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 108
5 Nutrition and Nutrition Programs INTRODUCTION The synergistic interactions between nutritional status and infectious diseases are an important part of the ecology of disease in Africa. These interactions complicate the evaluation of the effects of health programs because individual disease episodes cannot be treated as independent events. Preventing or treating a disease episode might reduce its nutritional effect, which in turn could reduce the severity of the next disease episode. Because the later episode might involve a different disease than the earlier one, a program to prevent or treat one disease could have implications for mortality due to other diseases. Therefore, health programs might be able to reduce mortality by employing two complementary approaches: preventing or treating infections, and preventing or treating malnutrition. The term malnutrition often implies one particular nutritional deficiency: protein-energy malnutrition, PEM, which results from inadequate consumption of calories or protein. However, deficiencies of numerous vitamins and other nutrients can be equally serious. It is often difficult to determine which is the major cause of malnutrition: inadequate consumption of specific nutrients (e.g., protein, vitamin A, or iron) or consumption of inadequate quantities of food (usually measured in calories). This section examines programs designed to reduce the prevalence of protein-energy malnutrition, low birthweight, and vitamin A deficiency. Although other nutritional deficiencies might be equally important, these are the problems
OCR for page 109
most frequently addressed by current program options. We have limited this review to studies that examine whether these interventions reduce mortality in children. PROTEIN-ENERGY MALNUTRITION (PEM) PEM can result from inadequate or inappropriate intake of energy, protein, or one or more essential amino acids. It also may be due to temporarily decreased dietary intake resulting from anorexia or malabsorption of nutrients. Many infectious diseases, especially diarrhea, threaten a child's nutritional status by decreasing appetite or reducing the ability to absorb nutrients. In addition, the immunological responses to infection increase nutritional requirements. Severe PEM can impair a child's response to infectious assaults. For an infection to occur, a pathogen has to overcome the host's immune system. Malnutrition can weaken these defense mechanisms. For example, children with severe clinical malnutrition (kwashiorkor, characterized by lethargy, edema, and dermatitis; and marasmus, characterized by severe wasting associated with depletion of fat and muscle reserves) are more susceptible to gastrointestinal infections because of a reduction in gastric acid (Gracey et al., 1977). PEM also can slow the speed with which an immunological response occurs and can reduce the rate of epithelial replication and tissue repair. Therefore, malnourished children may have longer, more severe cases of what might otherwise be simple childhood infections. In addition, deficiencies in tissue repair may be especially problematic with mucosal surfaces, as in the nasal tract and the intestine. This phenomenon can increase the child's risk of new infections. Mild to moderate. PEM is marked by growth retardation and reduced motor activity. However, most research on PEM now relies on standardized ratios among weight, height, and age. The most commonly used anthropometric indices are height-for-age (Ht/Age), weight-for-height (Wt/Ht), and weight-for-age (Wt/Age). In addition, indices such as upper arm circumference-for-age and -for-height are used to assess nutritional status. There are several ways of comparing these ratios to international standards. The Gomez classification (Gomez et al., 1956) compares a child's weight to a standard schedule of expected weight for a given age. The Waterlow (1972) classification differentiates between stunting (low Ht/Age) and wasting (low Wt/Ht). Stunting results from long-term nutritional insult. Wasting, on the other hand, is a measure of acute undernutrition. Low Wt/Age can result from either stunting or wasting. A child who is two standard deviations (s.d.) below the reference population median for any of the three indexes is classified as undernourished (Hamill et al., 1977; Administrative Committee on Coordination—Subcommittee on Nutrition, United Nations, 1987).
OCR for page 110
Most early anthropometric studies used the Harvard reference population for comparison (as given in Jelliffe, 1966). More recent studies use the National Center for Health Statistics (NCHS) reference population for comparison (Hamill et al., 1977). National surveys of nutritional status in sub-Saharan Africa show wide variation in the prevalence of malnutrition, based on the three commonly used indices. Figure 5-1 presents estimates of the proportions with low Wt/Age from the Demographic and Health Surveys (DHS) in eight African countries. Malnutrition tends to be lower during the first year of life, when breastfeeding is more common, than in later years. In fact, in most studies in Africa, children less than 3 months of age are on average heavier than the standard. This finding probably results from excessive mortality rates among low-birthweight infants. From 6 to 23 months of age, malnutrition becomes more common as children experience repeated bouts of diarrhea and other childhood diseases. Figures 5-2 and 5-3 present age patterns of stunting and wasting. The prevalence of wasting (acute malnutrition) peaks during the second year of FIGURE 5-1 Prevalence of children malnourished (low weight-for-age) by age of child, sub-Saharan African countries. NOTE: Intervals during first year of life vary by country: Burundi, Ghana, Mali, and Zimbabwe: 3-11 months; Togo and Uganda: 0-11 months; Senegal: 6-11 months; Nigeria: 0-5 and 6-11 months. Intervals after first year of life are: 12-23 months (shown as 18 months), 24-35 months (shown as 30 months), 36-47 months (shown as 42 months), and 48-60 months (shown as 54 months). SOURCE: Demographic and Health Survey reports (see Appendix B).
