests that combined interventions to improve both physical growth and psychological development have the greatest impact when jointly targeted at disadvantaged populations. It contends that it is possible that very early interventions - prenatal, during infancy, and early childhood-are likely to have the greatest impact (WHO 1999). 71 A.57 Programmatic implications are not easy to define. A model for combined interventions needs to be developed, and more research is necessary. Accident Prevention A.58 Prevention of injuries in children is a growing concern for many countries. In Brazil, car accidents and drowning are the most common causes of deaths in this category. However, interventions to prevent injuries are complex, by nature multi-sectorial, and the experience to date is difficult to summarize in order to identify best practices (Dowd 1999, Durbin 1999). 72 APPENDIX B: TARGETING AND IMPACT OF THE FAMILY HEALTH PROGRAM Targeting B.59 The federal and state governments attempted to select municipalities with high indices of social and medical needs when deciding where to encourage the establishment of the first PSF programs. Because establishing a PSF requires local management and municipal counterpart financing, however, it is possible that municipalities with greater managerial capacity, often associated with higher income and fewer social needs, would be more likely to take advantage of the opportunity. Therefore, it is not clear whether the additional funds associated with the PSF would go to more needy municipalities, or whether the reverse would be true. B.60 Table 12 below addresses the targeting question by comparing human development indicators for municipalities with and without PSF programs, using data from the 1991 census, subject to data availability.'6 The table shows that in the population of municipalities for which data were available, municipalities with PSF programs were not, according to these measures, significantly different from those without PSF programs, with one important exception: municipalities with PSF programs had higher infant mortality rates (in 1991). This might be the result of better mortality surveillance where the PSF is implemented, but it is also consistent with the emphasis that the government's Comunidade Solidaria placed on addressing infant mortality when the programs were being established (Viana 2000). When the sample is restricted to small municipalities, however, the differences are sharper: municipalities with PSF programs have lower per capita income, a greater percentage of people below the poverty line, a higher infant mortality rate, and more adult illiteracy. In other words, in the smaller municipalities the differences appear to be structural. In the Northeast, however, where the coverage of the PSF program is wider than in other regions, the differences between PSF and non-PSF municipalities are not as systematic. Northeast municipalities with PSF programs actually have higher mean incomes and lower mean poverty rates than those without the PSF; but small PSF municipalities in the Northeast have higher illiteracy rates, and all Northeast PSF municipalities have significantly higher official infant mortality rates. Although the higher infant mortality rates might be the result of better surveillance, the higher illiteracy rates probably are not. The inference, then, is that although there does appear to be some level of effective targeting in the Northeast, based on relative infant mortality and illiteracy rates, the absence of minimal municipal capacity appears to be preventing some very poor municipalities from implementing the program. 16 More recent data on municipal income and social characteristics were not available; but because poverty is structural in many municipalities in Brazil, income and other social characteristics of most municipalities are not likely to have changed significantly, relative to other municipalities, since 1991. When new municipalities did not have data, whether because they were created after 1991 or for any other reason, observations were dropped. 73 B.61 It should be noted that "targeting" is here understood only in the limited sense that municipalities with the program are, on average, worse off: in reality there is a great deal of similarity in social conditions among municipalities with and without the PSF, as the large standard deviations in the table demonstrate. It is also true, moreover, that this analysis of targeting is relevant if the PSF is understood to be a program that aims to improve health care in the municipalities that are worse off, which was the idea when the program was first conceived. Since then, however, the PSF has become a model for the delivery of health care in all municipalities in Brazil. The targeting question might well decline in importance as the program's coverage increases. It remains to be seen whether delivering all primary health care through the PSF will be a