7
Conclusions and Recommendations

Health programs in Africa encompass a wide range of medical and public health efforts. Ministries of health often oversee services, such as vaccination programs, maternity services, treatment of chronic psychiatric illnesses, and surgical wards. Given the extremely limited budgets for government expenditures and the pressures for controlling public spending, health sector planners and bilateral or multilateral funding agencies need evidence about whether the programs they support are having the desired effects and guidance on which of these programs and should be expanded.

The circumstantial evidence is clear: The coverage and range of health programs has increased substantially in Africa during the past 20 years, and infant and child mortality rates have declined. However, we found only a few studies that actually measured the effect of any component of a national health program on child mortality in Africa. The vast majority of research comes from studies of defined interventions (e.g., measles vaccination or antimalarial spraying) in small areas. Most of these studies examine the effects of a single intervention, and many are closer to carefully controlled clinical trials than they are to evaluations of large-scale programs. Therefore, we cannot make strong statements about the overall effectiveness of health programs in Africa. However, we can conclude that most national health programs include interventions that have been shown to reduce mortality in small test programs.

Our first and most important finding is that many of the central elements of most national health programs in Africa have never been evaluated



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 146
7 Conclusions and Recommendations Health programs in Africa encompass a wide range of medical and public health efforts. Ministries of health often oversee services, such as vaccination programs, maternity services, treatment of chronic psychiatric illnesses, and surgical wards. Given the extremely limited budgets for government expenditures and the pressures for controlling public spending, health sector planners and bilateral or multilateral funding agencies need evidence about whether the programs they support are having the desired effects and guidance on which of these programs and should be expanded. The circumstantial evidence is clear: The coverage and range of health programs has increased substantially in Africa during the past 20 years, and infant and child mortality rates have declined. However, we found only a few studies that actually measured the effect of any component of a national health program on child mortality in Africa. The vast majority of research comes from studies of defined interventions (e.g., measles vaccination or antimalarial spraying) in small areas. Most of these studies examine the effects of a single intervention, and many are closer to carefully controlled clinical trials than they are to evaluations of large-scale programs. Therefore, we cannot make strong statements about the overall effectiveness of health programs in Africa. However, we can conclude that most national health programs include interventions that have been shown to reduce mortality in small test programs. Our first and most important finding is that many of the central elements of most national health programs in Africa have never been evaluated

OCR for page 146
in terms of their likely impact on mortality. In some cases, the nature of the programs and of the interventions precludes direct measurement of effectiveness in reducing mortality. However, there are many types of programs that could be evaluated but have not been studied or have been studied only in small-scale trials. In particular, there are very few studies of the effects of health centers and integrated programs on mortality, and no studies that attempt to estimate the effectiveness of hospitals in reducing population levels of mortality. Given this conclusion, most of the review of the evidence—and, therefore, most of the conclusions below—concern interventions aimed at individual diseases. We will summarize here our findings about the four most common causes of death among children in Africa: measles, diarrhea, malaria, and acute respiratory infections. DISEASE- AND INTERVENTION-SPECIFIC CONCLUSIONS Measles Vaccination programs currently provide measles vaccination to about half of all children in the region. The protective effects of measles vaccination are well documented in several studies in coastal areas of West Africa and one study in Zaire. In these locales, vaccination was responsible for substantial declines in mortality. There is evidence from several other areas that vaccination has reduced the incidence of measles. However, wide variations in vaccine efficacy (often associated with cold chain failures) and substantial differences in the epidemiology of the disease (e.g., frequency of epidemics, proportion secondary cases) make it difficult to estimate the impact of measles vaccination in Africa. Vaccination is not likely to eliminate measles cases in Africa in the near future. Therefore, expanded case management should be considered as a supplemental strategy to lower the case-fatality rates in serious cases. Diarrheal Diseases Diarrheal diseases are among the leading causes of death of infants and children in sub-Saharan Africa, as they are throughout the developing world. They also reduce the health of children by imposing a high burden of morbidity and by contributing substantially to malnutrition. The primary intervention for control of diarrheal disease mortality in the past decade has been management of acute dehydrating diarrhea by using oral rehydration therapy (ORT) and continued feeding. This approach to case management has proven efficacious in clinical settings outside of Africa and is likely to prevent mortality in community-wide programs, but the effects in community-based