OCR for page 111
FIGURE 5-2 Prevalence of stunting (low height-for-age) by age of child, sub-Saharan African countries. NOTE: Intervals during first year of life vary by country: Burundi, Ghana, Mali, and Zimbabwe: 3-11 months; Togo and Uganda: 0-11 months; Senegal: 6-11 months; Nigeria: 0-5 and 6-11 months. Intervals after first year of life are: 12-23 months (shown as 18 months), 24-35 months (shown as 30 months), 36-47 months (shown as 42 months), and 48-60 months (shown as 54 months). SOURCE: Demographic and Health Survey reports (see Appendix B). life, with national levels for this age group ranging from about 2 to 16 percent. The accumulation of these deficits in growth leads to high levels of stunting among children over age 18 months. The proportions stunted at ages 24-35 months (shown as 30 months on the figure) varies from 28 percent in Senegal to 60 percent in Burundi. National estimates of the prevalence of malnutrition hide large variations in the prevalence among regions or among social, cultural, and economic groups. Figure 5-4 shows regional estimates of the prevalence of stunting at ages 6-35 months from the DHS in West Africa. Estimates for Togo range from 27 percent stunted in the southern Maritime Region to 48 percent in the northern Savanes Region. Ghana and Nigeria show a similar pattern of higher levels of stunting in the north. In western Mali, the proportion stunted is very similar to neighboring Senegal. The eastern and northern parts of Mali have higher levels of stunting. These regional patterns reflect ecological zones that transcend national boundaries. The prevalence of malnutrition can vary substantially over quite short distances and among social and ethnic groups. For example, a survey in
OCR for page 112
FIGURE 5-3 Prevalence of wasting (low weight-for-height) by age of child, sub-Saharan African countries. NOTE: Intervals during first year of life vary by country: Burundi, Ghana, Mali, and Zimbabwe: 3-11 months; Togo and Uganda: 0-11 months; Senegal: 6-11 months; Nigeria: 0-5 and 6-11 months. Intervals after first year of life are: 12-23 months (shown as 18 months), 24-35 months (shown as 30 months), 36-47 months (shown as 42 months), and 48-60 months (shown as 54 months). SOURCE: Demographic and Health Survey reports (see Appendix B). northeastern Timbuktu Region of Mali in 1985 after two consecutive crop failures found that 43 percent of nomadic children under age 5 were stunted, compared to only 20 percent of children in sedentary families (Carnell and Guyon, 1990). A study by Mbithi and Wisner (quoted in Kenya Bureau of Statistics, 1979) examined variations in the prevalence of malnutrition among children living on the eastern side of Mt. Kenya. In areas with the best agricultural land at higher altitudes, only 10 percent of children under age 3 years were less than 70 percent of the standard Wt/Age. At lower altitudes, 38 percent of children living in an area with poorer quality agricultural land were below 70 percent of the standard. Causes of Malnutrition Malnutrition may arise from a number of conditions, such as lack of food, cessation of breastfeeding, and infection. These factors are often interrelated, making it difficult to determine exactly how a child becomes malnourished. Food production has not kept pace with population growth in sub-Saharan
OCR for page 113
FIGURE 5-4 Proportions of children stunted in regions of Senegal, Mali, Ghana, Togo, and Nigeria. NOTE: Data for Senegal, Mali (Bamako shown in parentheses), Ghana (Accra shown in parentheses), Togo, and Ondo State, Nigeria (in parentheses) are age standardized, with 42 percent aged 6-17 months and 58 percent aged 18-36 months. The other data for Nigeria are ages 1-59 months. SOURCE: Demographic and Health Survey reports (see Appendix B). Africa. To achieve the needed level of food security, food production in sub-Saharan Africa must grow at about 4 percent per year, but has grown only 2 percent per year since the 1950s (World Bank, 1989). Drought conditions, lack of infrastructure, and political instability further exacerbate the situation. It is estimated that about 25 percent of the population of sub-Saharan Africa consumes less than 80 percent of the caloric intake recommended by the Food and Agricultural Organization and the World Health Organization (WHO) (World Bank, 1989). Because studies of caloric intake tend to be of limited value due to the difficulty in determining the quantity and nutritional content of foods that are eaten, they are not reviewed here. Among infants, poor breastfeeding patterns may lead to malnutrition. Breastfeeding is very prevalent for long periods in sub-Saharan Africa. Table 5-1 presents data from the DHS in sub-Saharan Africa on mean duration of breastfeeding and the proportion still breastfeeding during the first two
OCR for page 114
TABLE 5-1 Breastfeeding Patterns in Sub-Saharan Africa Country Mean Duration (months) Breastfed Child During First Two Months of Life (%) Breastfed Child Beyond First Year (%) Botswana 18.8 90.8 73.0 Burundi 23.8 97.0 91.3 Ghana 20.4 92.7 87.2 Kenya 19.4 96.0 81.7 Liberia 17.0 93.4 60.9 Malawi 21.6 90.4 81.5 Nigeria 20.1 96.9 88.4 Senegal 18.8 89.0 86.0 Togo 22.6 95.4 84.2 Uganda 18.6 90.9 84.7 Zimbabwe 19.3 95.0 87.6 SOURCE: Demographic and Health Survey reports (see Appendix B). months of life and beyond the first year of life. Mean duration is among the longest in the world, with a range of 17.0 to 23.8 months. Although duration is long, it does not mean that all children are breastfed. The table shows that in all the countries of sub-Saharan Africa included in the DHS (with the exception of Senegal), more than 90 percent of the children are breastfed during the first two months of life. After the first birthday, the proportion still breastfeeding ranged from 61 to 91 percent. Despite the common practice of breastfeeding, it is not uncommon for mothers to stop when their children become ill. Thus, the cessation of breastfeeding may enhance the possibility that an already-sick child will become malnourished, thereby increasing the probability of death. Infection is one of the common underlying causes of malnutrition. Measles, for example, puts many children below the local standard for Wt/Age and Wt/Ht in the Kasongo Project (1986). Rowland et al. (1977) found that gastroenteritis contributed significantly to growth faltering among Gambian children ages 6 to 35 months, reducing height gain at a rate of 4.2 millimeters per month and weight gain by 746 grams per month, compared to the growth that occurred among unaffected children. Malaria, although much less prevalent than gastroenteritis in the study, reduced weight gain by 1,072 grams per month. Another study conducted in The Gambia by Rowland et al. (1988) examined Wt/Age among children to assess the relationship between growth and morbidity during the first two years of life. During their first six months, the children exceeded the NCHS reference (see Hamill et al., 1977).