useful model for large municipalities with greater resources. Some analysts claim that is better suited for smaller, poorer municipalities where there no exist options to by-pass the PSF team and go directly to other facilities. Whether or not that is true, it is the case that currently poorer municipalities utilize community health care agents, including the PSF and PACS, to a greater extent than richer municipalities. In a regression not shown here, mean municipal income was found to be a highly significant explanator of, and negatively correlated with, the fraction of primary care delivered through the PACS in both large and small municipalities (Pearson correlations of -0.28 and -0.43, respectively). In other words, the poorer the municipality, the more its primary care is being delivered through PACS. 74 Table 14: Targeting of the PSF Municipalities, Mean per Mean poverty Mean Mean adult Mean access 1998 capita rate, percent, registered illiteracy, to safe water, income, 1991 infant percent, 1991 percent, 1991 salarios mortality minimos, rate, 1991 1991 With PSF 0.732 63.6 57.9 30.1 72.0 (.462) (21.1) (36.1) (16.4) (24.3) n=849 n=849 n=849 n=849 n=846 Without PSF 0.724 62.6 51.0 29.6 70.6 (.424) (22.0) (31.2) (16.4) (25.6) n=3611 n=3611 n-=3611 n=3611 n=3559 Twotailedt= 0.49 1.15 5.56 0.73 1.42 With PSF: 0.607 68.7 60.7 33.0 69.5 population less (.321) (17.9) (37.6) (16.1) (25.1) than 45,000 n=649 n=649 n=649 n=649 n=646 Without PSF: 0.685 64.3 51.5 30.5 69.7 population less (.388) (21.1) (31.6) (16.4) (26.0) than 45,000 n=3244 n=3244 n=3244 n=3244 n=3192 Two tailed t= -4.78 4.99 6.55 3.56 -0.19 With PSF: 0.407 80.7 100.0 47.2 54.3 Northeast (.206) (11.1) (28.6) (11.7) (23.2) n=283 n=283 n=283 n=283 n=280 Without PSF: 0.364 83.3 84.8 47.2 53.2 Northeast (.163) (9.4) (27.6) (10.4) (23.4) n=1206 n=1206 n:=1206 n=1206 n=1157 Two tailed t= 3.74 -4.04 8.28 0.01 0.71 With PSF: 0.357 83.3 102.9 49.9 51.9 Northeast and (.135) (9.2) (28.3) (10.1) (23.6) population less n=229 n=229 n-229 n=220 n=226 than 45,000 Without PSF: 0.341 84.5 85.5 48.3 52.2 Northeast and (.122) (7.7) (27.8) (9.3) (23.5) population less n=1092 n=1092 n-=1092 n=1092 n=1043 than 45,000 Two tailed t 1.67 -2.19 8.60 2.28 -0.16 Source: Author's calculations based on data from MS, IBGE, and background paper (Vianna 2000). Standard deviations are in parentheses. A two-tailed t-test is used to determine if the mean values of the indicators are different for municipalities with and without the PSF. Municipal population data are based on the 1991 census. 75 Impact B.62 Observations from field visits and from the focus groups found that despite the issues and problems enumerated above, the PSF is having a positive impact in several locales. Accounts like this one from a woman in Lauro de Freitas, Bahia, were repeated several times: Things got better. We have a doctor here that we can see here, directly. We've also got a community agent and didn't have one before. Now we've got these community health agents that are always thinking about our kids, visiting us in our houses. There's this doctor who sees people here in the church, she's a great person (Trad and Bastos 2000). B.63 Several studies have contended that improvements in services underlie positive evaluations like the one above, and that better municipal health care indicators are associated with the PACS/PSF. Case studies presented at the National Conference on Family Health in November, 1999, showed dramatic impact on registered infant mortality rates, vaccination coverage, childhood nutrition, and hospitalization rates in several municipalities, including Campina Grande in Paraiba, Camaragibe in Pernambuco, Sobral in Ceark, and Sao Jorge do Patrocinio in Parana (Ministerio de Sauide 1999, Alves 1998). A qualitative evaluation of the PACS in eight Northeast states (all but Ceara) found that families that knew of agentes and accepted them into their homes had on average received more prenatal care, had higher vaccination rates, used oral rehydration for diarrhea more often, and were more likely to filter their water, than families that did know about PACS (Ministerio de Saude 1994). The MS recently published results from a comprehensive evaluation of the PSF in which it surveyed the 1,219 municipalities that had implemented the program as of December, 1998 (Ministerio de Sauide 2000). It found that the adoption of the PSF led to dramatic increases in the frequencies with which municipalities provided a large number of health promotion and prevention activities in the communities covered, including prenatal care, family planning, cervical cancer screening, and diabetes and high blood pressure monitoring. Finally, a series of evaluations of seven municipalities that had adopted the PSF found substantially lower hospitalization rates in some municipalities but little impact in others (Viana and Dal Poz 1999). The authors' hypothesized that the embeddedness of a municipal PSF