OCR for page 146
programs have not been well documented in Africa. To achieve the maximum benefit from this intervention in sub-Saharan Africa, it would be necessary to increase coverage beyond the currently estimated 36 percent of episodes treated, and to improve both the targeting of treatment and the quality of the treatment given. Furthermore, because acute dehydrating illnesses may be responsible for half or less of diarrhea-associated mortality, a more comprehensive approach to case management will also be needed to reduce the mortality due to dysentery and persistent diarrhea, often associated with malnutrition. Malaria Presumptive treatment of fever with chloroquine is standard practice in many parts of Africa. It is not only the practice in health centers; presumptive treatment is practiced in many homes before or instead of attending a health center. There are no studies demonstrating that presumptive treatment actually reduces childhood mortality. Although studies demonstrate the effect of treatment of malaria with chloroquine on the progression of the disease, there are no studies that test the basic assumption that reducing case severity will in the long run reduce mortality. It may not be possible to design studies to compare the effectiveness of presumptive treatment with no treatment of fevers. It would be unethical to remove this treatment from a control population that already relies on chloroquine, and it might not be possible to find endemic or holoendemic areas where presumptive treatment is not already practiced. However, it would be possible to test the effect of increasing the proportion of cases treated presumptively or the effect of improving dosages or promptness of treatment. The two studies designed to do so failed to bring about sizable changes in the frequency of treatment. The effectiveness of presumptive treatment with chloroquine is also uncertain because of increases in the prevalence of chloroquine resistance. Although the mere presence of resistant strains does not mean that chloroquine is totally useless, it does reduce the level of effectiveness. As long as chloroquine was effective, the arguments for presumptive treatment were persuasive. However, the reduced efficacy of chloroquine and the lack of equally safe, low-cost alternative drugs introduce additional uncertainty into the estimation of effectiveness. Home-based use of chloroquine and other drugs for presumptive treatment of fevers presents problems for research on the effectiveness of other strategies for reducing mortality due to malaria. Researchers cannot assume that the level of treatment or chemoprophylaxis is minimal in designated control areas or at the time of baseline surveys in intervention areas. It is necessary to document the frequency and adequacy of treatment (home based

OCR for page 146
and at clinics, presumptive and diagnosed). Without documentation of the level of treatment in control areas or baseline comparisons of program and control areas, study results cannot be extrapolated to areas in which the baseline conditions may be quite different. The evidence substantiating the use of chemoprophylaxis among pregnant women is not very strong. Although there are numerous studies demonstrating beneficial effects of prophylaxis on intermediate outcomes such as birthweight, there is little direct evidence that prophylaxis actually increases child survival. The use of insecticide-impregnated bed nets has been demonstrated to reduce mortality in one area of sub-Saharan Africa with high levels of seasonal malaria. There is a need for studies of acceptability and efficacy in populations with other patterns, such as areas of high endemicity and areas of infrequent epidemics. Acute Respiratory Infections Acute respiratory infections (ARIs), particularly pneumonia, are also a major cause of childhood mortality in sub-Saharan Africa. Although their importance has been recognized for some time, it is only recently that a control program strategy has been shown to be effective. This strategy uses a presumptive diagnostic algorithm of pneumonia based principally on respiratory rate and recognition of chest indrawing; treatment is given with antibiotics. In community-based intervention trials in a variety of settings, this strategy has been shown to reduce under-5 mortality by approximately 25 percent. In the only trial in sub-Saharan Africa, the probability of dying by age 5 was reduced by 39 per 1,000. Clearly, some antibiotic treatment of suspected pneumonia is currently occurring in Africa, but systematic programs to improve diagnosis and provide correct therapy are just beginning. Although it is too soon to determine an effect on child mortality of ARI treatment programs, there would seem to be the potential for substantial mortality reduction if the case-management strategy can be fully implemented. GENERAL OBSERVATIONS ABOUT THE EVALUATION OF HEALTH PROGRAMS IN AFRICA Given the wide range of disease environments in Africa, a worrying number of the best studies of the efficacy of basic interventions have been carried out in a very narrow range of ecological and cultural settings. In particular, a substantial proportion of the studies have been carried out in small areas of Senegal and The Gambia. The area of Senegal that has been studied by researchers from the Institut Français de Recherche Scientifique