OCR for page 115
However, during the second six months of life, they had an average deficit of 1.2 kilograms. Rowland et al. attributed the weight faltering principally to diarrhea and acute lower respiratory infections (ALRIs). Diarrhea was responsible for one-half of the deficit (14.4 grams per day among weaning infants) and ALRI for one-fourth (14.7 grams per day of infection). In a study of growth faltering among Sudanese children, diarrhea was an initiating factor in about 50 percent of cases of faltering (Zumrawi et al., 1987). Among children ages 3 to 6 months, one day of diarrhea reduced the average weight gain of 18 grams to an average loss of 13 grams, a net loss of 31 grams compared to the normal weight change. Colds and cough were associated with a loss of 16 grams per day compared to the normal weight change. These studies suggest that many common childhood infectious diseases can contribute to growth faltering and malnutrition. Studies of Malnutrition and Mortality in Other Regions Several studies have demonstrated that malnourished children are at increased risk of death. For example, Schroeder and Brown (personal communication, 1992) reviewed anthropometric studies in India, Bangladesh, and Papua New Guinea. They examined the relative risk of mortality of children identified as malnourished during the 6- to 24-month period after diagnosis. They concluded that midly or moderately malnourished children aged 6 to 60 months had a risk of death 2.1 times that of well-nourished children. Severely malnourished children had a risk of dying 6.5 times that of well-nourished children. The amount of excess risk associated with a given level of malnutrition varies across ecological and social environments. After reviewing a number of studies, Pelletier (1991) concluded that the response of mortality to malnutrition is a function of the baseline level of mortality, with malnutrition having a exacerbating effect on child mortality for any level and type of morbidity that exists in a population. Studies of the Relationship Among PEM, Morbidity, and Mortality in Africa The relationship between nutritional status and mortality might be different in Africa than in Asia and Latin America. First, the attributable risk associated with malnutrition (i.e., the extent to which mortality would decline if all children had the mortality rates of well-nourished children) depends on both the percentage malnourished and the degree of malnutrition. If the distribution of children by nutritional status is different in Africa than elsewhere, the attributable risk associated with malnutrition could be different
OCR for page 116
as well. This variation could be true even if the risks associated with various levels of malnutrition are the same in Africa as elsewhere. Second, in some surveys, nutritional status may serve as a marker for social class. If studies do not control adequately for other risk factors associated with social class, such as education or residence, the estimates of the importance of malnutrition might be exaggerated. It is possible that the link between nutritional status and social class is less important in Africa than in Asia (Bairagi et al., 1985). Third, malaria is a more significant factor in morbidity and mortality in Africa than in most other regions. The interactions between malarial infection and malnutrition are complex and have not been investigated adequately. PEM and Morbidity A few studies have examined the link between malnutrition and morbidity in sub-Saharan Africa. A study of children in The Gambia (Tomkins et al., 1989) who were between 6 and 35 months of age at the baseline survey found that short and underweight children experienced an excess risk of illness from diarrhea or fever. The differences persisted after controlling for social, economic, and environmental conditions that might confound the association between the anthropometric index and the excess risk of morbidity. In the Malumfashi study in Nigeria, however, Tomkins and colleagues (Tomkins, 1981; Tomkins et al., 1991) found that malnutrition had relatively little impact on the incidence of diarrhea, but did increase its prevalence and presumably the average duration of illness. In the Sudan, El Samani et al. (1988) studied the association between malnutrition and diarrheal disease by weighing and measuring a group of children under age 5 every two months. They reported that children who had experienced an episode of diarrhea in the preceding two months and who were less than 90 percent of the Wt/Age based on the NCHS standard were more likely to have a subsequent diarrheal attack. Among those children who had not had an attack of diarrhea in the preceding two months, the incidence of diarrhea in the subsequent interval was higher among children with Ht/Age less than 95 percent of the standard. After controlling for a number of potentially confounding factors, El Samani et al. found that children with Wt/Age less than 75 percent of the standard were twice as likely to have diarrhea in the subsequent interval, regardless of whether they had had diarrhea in the preceding interval. Biritwum et al. (1986) found that children in Ghana with Wt/Age less than 80 percent of the WHO standard had a mean of 2.6 episodes of diarrhea per year compared to only 1.7 for other children. This difference was significant at the 95 percent level.