program in a larger state-wide strategy considerably assists municipal programs. B.64 Ideally an impact study would take the form of a two-group, before-and-after assessment, comparing changes in outcomes in municipalities that have the PSF to changes in those outcomes over the same period of time in municipalities that do not have the program. That design methodology would control for some of the unobserved variables that complicate interpretations of one-group, before-and-after studies and cross-sectional studies like those described above. The table below presents results from the first attempt to conduct such an assessment of the PSF. It compares changes in measures of health system performance and health outcomes between 1995 and 1998 for municipalities that had and did not have the PSF. Four caveats should be noted. First, some categories of data were not available for large numbers of municipalities (for instance, only six municipalities in all of Minas Gerais reported data on immunizations). The analysis below simply drops observations for which variables are missing. Second, the data lend themselves to an analysis of whether differences in rates of change existed among municipalities that had and did not have the PSF in 1998. How long they had the program (it is relatively new in most municipalities), as well as the size of the population covered, are not considered (though a few additional regressions, mentioned below, attempt to address the latter issue). Third, due to issues related to data quality and the decentralization and fragmentation of 76 data collection, health status measures (here registered infant mortality rates) were obtained only for the state of Pemambuco. Fourth, whether or not a municipality has a PSF program is endogenous, so the findings below might be the result of characteristics that led the municipalities to adopt the program and not to the program itself. Instrumental variables might address this concern, but it is hard to find municipal characteristics associated with the adoption of the PSF that would not also be associated with the outcome measures. B.65 Table 15 below shows that, on average, immunization rates, per capita hospital admissions, and per capita hospital admissions for children under 5 rose, while per capita hospital admissions due to perinatal causes and hospital mortality rates for children under 5 declined in municipalities both with and without the PSF between 1995-1998. It is not clear, a priori, whether the PSF should be associated with higher or lower hospital admissions rates - there might be one effect that increases hospitalizations from diagnosing more people, and an effect in the opposite direction from preventing illness. The table's figures for hospitalization rates are also somewhat deceptive for two reasons. First, national statistics show that hospitalization rates in almost all categories declined during the period, but the figures here are average municipal-level changes, which are positive because there were increases in a large number of small municipalities. Second, note also that the standard deviations are large: hospitalizations per capita, both generally and for children under five, actually declined in slightly more than half of the municipalities. The data show that per capita hospital admissions increased more slowly in municipalities with PSF: the t-values for the difference in the means approaches significance at the 5% level in both the overall sample and in the Northeast, and the t-values are significant when the sample is restricted to municipalities with less than 17,000 inhabitants (not shown in the table). In the overall sample, municipalities without PSF experienced steeper declines in hospitalization for perinatal causes than municipalities with the PSF, but this appears to be an effect of the largest municipalities (over 400,000 inhabitants): when those are dropped from the sample, the differences become statistically insignificant. Surprisingly, it appears that the PSF is associated with relatively higher average hospital mortality rates for children under five, particularly in the Northeast. This might be because the PSF programs are locating more ill children and helping them get in to hospitals when they would otherwise have died at home. Most strikingly, mean infant mortality rates fell 13% from 1995-98 in the 33 municipalities in Pernambuco that had the PSF but rose nearly 17% in the 98 municipalities that did not have the PSF. The difference is significant at the 10% level. Given that registered infant mortality rates for small populations can be unreliable, this result should be interpreted with caution. Finally, additional regressions, not shown here, found that in small municipalities with the PSF or PACS, the number of PSF or PACS activities, per capita, is significantly associated with lower overall hospital admissions. B.66 Summarizing, this impact analysis finds some evidence that the PSF is reducing infant mortality rates, and some evidence that it is reducing hospital admissions while also getting more ill children into hospitals. Although the PSF remains a relatively new program and the conclusions at this stage are preliminary, there reasons to believe that the PSF can improve health care and health status. 