OCR for page 146
pour le Développement en Coopération (ORSTOM), and the area of The Gambia that has been studied by researchers from the Medical Research Council of the United Kingdom, are within approximately 60 miles of each other in very similar ecological and cultural environments. Good research is being done in other areas of Africa. However, there is a desperate need for more research on the effectiveness of integrated programs and individual interventions in a wider range of environments. The high volume of excellent research from Senegal and The Gambia is evidence that long-term studies in defined populations are an effective means of studying health interventions. In addition to disease- and intervention-specific conclusions and recommendations, there are a number of general observations related to the state of health programs and research in sub-Saharan Africa. Declines in infant and child mortality rates should remain the primary indicator of the effectiveness of child health interventions in Africa. There are other useful indicators of the success of some specific health interventions, such as improved nutritional status and lower prevalence of chronic morbidity. However, the primary goal of most health programs in Africa must be the reduction of mortality. In addition, many of the interventions aimed at reducing mortality are also associated with reductions in morbidity. That having been said, it is not necessary to measure the change in mortality associated with every program. Once we have shown that an intervention reduces mortality when implemented properly, we can evaluate programs using coverage and promptness of services and compliance with program protocols. The goals stated for many programs suggest that program planners often have unrealistic expectations about the feasibility of measuring mortality changes associated with some kinds of interventions. There is large variation in the rapidity with which interventions can reduce mortality. It is feasible to measure the effects only of those interventions that can achieve large reductions in mortality quickly. Evaluation of interventions that reduce mortality by only a modest amount (e.g., less than 10 percent) requires very large sample sizes to achieve precise measurements of mortality. However, the larger the sample size, the more difficult it is to ensure precise measurements. For national programs, it may not be possible to measure the effect of programs that reduce mortality in an age group by less than 20 per 1,000. Similarly, it may not be possible to demonstrate that a program reduced mortality if the mortality decline occurred over several years. The slower the decline, the longer must data collection continue and the greater is the expense. A long, slow decline also makes it difficult to demonstrate that  

OCR for page 146
  the change was caused by the program rather than by other changes in the population. It is useful to consider three types of programs: Programs that are relatively easy to study because they have a dramatic effect on mortality: Included here are programs that can cause a quick, dramatic reduction in mortality, perhaps limited to one age group. In some cases, it is possible to document the effectiveness of these programs in large populations. Examples are programs that include measles vaccination and mass immunization of women of childbearing age with tetanus toxoid. Regular distribution of vitamin A may also belong in this group. Although evaluations of these programs can demonstrate short-term effects on mortality, studies of long-term, sustained effectiveness generally will have to rely on intermediate measures such as vaccine efficacy and surveillance studies of incidence. Programs whose effectiveness in reducing mortality can be measured only in small-scale studies: This group includes programs that have a more modest potential for reducing mortality or that reduce mortality at a slower rate. We can strengthen these studies by comparing survival rates for those who received the intervention and those who did not. Examples include programs for treatment of ARI and immunization of pregnant women with tetanus toxoid. Once the effectiveness of these programs has been demonstrated in small-scale studies, larger programs can be evaluated by using intermediate measures such as coverage rates, efficacy, and incidence rates. Programs for which direct measurement of effect on mortality is not feasible: This group includes programs that require an expansion of infrastructure, establishing a referral structure, a large change in staff duties, or changes in health-seeking behavior or treatment of disease by the population. These programs can take so long to be implemented fully that it is very difficult, if not impossible, to document their effects in populations. In addition, it is often difficult to document a single starting point for these programs or to measure changes in the level of program activity. This category includes programs to reduce the incidence of low birthweight and prematurity. In general, we can only estimate the effectiveness of these programs by using studies of intermediate output measures (e.g., coverage rates and changes in the incidence of low birthweight) combined with studies that demonstrate the association of these intermediate variables with excess mortality. Misunderstandings about the nature of program effect can lead to unrealistic expectations for measurable impact. For example, many discussions of the potential of home-based ORT programs imply that they can achieve quick, dramatic reductions in mortality. In fact, such programs belong in the second or even third group because of the time required to implement  