OCR for page 117
Lang et al. (1986) examined the association between ALRI and nutritional status in Burkina Faso. They found that children with a small arm circumference had a higher incidence and a longer duration of ALRI. The relationship between nutritional status and morbidity is complex and not well documented in Africa. However, it does appear that malnutrition probably increases the proportion of time a child suffers from diarrhea and may also complicate acute respiratory infections (ARIs). PEM and Mortality Lindskog et al. (1988) examined survival rates during the year following an anthropometric survey of children under age 5 in a rural area of Malawi. After adjusting for age, there was a consistent, significant relationship between mortality and nutritional status as measured by height-for-age, weight-for-height, or weight-for-age. For example, as shown in Table 5-2, children who were between 1 and 2 standard deviations below the standard Ht/Age had a relative risk of death 1.46 times that of children with a higher Ht/Age. Those who were more than four standard deviations below the standard had a relative risk of 3.3. The Kasongo Project (1983) in Zaire observed that the risk of mortality was 1.8 times greater among children with Wt/Age indices less than 80 percent of the Harvard standard median (see Jelliffe, 1966) than among other children. Those less than 60 percent of the median were 3.3 times more likely to die than better-nourished children. (They did not present confidence intervals or significance tests for these differences.) These risk ratios were lower than the values the authors reported based on the data for India provided by Kielmann and McCord (1978). TABLE 5-2 Effect on Relative Risk Estimates of Controlling for Age and Period on the Relationship Between Height-for-Age and Child Mortality, Malawi, 1983-1985 Ht/Age Score (standard deviations from median) Raw Relative Risks Adjusted Relative Risks Greater than -1 s.d. 1.00 1.00 -2 to -1 s.d. 0.88 1.46 -3 to -2 s.d. 0.76 1.71 -4 to -3 s.d. 1.15 2.79 Less than -4 s.d. 1.07 3.30 NOTE: Adjustments are made by introducing age (0-5, 6-11, 12-17, 18-23, 24-35, 36-39 months) and period into the log-linear regression analysis. SOURCE: Lindskog et al. (1988).
OCR for page 118
A study by Smedman et al. (1987) of children aged 6 to 59 months in Guinea-Bissau examined survival during the 8 to 12 months following weighing and measuring. They reported that Ht/Age was correlated with child survival after controlling for the age of the child. However, Wt/Ht was not significantly related to survival. They also noted that their findings showed less effect of nutritional status than studies in Asia. Briend et al. (1989) examined survival rates for children in rural Senegal during the six months following semiannual weighings. They concluded that survival is related to nutritional status, and the risk is most closely related to muscle mass rather than to the standard nutritional indices. These studies show that poor nutritional status is associated with higher mortality, although the relationship appears to be weaker than that found in similar studies in Asia. These studies do not prove that the relationship between nutritional status and mortality is causal. Although poor nutritional status may compromise the immune system, it is also possible that poor nutritional status and elevated risk of death are jointly affected by some other unmeasured characteristic of children such as child care practices, access to health care, or differences in socioeconomic class or housing, or by some other aspect of nutritional status such as vitamin A deficiency. PEM and Measles Mortality There is solid evidence that children with measles often develop malnutrition (Kasongo Project Team, 1986; Reddy et al., 1986). However, Aaby et al. (1984a-c, 1986) questioned whether malnutrition is associated with higher case-fatality rates for measles. Many studies (such as Kimati and Lyaruu, 1976) have examined the relationship between nutritional status and case fatality due to measles in hospitals. Generally, however, hospital-based studies cannot provide evidence of the temporal relationship among malnutrition, the onset of measles, and subsequent mortality. For example, many of the severe cases in hospitals are among children who were sick for several days before coming to the hospital. Among these cases, there might be an increased prevalence of low Wt/Age as a result of several days of illness. There have also been several studies based on long-term monitoring of anthropometry, measles cases, and mortality. For example, a study in Bangladesh (Koster et al., 1981) weighed and measured each child every two months. Children who died of measles had weights and heights comparable to those of controls matched for age, sex, and neighborhood. There was no evidence in the study that preexisting malnutrition increased the risk of death from measles. Aaby (1992) presented data from a measles epidemic in Bandim, Guinea-Bissau,
OCR for page 121
The mortality data from the Iringa Nutrition Programme include the number and proportion of deaths due to specific causes. Over the course of the study, deaths from respiratory infections and diarrhea were reported to have decreased, measles deaths were fairly constant, and deaths from fever (presumably malaria) increased. However, it is not possible to estimate death rates based on the data provided, because the authors did not supply adequate information on exposure to death. Were it possible to estimate age-specific rates from the data, the study would be more helpful in determining the effect of nutritional status on mortality. The team that evaluated the Iringa Nutrition Programme concluded that these data provide a strong indication of the program's effect. However, the lack of a control area made it difficult to reach such a conclusion. Moreover, the evaluation did not classify children as stunted or wasted, nor did it provide data by age. Perhaps hundreds of small- and large-scale nutrition programs have been undertaken in different parts of Africa. In a qualitative review of the factors contributing