77 Table 15: Impact of the PSF Municipalities, Mean percent Mean Mean Mean percent Mean percent Mean percent 1998 increase in per percent percent increase in per Increase in increase in capita total increase in increase in capita hospital hospital official infant immunizations, per capita per capita admissions for mortaliy rate mortality rate, 1998-95 hospital hospital perinatal of children 1998-95 admissions, admissions of causes, 1998- under 5, (Pernambuco) 1998-95 children 95 1998-95 under 5, 1998-95 With PSF 86.5 1.6 12.5 -39.8 -0.42 -13.3 (136.5) (57.0) (96.1) (82.1) (.60) (46.3) n=480 n=633 n=662 n=554 n=304 n=32 Without PSF 82.0 6.8 17.3 -49.7 -6.3 16.6 (122.7) (64.4) (104.9) (72.5) (.61) (93.2) n=2754 n=2443 n=2629 n=2110 n=841 n=89 Two tailed t 0.70 -1.87 -1.04 -2.79 1.45 -1.73 With PSF: 98.1 3.2 14.1 -48.5 2.3 population less (149.9) (63.0) (98.5) (77.5) (65.4) than 45,000 n=336 n=475 n=463 n=403 n=145 Without PSF: 83.7 8.1 18.1 -52.6 -6.1 population less (123.6) (66.1) (108.1) (72.6) (66.5) than 45,000 n=2250 n=2097 n=2127 n=1746 n=533 Two tailed t= -1.93 -1.45 -0.72 0.98 1.35 With PSF: 79.5 6.8 22.0 -38.8 -10.8 Northeast (152.6) (69.9) (97.6) (87.9) (67.3) n=275 n=221 n--203 n=164 n=73 Without PSF: 82.6 15.8 26.7 -45.7 -11.3 Northeast (125.4) (68.9) (119.3) (79.8) (63.8) n=1205 n=802 n=747 n=593 n=240 Twotailedt -0.36 -1.72 -0.51 0.95 2.56 With PSF: 89.1 8.1 22.0 -54.3 15.1 Northeast and (164.3) (78.1) (95.8) (75.3) (66.7) population less n=217 n=167 n=151 n=118 n=40 than 45,000 Without PSF: 84.0 17.8 29.3 -48.6 -15.8 Northeast and (126.6) (70.4) (123.4) (79.9) (68.1) population less n=1059 n=691 n=644 n=498- n=159 than 45,000 Two tailed t 0.52 -1.55 -0.69 -0.70 2.58 Source: Author's calculations based on data from MS, IBGE, and background paper (Vianna 2000). Standard deviations are in parentheses. A two-tailed t-test is used to determine if the mean values of the indicators are different for municipalities with and without the PSF. Municipal population data are based on the 1991 census. 78 REFERENCES Alisjahbana, A, Williams C, Dharmayanti D, and others. June 1995. "An integrated village maternity service to improve referral patterns in a rural area in West Java." International Journal of Gynecology and Obstetrics, 48: S83-S93. Alves, MA . 1998. "Programa de sauide da familia: Mutirao do Serrotao, Campina Grande - PB." EASP/FGV: unpublished manuscript. Albemaz, E, Giuliani ERJ, Victora CG. 1998. "Supporting breastfeeding: a successful experience." JHuman Lactation, 14283 - 5. American Academy of Pediatrics Policy Statement. Pediatrics 1997. "Breastfeeding and the use of human milk.", 100(6): 1035 - 39. Arilha, M, Barbosa RM. 1998. "Cytotec in Brazil: 'At least it doesn't kill." Outlook, 16:2. Ashworth, A, Feachem RG. 1985. "Interventions for the control of diarrhoeal diseases among young children: weaning education." WHO, 63: 1115 - 27. Ashworth, A, Feachem RG. "Interventions for the control of diarrhoeal diseases among young children: growth monitoring programmes." Bang RA, Baitule SB, and others. 1999. "Home-based neonatal care and management of sepsis to reduce neonatal mortality: a field trial in rural India." Lancet, 354(9194):1955-61. Barclay, AJG, Foster A, Sommer A. 1987. "Vitamin A supplements and mortality related to measles: a randomised clinical trial." British Medical Journal, 294: 294 - 6. Barreto, ML, Santos LM, Assia AM, and others. 1994. "Effect of vitamin A supplementation on diarrhea and acute lower respiratory tract infections in young children in Brazil." Lancet, 344 :228-31. Barros, FC, Vaughan JP, Victora CG, Huttly, SRA. 1991. "Epidemic of caesarean sections in Brazil." Lancet, 338: 167 - 9. Barros, FC, Vaughan JP, Victora CG . 1986. "Why so many caesarean sections? The need for a further policy change in Brazil." Health Policy and Planning 1: 19 - 29. 79 Barros E. 2000 . "A politica de aten,cao a saiude de mulheres e crian,as no Brasil: algumas informac6es basicas." Background paper commissioned for this study. Beaton, GH, Martorell A, Aronson K, and others. 1994. "Vitamin A supplementation and child morbidity and mortality in developing countries." Food and Nutrition Bull, 15(4):182 - 9. Beaton, GH, Ghassemi H. 1982. "Supplementary feeding programs for young children in developing countries." American Journal of Clin Nutrition. 