OCR for page 146
  them and to change behaviors. Similarly, programs that provide tetanus toxoid can achieve large reductions in some populations if coverage increases rapidly. However, programs that provide injections through antenatal car may not achieve a quick increase in coverage in populations where prenatal care is low. The foregoing does not mean that we should invest only in programs that have quick effects. However, we should ensure that the measures used to evaluate these programs are firmly linked to real effects. The trend toward stating program goals in terms of reduction in cause-specific mortality may be setting unrealistic expectations for evaluation. Measuring cause-specific mortality rates is very difficult in Africa. Even carefully performed verbal autopsies using locally validated methods can only provide estimates of the distribution of deaths by the most common immediate causes. Assigning associated or underlying causes exceeds the capabilities of large studies and national systems. In general, it is not possible to measure cause of death precisely enough for the evaluation of large-scale programs. However, it is important to produce better national and subnational estimates of the relative importance of major causes of death and to develop systems for monitoring long-term trends in the cause-of-death structure. The verbal autopsy components of the Demographic and Health Surveys and other surveys should begin to give some ideas of the mortality profiles of infants and children in the general population. More emphasis should be given to age-specific mortality rates in stating program goals. It is far easier to measure achievement of goals for reduction of neonatal, postneonatal, or second-year mortality, for example, than goals for reducing disease-specific mortality. It is often possible to identify age (or age-sex) groups that should benefit most from a program. There is a need for more evaluations of various packages of interventions. For example, there may soon be a need to evaluate the combination of vaccination programs and vitamin A distribution. The purpose of these evaluations should be to measure the total impact of the package. It is rarely possible to determine which elements of a package are responsible for the largest share of the effect. The synergies between diseases make it as difficult to estimate the joint effect of two interventions from separate evaluations as it is to disentangle their separate effects when they are studied in combination. Because the impact of various packages can differ greatly across disease environments, we should evaluate these packages in several types of populations. More empirical evaluation of program effects are needed in order to test predictions from models. Although we have data on many components of these models, there are no direct tests of the effects of the recommended practices. Examples include the training of traditional birth attendants and several of the common recommendations on malaria. We are  

OCR for page 146
  not suggesting that programs stop training traditional birth attendants or change their recommendations on malaria. Whenever possible, the conclusions from simplified biological models should be tested in studies that actually measure the effectiveness of the proposed intervention in reducing mortality. There is a need for more long-term studies that include regular collection of vital statistics and routine surveys of service utilization and quality of care. These studies require long-term commitments of funds and personnel. However, their potential is apparent from the number of important studies carried out in Senegal and The Gambia. Without these study areas, we would know much less about such diseases as measles and malaria, and the potential effect of interventions to combat them. We need similar types of study areas in other parts of Africa. Similarly, on-going data collection programs, such as the Demographic and Health Survey program, should continue gathering information on health services and conditions. These surveys can contribute to the knowledge base of long-term health and population changes in sub-Saharan Africa. Evaluation studies should include detailed measurement of both the coverage and the promptness of services, as well as compliance with program protocols. We cannot adequately evaluate measurements of mortality change associated with programs unless we also have data on the program activities. During the last 10 years there has been a great deal of attention paid to the health needs of Africa. Many programs have been started or expanded, and many new approaches have been tested in clinical trials and small study areas. Despite this increased effort, we know very little about the effects of single interventions in large-scale national programs or the overall effectiveness of integrated health programs. There are also many parts of the continent where we know very little about the effect of programs or even the efficacy of treatments. The need for program evaluations will increase as the recommended combinations of interventions become more elaborate. In particular, if we begin to consider adding large-scale intervention against acute respiratory infections or vitamin A supplementation to the existing vaccination and diarrhea control programs, it will be increasingly difficult to evaluate programs without direct measurements of their effect. As the size and the complexity of programs increase, so will the need for more elaborate systems to evaluate and monitor their effectiveness, and to set new priorities.

OCR for page 146
This page in the original is blank.