to successful nutrition programs in the region, Kennedy (1991) discussed seven key elements: community participation, program flexibility, institutional structure, recovery of recurrent costs, multifaceted program activities, well-trained and qualified staff, and the presence of infrastructure. However, because most nutrition programs do not gather mortality data, it is not possible to conclude much about the effect of these programs on survival. Supplementary Feeding Programs Beaton and Ghassemi (1982) reviewed the effect on supplementary feeding programs on nutritional status and reported that such programs should improve nutritional status of children, but often do not because of low coverage, low levels of supplementation, food sharing, and food substitution. Few programs collect data on the morbidity (other than relief of malnutrition itself) and mortality consequences of food supplementation programs. Studies in Guatemala, India, and Peru found a significant reduction in infant and/or child mortality due to supplementary feeding alone (Ascoli et al., 1967; Scrimshaw et al., 1968; Baertl et al., 1970; Kielmann and McCord, 1978). However, no evidence of such effects of programs implemented in Africa was uncovered. Growth Monitoring Monitoring the growth of infants and children may increase the effect of programs that provide nutrition rehabilitation or other services that can be targeted to malnourished children. In Jamaica, a reduction in mortality
OCR for page 122
was observed following increased access to primary health care in association with growth monitoring and targeted provision of food (Alderman et al., 1978). However, this program was tested in an area with mortality rates comparable to the lowest rates in Africa—mortality at ages 1-48 months of 14.5 deaths per 1,000 before the start of the program. In Malawi (Cole-King, 1975), there was a decline in the proportion malnourished following the expansion of the national system of under-5 clinics. The percentage of undernourished children (based on the Wt/Age index) dropped from 37 in the first year to 29 percent in the second and third years. However, the clinics offered a wide range of services in addition to growth monitoring, so it is not possible to determine what part of the change in nutritional status might have been caused by nutrition activities. The potential effect of growth monitoring programs is limited by the sensitivity and specificity of nutritional status as a screening tool for identifying those with the highest risk of death. For example, the Kasongo Project Team (1983) emphasized that the sensitivities of the various anthropometric measures are very low. Their data suggest that if a program in Kasongo targeted children below the tenth percentile in Wt/Age using a local standard and reduced their mortality rate to the average for other children, mortality would drop by only about 10 percent. These calculations suggest that even if a program succeeded in eliminating all the excess mortality associated with malnutrition, the effect on mortality would be so small that it would be hard to measure. Although some other areas of Africa have a higher prevalence of low Wt/Age than Kasongo, it would still be difficult to measure the effects of the most successful programs. Studies that have examined the effect of growth monitoring have found little benefit. At a recent UNICEF meeting, it was suggested that growth monitoring not be adopted as a global strategy, but that growth promotion should be. Weighing or measuring all children is a difficult undertaking and the information is often used inappropriately by individuals, households, or communities. Thus, UNICEF has developed a three-step program for growth promotion strategies, which includes activities such as nutrition education, surveillance, and paying special attention to children identified as high risk in community-based nutrition surveillance (United Nations Children's Fund, 1992). Programs Designed to Change Breastfeeding and Weaning Weaning education can modify behavior, but it must address the cultural norms and social and economic conditions of the groups to which it is directed. Several strategies have been shown to result in improved nutritional status in sub-Saharan Africa. These include rehabilitation centers for
OCR for page 123
malnourished children in Zaire (Brown and Brown, 1979); home-based training of mothers in various aspects of food production and preparation, food hygiene, and basic health care in Uganda Hoorveg and McDowell, 1979); and community-based activities such as demonstrations and group lessons on low-cost, homemade, weaning foods carried out by village-based volunteer monitors in Burkina Faso (Zeitlin, 1981). These studies showed improvements in nutritional status of children, but did not assess the effects on mortality. Weaning education may also contribute to a reduction in mortality from other causes in which malnutrition is a complicating factor. The promotion of breastfeeding is another type of nutrition intervention that is related to improved probability of child survival. Breastfeeding ensures that a child receives adequate nutrition in early infancy. Moreover, it protects the child from diarrhea, ARI, and other diseases (Feachem and Koblinsky, 1984; Huffman et al., 1991). Strategies for increasing the initiation and duration of exclusive breastfeeding include training and education of health professionals (Potter et al., 1987), changes in hospital practices that facilitate immediate initiation (Klaus and Kennel, 1976), keeping mother and child in the same room (Mata, 1983), and restriction of infant formula samples (Bergevin et al., 1983). Although a number of studies have examined the effect of breastfeeding on nutritional status and disease prevention, they generally do not measure its effect on mortality. LOW BIRTHWEIGHT Low birthweight (LBW), typically defined as weighing less than 2,500 grams at birth, has been reported to be the strongest predictor of infant mortality, especially in the neonatal period (Susser et al., 1972). LBW appears to affect mortality through direct and indirect mechanisms. Children with the condition are more likely to have impaired cellular immunity, which may increase their risk of early cases of diarrhea, respiratory infection, and other infections. Studies of the Relationship Between Low Birthweight and Mortality The increased risk of mortality among LBW infants has been demonstrated in a number of studies in other regions (Shapiro, 1968; Puffer and Serrano, 1973; De Vaquera et al., 1983; Victora et al., 1988). There are few similar studies in Africa. Mbacké and van de Walle (1992) examined the role of birthweight on survival using a cohort study of all births in maternity hospitals in the town of Bobo-Dioulasso, Burkina Faso, between April 1981 and March 1984. They reported neonatal, postneonatal, and second-year death rates by birthweight