35: 863 - 916. Bergman, NG, Jurisoo LA. 1994. "The Kangaroo method" for treating low birth weight babies in a developing country. Trop Doct, 24(2): 57 - 60. "Breastfeeding counseling: a training course." WHO/CDR/93.3-6 and UNICEF/NUT/93.1 - 4. Billings, DL, Ankrah V, Baird TL, Taylor JE. 1999. "Ababio KPP and Ntow S. Midwives and comprehensive postabortion care in Ghana." In: Huntington D, Piet-Pelon NJ, eds. Postabortion Care: Lessons from Operations Research, Population Council, NY. 141-158. Boulanger, LL, Lee LA, Odhacha A. 1999. "Treatment in Kenyan rural health facilities: projected drug costs using the WHO-UNICEF Integrated management of childhood illness (IMCI) guidelines." WHO Bull, 77(10): 852 - 8. Brown, LV, Zeitlin MF, Peterson KE, and others. 1992. "Evaluation of the impact of weaning food messages on infant feeding practices and child growth in rural Bangladesh." American Journal of Clin Nutrition, 56: 994-1003. Brown, KH, Creed-Kanashiro H, Dewey KG. 1995. "Optimal complementary feeding practices to prevent childhood malnutrition in developing countries." Food and Nutrition Bull, 16(4): 320 - 38. Campbell, 0, Koblinsky M, Taylor P. June 1995. "Off to a rapid start: appraising maternal mortality and services." International Journal of Gynecology and Obstetrics, 48 : S33-S52. Campos, FE. 1997. "Educacao Medica no Brasil." Paper commissioned by the Operations Evaluation Department, The World Bank. Carter, GM, Newhouse JP, and Relles DA. 1990. "How Much Change in the Case Mix Index Is DRG Creep?" Journal of Health Economics, 9: 411-28. Ceesay, SM, Prentice AM, Cole TJ and others. 1997. "Effects on birth weight and perinatal mortality of maternal dietary supplements in rural Gambia:" 5 year randomized controlled trial. British Medical Journal, 315 : 786 - 90. Charpak, N, Ruiz-Pelaez JG, 1994. "Charpak Y. Kangaroo Mother Programme: an alternative way of caring for low birthweight infants?" One year mortality in a two cohort study. Pediatrics 94: 804 - 10. Correa S, Piola S, Arilha M. October 1998. "Reproductive Health in Policy and Practice Brazil." Population Reference Bureau. 80 Cattaneo, A, Davanzo R, Worku B, and others. 1998. "Kangaroo mother care for low birthweight infants:" a randomized controlled trial in different settings. Acta Paediatrica, 87(9) : 976-85. Caulfield, LE, Huffman S, Piwoz E. 1998. "Interventions to improve complementary food intakes of 6-12 month old infants in developing countries: impact on growth, prevalence of malnutrition, and potential contribution to child survival." Academy of Educational Development Linkages Progrm. Washington DC. Cleland, J, Bicego G, and Fegan G. "Socioeconomic inequalities in childhood mortality." Health Transition Review. 2: 1 - 18. Dal Poz MR 2000. "Recursos humanos e o programa de saude da familia no Brasil: politica e problemas." Background paper commissioned for this study. De Brouwere, V, and others. 1998. "Strategies for reducing maternal mortality in developing countries: what can we learn from the hisotry of the industrialized West?" Tropical Medicine and International health 3(10): 771-82. De Quadros, CA, Olive' JM, Nogueira C, and others. 1998. "Expanded Program on Immunization" (EPI). "In: Maternal and child health activities at the local level. Washington: Pan American Health Organization. Diaz, M, Simmons R, Diaz J and others. 1999. "Expanding contraceptive choice: findings from Brazil." Studies in Family Planning, 30(1): 1 - 16. Dowd, MD. 1999. "Childhood injury prevention at home and play." Current Opininion in Pediatrics. 11: 578 - 82 Durbin, DR. 1999. "Preventing motor vehicle injuries." Current Opinion in Pediatrics. 11:583-7. Ellertson, C, Winikoff B, Armstrong E and others. 1995. "Expanding access to emergency contraception in developing countries. Studies in Family Planning. 26: 5 : 251-263. English M and others. 1996. "Clinical overlap between malaria and severe pneumonia in African children in hospital." Trans Royal Soc Trop Med Hyg 90: 658 - 62. Esrey, SA, Feachem RG, Hughes JM. 1985. "Interventions for the control of diarrheal diseases in young children: improving water supply and excreta disposal facilities." Bulletin of the World Health Organization 63 : 757 - 72. Estado de Goias. 1999. "Pacto dos 1.000 dias a favor da saude da mulher, da crianca e do adolescente, no estado de Goias." Estado de Goias, Secretaria de Estado da Sauide, Superintendencia de Acoes Basicas de Saude. Filmer, D, Pritchett, L. 1999. "The impact of public spending on health: does money matter?" Social Science and Medicine 49: 1309-23. 81 Funkhouser, A, Steinhoff MC, Ward J. 1991. "Heamophilus influenza disease and immunization in developing countries." Review of Infectious Diease 13(suppl 6): S542 - 54. Gaffikin, L, Blumenthal PD, Davis C, Brechin SJG. 1997. "Alternatives for cervical cancer screening and treatment in low-resource settings": Proceedings of the JHPIEGO Board of Trustees Workshop held 21 - 22 May 1997. JHPIEGO Corporation. Baltimore, Maryland. Galloway, R. Feb. 2000. "Improving the quality of iron supplementation programs: the MotherCare experience", MotherCare. John Snow, Inc.:Washington. Gallup, JL, Sachs JD. 1998. "The Economic Burden of Malaria." Harvard Center for International Development: unpublished manuscript. Galvao, L, Diaz J, Diaz M, and others. 1999. "Emergency contraception: knowledge, attitudes and practices among Brazilian obstetrician-gynecologists." Family Planning Perspectives 25(4): 168 - 171. Gerein, NM, Ross DA. 1991. "Is growth monitoring worthwhile? An evaluation of its use in three child health programmes in Zaire." Soc Sci Med, 32: 667 - 75. George, SM, Latham MC, Abel R, and others. 1993. "Evaluation of effectiveness of good growth monitoring in South Indian villages." Lancet. 342 : 348 - 52. Glewwe, Paul. 1999. "Why does mother's schooling raise health in developing countries? Evidence from Morocco." The Journal of Human Resources 34(1): 124-59. Griffiths, M, Dickin K, Favin M. 1996. "Promoting the growth of children: what works". The World Bank. Gorman, KS. 1995. "Malnutrition and cognitive development: evidence from experimental/quasi experimental studies the mild-to-moderate malnourished." Journal of Nutrition 125(8S) : 2239S - 44S. Gove, S. and others. 1997. "Integrated management of childhood illness by outpatient health workers: technical basis and overview." WHO Bull 75(Suppl 1) . 7 - 24. Governo do Estado de Pernambuco. 1999. "Pernambuco cidadao: uma proposta de reducao da mortalidade infantil em Pernambuco." Governo do Estado de Pernambuco, Secretaria de Saude. Governo de Pemnambuco. 1998. "AIDPI Atencao Integrada as Doencas Prevalente na Infancia. Processo de implantacao da estrategia em Pernambuco." Recife. Habicth, JP, Butz WP. 1979. "Measurement of health and nutrition effects of large-scale nutrition intervention projects." In: Klein Re and others. "Evaluating the impact of nutrition and health programs." Plenum Publishing Corporation 133 - 82. 82 Henderson, RH. 1998. "Immunization: going the extra mile. In: The progress of nations." New York, UNICEF. Husaini, MA, Karyadi L, Sandjaja KD, and others. 1991. "Developmental effects of short term supplementary feeding in nutritionally at risk Indonesian infants." American Journal of Clinical Nutrition 54: 799 - 804. Janowitz, B, Wallace S, Araujo G, Araujo L. July 1985. "Referrals by traditional birth attendants in Northeast Brazil", American Journal of Public Health 75(7) : 745 - 748. John Snow Inc. 1998. "Assessing safe motherhood programs: India and Bangladesh." MotherCare Matters 7: 1. John Snow Inc. 1997. "Learning and action in the first decade: The MotherCare experience." MotherCare Matters, Special edition, 6 : 4. Kambarami, RA, Chidede 0, Kowo DT. 1998. "Kangaroo care versus incubator care in the management of well preterm infants - a pilot study." Annals of Tropical Pediatrics 18(2): 81 - 6. Koblinsky, MA, Campbell 0, Heichelheim J. 1999. "Organizing delivery care: what works for safe motherhood?" Bulletin of the WHO 77(5) : 399 - 406. Kramer, MS. 1993. "Effects of energy and protein intakes on pregnancy outcome: an overview of the research evidence from controlled clinical trials." American Journal of Clinical Nutrition 58 : 627 - 35. Lennox, CE, Kwast BE, Farley TM. 1998. "Breech labor on the WHO partograph." International Journal of Gynecology and Obstetrics 62(2): 117 - 27. Levine, OS, Lagos R, Munoz A, and others. 1999. "Defining the burden of pneumonia in children preventable by vaccination against Haemophilus influenzae type b." Pediatric Infectious. Disease Journal. 18: 1060 - 4. Lewis M, Eskeland GS, Traa-Valerezo X. 1999. "Challenging El Salvador's rural health care strategy." Unpublished manuscript, The World Bank. Liljestrand, J. 1999. "Reducing perinatal and maternal mortality in the world: the major challenges." British Journal of Obstetrics. and Gynecology 106: 877 - 880. Ludwig, H. 1994. "Women and cancer." International Journal of Gynecology and Obstetrics 46: 195 - 202. Lutter, CK, Perrez-Escamilla R, Segall A, and others. 1997. "The effectiveness of a hospital based program to promote exclusive breastfeeding among low-income women in Brazil." American Journal of Public Health 87(4):659 - 63. Maine, D, Rosenfield A. 1999. "The safe motherhood initiative: why has it stalled?" American. Journal of Public Health 89(4): 480 - 482. 83 Maine, D. 1991. "Safe motherhood programs: Options and issues." Centerfor Population and Family Health. Columbia University: NY. Martorell, R. 1995. "Promoting healthy growth: rationale and benefits. In: Pinstrup P, Pelletier D, Aldman H eds. Child growth and nutrition in developing countries." Ithaca, NY Cornell University Press. McAuliffe, JF, Falcao L, Duncan B. 1993. "Understanding of growth monitoring charts by literate and illiterate mothers in Northeast Brazil." Tropical Pediatric Journal 39 : 370 - 2. Medici, AM. 1998. "Uma Decada de SUS (1988-1998): Progressos e Desafios." Unpublished manuscript. Meeks, -Gardner JM, Grantham SM, Chang SM, and others. 1995. "Activity and behavioral development in stunted and non-stunted children and response to nutritional supplementation." Child Development 66: 1785 - 97. Ministerio de Sauide. 2000. "Avaliacao da Implamantacao e Funcionamento do Programa de Sauide da Familia- PSF." Ministerio de Sauide, Secretaria da Assistencia da Sauide, Coordenacao de Aten,ao Basica: Brasilia. Ministerio de Sauide. 1994. "Avaliacao Qualitativa do PACS. " Ministerio de Sauide, Fundacao Nacional de Saudde, Programa de Agentes Comunitarios de Saudde. Brasilia DF. Ministerio de Sauide. 1999. "Livro de Resumos: Primeira Mostra Nacional de Producao em Saiide da Familia." Brasilia DF. Monteiro, CA, D'Aquino Benicio MH, and Martins de Freitas IC. 1997. "Changes in poverty-related health indicators in Brazil: causes and impact on regional inequities." Paper commissioned by the Operations Evaluation Department, The World Bank. Monteiro, CA, Benicio MH, lIunes R, and others. 1992. "Nutritional status of Brazilian children: trends from 1975 to 1989." WHO Bull 70: 657 - 66. Mosley, WH, Chen LC 1984. "An analytical framework for the study of child survival in developing countries," in Mosley WH, Chen LC. "Child survival: strategies for research." Population and Development Review 10(supplement): 25 - 48. Mullholland, K, Hilton S, Adegbola R, and others. 1997. "Randomised trial of Haemophilus influenza type b for prevention of pneumonia and meningitis in Gambian infants." Lancet 349: 1191-7. Mulholland, K. 1998. "Serious infections in young infants in developing countries." Vaccine, 16: 1360 - 2 Mullholland, K, Levine 0, Nohynek H, and others. 1999. "Evaluation of vaccines for the prevention of pneumonia in children from developing countries." Epidemiology Review 21 43 - 55. 84 Musgrove, P. 1990. "Do nutrition programs make a difference? The case of Brazil." International Journal of Health Services 20 : 691 - 715. Narayanan, I, Prakash K, Murthy NS, at al. 1984. "Randomised controlled trial of effect of raw and holder patourized human milk and of formula supplements on incidence of neonatal infection." Lancet 2(8412): 1111- 3. "National Center for Health Statistics". 1999. Health, United States. Nations MK. 2000. "'Inventing' or 'inverting' health care quality? Popular perspectives on family health program (PSF) in Northeast Brazil." Background paper commissioned for this study. "Nuicleo de Estudos de Politicas Puiblicas". 1999. Avaliacdo qualitativas de programas sociais prioritarios. Campinas: Universidade de Campinas. PAHO, 1999 "Health in the Americas", 1998 Edition. Washington: Pan American Health Organization. Palme-Kilander. 1992. "Methods of resuscitation in low-Apgar score newborn infants - a national survey." Acta Paediatrica 81: 739 - 44. Pearson, R. 1995. "Thematic evaluation of UNICEF support to growth monitoring." Working paper series n 2. New York, UNICEF. Pellettier and others. 1993. "Epidemiological evidence for a potentiating effect of malnutrition on child mortality." American Journal of Public Health. 83: 1130 - 33. Rea, MF, Venancio SI, Martines JC, and others. 1999. "Counseling on breastfeeding: assessing knowledge and skills." Bulletin of WHO 77: 492 - 8. Rea, MF. 1990. "The Brazilian National Breastfeeding Program: a success story." International Journal of Gynecology and Obstetetrics 3 1(suppl 1): 79 - 82. Ravallion, M, van de Walle D. 1991. "Urban-rural cost-of-living differentials in a developing economy." Journal of Urban Economics 29: 113 - 127 Rede Nacional Feminista de Saude e Direitos Reprodutivos. 1999. "Adolescentes: Sauide Sexual e Reprodutiva", Dossier, Brasil. Rose, Elaina. 1997. "Health in Brazil: results from the Brazil Living Standards Measurement Survey." Paper commissioned by the Operations Evaluation Department, The World Bank. Rivera, JA, Martorell R, Ruel MT, and others. 1995. "Nutritional supplementation during the preschool years influences body size and composition of Guatemalan adolescents." Journal of Nutrition 125:(4S): 1068 - 77. 85 Rooks, J, Winikoff, B. February 12 - 13, 1990. "A reassessment of the concept of reproductive risk in maternity care and family planning services." Proceedings of a seminar presented under the Population Council's Robert H. Ebert Program on Critical Issues in Reproductive Health and Population. New York. Ruel, MT. 1995. "Growth monitoring as an educational tool, an integration strategy, and a source of information: a review of experience." In: Pinstrup-Anderson P, Pellettier D, Alderman H, eds. "Child growth and nutrition. Ithaca, NY, Cornell University Press, 78 - 96. Santos, IS, Victora CG, Martines JC, and others. "Nutritional counseling increases weight among Brazilian children." Submitted for publication. Sastry, N (1996). "Community characteristics, individual and household attributes, and child survival in Brazil." Demography 33 : 211 - 29. Saugstad, OS, Rootwell T, Aalen 0. 