OCR for page 124
for 6,091 births who either survived to age 2 years or died before that age. More than 13 percent of these children had birthweights less than 2,500 grams. The infant mortality rate for these low-birthweight infants was 3.88 times (95 percent confidence interval (C.I.) 3.32-4.54) the rate among births with weights greater than 2,500 grams (250 and 64.4 per 1,000 live births).1 Extending the analysis through the second year of life reduces the relative risk somewhat. The risk of dying before age 2 years was 2.84 times (95 percent C.I. 2.48-3.24) as high for low-birthweight children as for other children. (The probability of dying by age 2, 2q0, were 285 and 101 per 1,000 live births.) Mbacké and van de Walle also tested whether the difference in postneonatal and second-year mortality rates remained after controlling for socioeconomic factors (father's income, mother's education, type of home, etc.) and other risk factors (sex, twins, month of birth, birth order, housing density, number and timing of prenatal visits, and use of measures against malaria). They found that those children with weights greater than 3,000 grams still had a postneonatal mortality rate significantly lower than low-birthweight children (odds ratio of 0.53; probability less that or equal to .001), and those weighing 2,500-2,599 grams still had a lower postneonatal mortality rate than low-birthweight children (odds ratio of 0.84; not significant). After controlling for other factors, there was no significant difference by birthweight in the second-year mortality rate. Low birthweight is a relatively common condition. It is estimated that approximately 17 percent of all births in developing countries are LBW (World Health Organization, 1980). The condition varies a lot by region, with higher prevalence observed in Asia and lower prevalence in Latin American and Africa (World Health Organization, 1980a). The underlying cause of LBW is also different when less and more developed countries are compared. Intrauterine growth retardation is estimated to account for more than half of LBW infants in developing countries, whereas preterm delivery is the major cause of LBW in developed countries (Kramer, 1987). The prevalence of LBW in many sub-Saharan African countries is unknown because of unattended births and poor registration systems. In the Machakos study in Kenya, 7.0 percent of the births were LBW (Muller and van Ginneken, 1991). In a study of the effect of malaria prophylaxis on birthweight in Burkina Faso, the incidence of LBW was 16.4 percent for both the test and the control groups (Cot et al., 1992). There are several studies of birthweight among births in African hospitals. However, hospital births may not provide an unbiased sample of all births. Although the 1 These estimates are based on the data presented by Mbacké and van de Walle (1992:132).
OCR for page 125
prevalence of LBW in sub-Saharan African exceeds that of developed countries, it may be lower than that observed in Asia (Kramer, 1987). Risk Factors for Low Birthweight A number of factors, such as inadequate prenatal care, inadequate maternal weight gain, physically demanding work, short birth intervals, and tobacco or alcohol consumption have been associate with LBW. These conditions can be modified through interventions prior to or during pregnancy. These types of factors tend to be correlated with LBW, but may not cause it. LBW and factors such as inadequate prenatal care and short birth intervals may be jointly influenced by unobserved characteristics such as access to health care and socioeconomic status. 2 Other factors, such as multiple births, maternal height, and birth order also affect birthweight, but are not as easily addressed through interventions. Physically Demanding Work Two studies in Africa suggest that a seasonal increase in the energy expenditure of pregnant women may affect birthweight more than a seasonal decrease in caloric intake. The Keneba study in The Gambia noted a decrease in birthweight preceding the decrease in seasonal intake and paralleling the increase in physical work (Roberts et al., 1982). Similarly, a study in Tanzania did not see any seasonal decrease in birthweight when rains delayed the beginning of field work (Bantje, 1983). Short Birth Intervals One of the motives for family planning programs has been the promotion of longer birth intervals to reduce the prevalence of LBW. The Demographic and Health Surveys conducted in sub-Saharan Africa reinforce the strength of the association between short birth intervals and infant mortality. Figure 5-5 illustrates that as birth intervals become longer, perhaps through the use of contraceptives, the infant mortality rate decreases, although the analysis does not control for any potential confounding factors. This negative association between short birth intervals and survival has been widely studied (e.g., Hobcraft et al., 1985), but the biological mechanism is not well understood. The relationship between socioeconomic status and birth intervals might explain part of the apparent effect of intervals 2 Biased estimates may result in estimating the effects of prenatal care or birth intervals on birthweight if improper procedures are used. See Rosenzweig and Schultz (1983) and Schultz (1984) for further discussion of this bias and procedures for minimizing it.
OCR for page 126
FIGURE 5-5 Infant mortality rate by length of preceding birth interval. NOTE: BWA—Botswana, BDI—Burundi, GHA—Ghana, KEN-Kenya, LBR—Liberia, MLI-Mali, NGA—Nigeria, SEN—Senegal, TGO-Togo, UGA—Uganda, ZWE—Zimbabwe. SOURCE: Demographic and Health Survey reports (see Appendix B). on mortality. We do not have any evidence of how much infant mortality might be reduced by changes in breastfeeding and contraceptive use that increase birth intervals. Prenatal Care Studies in Nigeria (Oruamabo and John, 1989; Wright, 1990) report that prenatal care is associated with a lower risk of LBW because some of the biological factors that contribute to the condition can be controlled or monitored through regular medical attention. Onyemunwa (1988) reported that in a study of Nigerian women, the majority (92 percent) of the women received at least one prenatal visit. However, the significant association with infant mortality was the timing of initiating prenatal care. Women who began receiving care during the sixth month of pregnancy or later were 48 percent more likely to experience the death of the child than those women who began prenatal care during the first through fifth months. However, it may be that some unmeasured variable was responsible for both the use of prenatal care and child death. Results from the DHS in sub-Saharan Africa indicate that the use of prenatal care varies across countries. Figure 5-6 shows that in most countries,
OCR for page 127
FIGURE 5-6 Prenatal care received in pregnancies over previous five years, selected countries of sub-Saharan Africa. NOTE: BWA—Botswana, GHA—Ghana, KEN-Kenya, LBR—Liberia, NGA—Nigeria, UGA—Uganda, ZWE—Zimbabwe. SOURCE: Demographic and Health Survey reports (see Appendix B). the majority of women report some prenatal care from either a doctor or a nurse. In Nigeria and Mali, however, 35 and 62 percent, respectively, did not receive any prenatal care.3 We do not know whether the number or timing of the visits or the type of advice given or heeded was sufficient to have any impact on the outcome of pregnancies or child survival. Malaria As discussed earlier, malarial infections in pregnant women are associated with increased risk of low birthweight (McGregor, 1984; Greenwood et al., 1989; Cot et al., 1992). This problem is most common among first births (McGregor, 1984; Greenwood et al., 1989). It is not clear what proportion of LBW in Africa is attributable to malaria. However, it may be the cause that can be addressed most successfully by health programs. Programs to address malaria during pregnancy are reviewed in Chapter 4. 3 In most countries, a few percentage of women reported receiving prenatal care from a traditional birth attendant or someone else. In Liberia, 17.1 percent reported another source.
OCR for page 128
Effects of Programs on the Incidence of Low Birthweight There are a number of studies that examine the effect of various interventions on the incidence of low birthweight. Since these studies rarely examine the resulting effect on infant mortality and most are small-scale studies, we have not made a complete survey of all of them. However, it is useful to mention a few to indicate the types of programs that have been tested and the conclusions reached. Small-scale studies of pregnant women have reported significant improvements in birthweights from nutrition supplementation. A study in Guatemala (Lechtig et al., 1975a,b) examined the effect of two supplements (one with high protein and caloric value, the other with a lower caloric value and no protein). The infants of the better-supplemented group had a mean birthweight that was 111 grams higher than the poorly supplemented group. The prevalence of LBW was 17 percent in the poorly supplemented group and 8 percent in the better-supplemented group. The study also demonstrated that caloric intake rather than protein appears to be the principal factor limiting fetal development. It also suggested that the total additional calories consumed during pregnancy appear to have been more important than the calories consumed during the trimester when supplementation was begun. In the village of Keneba in The Gambia, a food supplementation program for pregnant women produced a significant increase in birthweight (Prentice, 1983; Prentice et al., 1987). Women were given biscuits and fortified tea six days a week, providing 950 calories per day during the dry season and 1,110 calories per day during the wet season when food was less plentiful. Moreover, the supplement tasted good and was offered early in the morning when most of the women would not normally have eaten at home. Among the inadequately nourished mothers, the supplementation increased the mean birthweight by 225 grams and decreased the prevalence of LBW from 28 to 5 percent—a reduction of 82 percent. Among all women, the increase in birthweight was 120 grams, and a 68 percent reduction in the percentage of LBW infants was observed. VITAMIN A DEFICIENCY Vitamin A deficiency is widespread in the developing regions of the world, especially Africa and Asia. The condition results in a number of health disorders and is often manifested in problems with the eye and vision. Vitamin A has important functions in the human body. One of its physiological functions is in the formation and maintenance of epithelial tissue, which contributes to the body's immune system. Vitamin A is essential to growth of the skin as well as the mucous membranes lining the ocular and
OCR for page 129
oral cavities, and the respiratory, genitourinary, and gastrointestinal tracts. When vitamin A is deficient, the epithelial cells become dry and flat, hardening so that absorption of nutrients is reduced. Moreover, vitamin A deficiency may increase the risk of bacterial colonization or delay recovery. It is estimated that each year, approximately 10 million cases of childhood xerophthalmia (dryness of the conjunctiva and cornea) occur worldwide, with more than 500,000 resulting in blindness (Feachem, 1987). Vitamin A deficiency is often associated with specific regions of the world, such as Africa and Asia, where diets often lack carotene-containing foods. Studies of Vitamin A Deficiency, Morbidity, and Mortality The lack of data makes it difficult to assess the prevalence of vitamin A deficiency for sub-Saharan Africa or for individual countries. However, a few studies in sub-Saharan Africa provide some estimates. In the Lower Shire River Valley of Malawi, for example, 5.4 percent of children less than 6 years of age experienced night blindness and 3.9 percent of children suffered active corneal disease (Tielsch et al., 1986). In southern Ethiopia, De Sole et al. (1987) reported an average prevalence of vitamin A deficiency of 5.4 percent in boys aged 6 months to 6 years, and 5.5 percent in girls of the same age. A few studies in sub-Saharan Africa support the association between vitamin A deficiency and