1998. "Resuscitation of asphyxiated newborn infants with room air or oxygen: an international controlled trial." Pediatrics 102: 1-. Starrs, A. 1997. "The Safe Motherhood Action Agenda: Priorities for the next decade. Report on the Safe Motherhood technical consultation" 18-23 October, 1997, Colombo, Sri Lanka. Strauss, J, Thomas D. 1998. "Health, Nutrition, and Economic Development." Journal of Economic Literature 36: 766-817. Sazawal, S, Black RE. 1992. "Meta-analysis of intervention trials on case management of pneumonia in community settings." Lancet 340: 528 - 33. Sazawal, S, Black R, Jalla S, and others. 1998. "Zinc supplementation reduces the incidence of acute lower respiratory infections in infants and preschool children: a double-blind controlled trial." Pediatrics, 102(1): 1-5. Susser, M, Stein Z. 1994. "Timing in prenatal nutrition: a reprise of the Dutch Famine Study." Nutrition Review 52(3): 84 - 94. Trad, LA, Bastos AC. 2000. "Avaliacao qualitativa de satisfacao do usuario do programa de sauide da familia- PSF." Background paper commissioned for this study. Tulloch, J. 1999. "Integrated approach to child health in developing countries.". Lancet 354(Suppl 2) :16 - 20. Viana, AL, Dal Poz MR. 1998 "Reforna em saude no Brasil: Programa de saude da familia." UERJ: Estudos em Sazude Coletiva 166. Viana, AL and others. 2000. "Projeto avaliacao do programa de sauide da familia." Background paper commissioned for this study. Victora, C. 2000. "Potential interventions to improve the health of mothers and children in Brazil." Background paper commissioned for this study. 86 Victora, CG, Olinto MTA, Nobre LC, and others. 1996. "The recent fall in diarrhea mortality in Northeastern Brazil: did ORT play a role?" Health Policy and Planning 11:132 - 41. Victora, CG, Smith PG, Vaughan JP, and others. 1987. "Evidence for protection by breastfeeding against infant deaths from infectious diseases in Brazil." Lancet 2 : 319 - 22. Viviani, S, Jack A, Hall AJ, and others. 1999. "Hepatitis B vaccination in infancy in The Gambia: protection against carriage at 9 years of age." Vaccine 17(23-24): 2946 - 50. Wagstaff, A. 2000. "Socioeconomic inequalities in child mortality: comparisons across nine developing countries. " Bulletin of the World Htealth Organization 78(1):19-29. Walker, SP, Powell CA, Grantham McGregor SM., and others. 1991. "Nutritional supplementation, psychosocial stimulation, and growth of stunted children: the Jamaican study." American Journal of Clinical Nutrition 54: 642 - 48. West KP, Jr., Katz J. Khatry SK and others. 1999. "Double blind, cluster randomized trial of low dose supplementation with vitamin A or beta carotene on mortality related to pregnancy in Nepal." British Medical Journal 318: 570 - 5. Westphal, MF, and others. 1995. "Breastfeeding training for health professionals and resultant institutional changes." WHO 73 (4) : 461 - 8. Whittle, HC, Maine N, Pilkinton J, and others. 1995. "Long term efficacy of continuing hepatitis B vaccination in infancy in two Gambian villages." Lancet 345(8957): 1089 - 92. World Bank. 2000. "Brazil: Selected Issues in Social Protection" Report No. 20054-BR. (Confidential Draft), Brazil Country Management Unit, Latin America and the Caribbean Region, Washington DC: The World Bank Group. World Bank. 1998. "Public Expenditures for Poverty Alleviation in Northeast Brazil: Promoting Growth and Improving Services." Report No. 18700-BR. Washington DC: The World Bank Group. World Bank/PAHO Partnership for School Health in Latin America and the Caribbean. 2000. "Situational analysis of school health and nut,rition in Brazil." Unpublished paper. World Health Organization Young Infant Study Group. 1999. "Bacterial etiology of serious infections in young infants in developing countries." Pediatric Infectious Disease Journal 18: S17 - 22. World Health Organization. 1999. "Interventions for physical growth and psychological development - a review." WHO/CHS/CAH/599.3 World Health Organization Young Infant Study Group. 1999. "Clinical prediction of serious bacterial infections in young infants in developing countries." Peciatric Infectious Disease Journal 18: S23 - 31. 87 World Health Organization Collaborative Study Team. 2000. "On the role of breastfeeding on the prevention of infant mortality. How much does breastfeeding protect infant and child mortality due to infectious diseases? A pooled analysis of six studies from less developed countries." Lancet (in press). World Health Organization, 1999. "Reduction of Maternal Mortality: A Joint WHO/UNFPA/UNICEF/World Bank Statement." Geneva: WHO. World Health Organization. "Complementary feeding of young children in developing countries: a review of current scientific knowledge". WHOINUT/98. 1 World Health Organization. 1998. "The WHO position paper on Haemophilus influenza type b conjugate vaccine." Weekly Epidemiological Record 73 : 64 - 71. World Health Organization. "Evidence for the ten steps to successful breastfeeding." Geneva, WHO/CHD/98. 9. World Health Organization. 1996. "Expanded Program on Immunization. Immunization Policy." WHO/EPI/95. 03. World Health Organization. 1994. "Mother-baby package: implementing safe motherhood in countries." FHE/MSM/94. 11 88