more severe cases of diarrhea, measles, and respiratory infections. In southern Ethiopia, De Sole et al. (1987) reported that the prevalence of severe cases of diarrhea in the previous year was twice as high among children with vitamin A deficiency as among children without the deficit. A number of studies have indicated that children with severe vitamin A deficiency and xerophthalmia experience elevated rates of mortality. Death rates among children hospitalized with these conditions range on average from 15 to 25 percent (Kuming and Politzer, 1967; Sommer et al., 1975; Brown et al., 1979; Sommer, 1982). Most studies do not attribute the deaths to the deficiency of vitamin A but rather to concurrent illnesses and conditions that are exacerbated by reduced levels of vitamin A, such as PEM, diarrhea, and respiratory infections. Studies of Vitamin A Supplementation and Mortality Reduction A number of vitamin A supplementation studies have been conducted, principally in Asia, that demonstrate a positive effect on reducing infant and child mortality. In a meta-analysis of vitamin A supplementation studies conducted in Asia, Tonascia et al. (personal communication, 1992) estimated the weighted reduction in mortality attributable to vitamin A deficiency
OCR for page 130
for children 6 months or older (up to approximately 83 months at follow-up) to be 34 percent. They noted that the magnitude of the effect may depend on the extent and severity of nutritional deficiencies, the cause-of-death structure of infant and child mortality, cultural and environmental factors, and the study design and implementation. Sommer et al. (1986) conducted a randomized controlled community trial of periodic large-dose vitamin A supplementation in northern Sumatra, Indonesia. During the follow-up period, almost all deaths in program villages were among children who had not received the supplement. Preschool children in the treatment group experienced a 34 percent reduction in the noninfant mortality rates compared to the comparison group. In a reanalysis of the same data, Tarwotjo et al. (1987) reported that among children between ages 3 and 11 months, mortality rates were 0.9 per 1,000 in the treatment group, compared to 12.0 in the nontreatment group and 6.0 in the control areas. They concluded that the 34 percent reduction reported in the previous article may be an underestimate because the early analysis was based on intent to treat (i.e., results for all children allocated to one regimen are compared with those allocated to the other, regardless of whether they received the regimen assigned) at the community level, rather than on those actually receiving the intervention at the individual level. In southern India, Rahmathullah et al. (1990) reported that vitamin A supplements equivalent to the level recommended by international groups, when given on a weekly basis, reduced the relative risk of mortality in children under 5 years of age by 54 percent. Mortality rates in the control group were 10.5 per 1,000 compared to 4.8 per 1,000 in the treatment group. West et al. (1991) conducted a trial of the efficacy of vitamin A supplementation in reducing childhood mortality in Nepal. In a randomized, double-blind community trial of almost 29,000 children aged 6-72 months, supplemented children received 60,000 retinol equivalents every four months and the placebo-treated group received 300 retinol equivalents. After 12 months, the relative risk of death among the supplemented group was 70 percent that of the control group, supporting the hypothesis that vitamin A supplementation can contribute to lower overall child mortality. Daulaire et al. (1992) found that the risk of death for Nepali children aged 1 to 59 months in supplemented communities was 26 percent lower than in unsupplemented communities, with the largest reduction of 49 percent among children 6 to 11 months of age. Although most studies find an association between vitamin A supplementation and a reduction in mortality, a prospective double-blind placebo-controlled study conducted in Hyderabad, India, reported no significant difference when supplementation occurred. The study, conducted by Vijayaraghavan et al. (1990), indicated that mortality rates were similar in the groups receiving the supplement and the placebo.
OCR for page 131
Herrera et al. (1992) also found no mortality effect in a controlled, masked randomized trial of 29,615 children in northern Sudan where there was a clear association between dietary vitamin A and mortality, based on a nonsignificant difference in the number of deaths between the comparison and treatment groups. In that study, some children were assigned to a group that received 200,000 international units (IU) of vitamin A and 40 IU of vitamin E every six months. The control group received only 40 IU of vitamin E. Over the 18 months of the study, there was no apparent effect of large-dose vitamin A supplementation on mortality. It is not understood why the findings of this study in northern Sudan differ from those conducted in other parts of the world. Perhaps political and social conditions were impediments to the study. On the other hand, the nonsignificant effect of vitamin A supplementation may be based on differing disease epidemiologies due to ecological differences. For Africa, the most recent evidence on the effect of vitamin A supplementation, morbidity, and mortality comes from an experiment in northern Ghana. There, in an intervention trial among mildly vitamin A-deficient children conducted by Ross et al. (Beaton et al., 1992) a reduction of 20 percent in mortality is reported among the supplemented group compared to the control group (probability less than .003). SUMMARY A variety of nutrition-related conditions exacerbate the mortality effect of several diseases. Protein-energy malnutrition, although rare during the first six months of life, is common among children between 6 and 23 months of age, and is associated with increases in diarrheal and other diseases. Wide variations in the prevalence of malnutrition are observed in sub-Saharan Africa, and a number of studies suggest that the effect of nutritional status is less important in Africa than in Asia. Many programs that address malnutrition do not necessarily reduce mortality. Most nutrition interventions do not gather information on mortality. Supplemental feeding programs often do not reach the target populations; growth monitoring may make mothers aware of how their children are developing, but it has little direct effect on reducing mortality. Mild and severe vitamin A deficiency seems to be associated with excess mortality. Studies conducted principally in Asia report higher overall and cause-specific mortality among children with this deficiency. Again, little evidence exists to make conclusions about the effect of vitamin A supplementation in sub-Saharan Africa.
Representative terms from entire chapter: