Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Hypotheses About Reproductive Patterns and Women's and Children's Health In developing countries, where resources are scarce, health services are poor and not readily available, and infectious diseases are common, it is difficult to maintain good health. Although the primary causes of poor health in such an environment are often beyond the control of individual families (at least in the short run), there are ways in which they can act to reduce the risk of illness and death. Choosing to limit fertility, to delay the onset of childbearing, to space births, and to breastfeed are among the actions that may reduce the risk of illness and death to women and children. The central objective of this report is to assess the effects of particular repro- ductive patterns for the health of women and children. Of course, reproductive patterns are not a direct cause of death, in the way hemorrhage or infectious diseases are, but rather they may be associated with conditions leading to death. The chapter begins by examining hypotheses about the direct effects of reproduc- tive patterns, first on women's health, then on children's health. By direct effects we mean the biological and behavioral mechanisms through which a change in childbearing patterns might affect women's and children's health directly. Next, we discuss hypothesized indirect effects of changes in reproduction on women and children's health. Indirect effects refer to changes in household structure and parental roles and new or foregone opportunities that are a consequence of changes in the number and timing of births that may themselves influence health. While the report focuses primarily on possible direct and indirect effects of reproductive patterns on women's and children's health, interaction or synergistic effects may also be operating. In these cases, the presence of two conditions or behaviors alters their effects. For example, the joint detrimental consequences of 12
HYPOTHESES 1 3 using oral contraceptives and smoking are noted in Chapter 4. Numerous other interaction effects may be occurring, but they are beyond the scope of this report We then review other possible explanations for relationships between repro- ductive patterns and health, such as the possibility that other health behaviors (e.g., use of modern medical care) affect both reproductive patterns and health. Finally, we discuss the types of evidence on which our findings are based. DIRECT EFFECTS OF REPRODUCTIVE PATTERNS ON HEALTH Reproductive Patterns and Women's Health Practicing physicians and midwives have observed that women who have many children and women with certain reproductive patterns are at higher risk of poor health and mortality. These observed relationships may be directly causal or may be due to confounding factors. In addition, it is possible that the poorer health of certain women or their children may lead to certain reproductive patterns, such as reduced fertility. Hypotheses concerning the effects of repro- ductive patterns on the health of women are summarized in Table 2.1. There are at least five reasons to believe that women who limit their fertility will have better health. Fast, each time a woman becomes pregnant, she is at risk of pregnancy complications and related complications that she does not face when she is not pregnant. Thus, over the course of a lifetime, women who have many children are more likely than women who have fewer children to experi- ence reproductive illness or mortali~merely because they are exposed more often to the risks associated with each pregnancy. The risks associated with pregnancy include pregnancy-induced hypertension, ectopic pregnancy, hemor- rhage, obstructed labor, infection, and, for those not wanting to be pregnant, the hazards of unsafe abortion. The health benefits of fewer pregnancies are likely to be greater in regions where prenatal and intrapartum care are poor, because the risks associated with each pregnancy are greater. These issues are discussed in greater detail in Chapter 3. Second, in addition to having fewer pregnancies, women who control their fertility can avoid pregnancies that may pose higher-than-normal risks to their health. Research suggests that women who have had many previous pregnancies may be at higher risk of poor health from a given pregnancy than women who are pregnant for the second, third, or fourth time. Women who have had many previous pregnancies may be at higher risk of morbidity and mortality because of the cumulative toll of previous pregnancies and because of previous reproductive injury. In particular, they may be more likely to experience complications such as uterine rupture and hemorrhage than women who have had fewer previous pregnancies. The first pregnancy also appears to have a higher risk than the second, third, or fourth pregnancy. Women who are having their first child appear to be at higher
i4 CO~RACE=ION ID REPRODUCTION TABLE 2.1 Hypothesized Direct Effects of Reproductive Patterns on Women's Health Reproductive Pattem Hypothesized Effect Number of pregnancies High-risk pregnancies First pregnancies High-order pregnancies Pregnancy at very young maternal ages Pregnancy at older ages Short interbirth intervals Unwanted pregnancies ending in unsafe abortion Large family size Pregnancies for women already in poor health Use of contraception Each pregnancy carries a risk of reproductive morbidity and mortality Adaptation to pregnancy for the first time Hemorrhage; uterine rupture; previous reproductive . . injury Inadequate development of reproductive system and incomplete growth Adequate prenatal and interpartum care less likely Body in poorer condition for pregnancy and childbirth Inadequate time to rebuild nutritional stores and regain energy level Abomons performed by unsafe means increase exposure to injury, infection, hemorrhage, and death Reduced availability of family resources for women's health and nutrition Aggravated health condition Direct health risks and benefits associated with ca~tracephve methods risk of obstructed labor, pregnancy-induced hypertension, and birth complica- tions. However, since any woman who has children must have a first birth, women who desire children obviously cannot choose to avoid first pregnancies in order to lower their risks of morbidity or mortality. Reproductive morbidity and mortality may also be more common for women who become pregnant at the very beginning and at the end of their reproductive years. Pregnancy may be more stressful physiologically to very young women,
HYPOTHESES 15 because their reproductive systems are not yet fully mature and they may not yet have completed their growth. Young girls may also be less likely to detect a pregnancy early on or, for a variety of reasons, they may deny the pregnancy. In both cases they may seek prenatal care later in pregnancy than desirable or delay having an abortion (in situations where either is available). Older women may encounter problems more frequently during pregnancy and birth because the ability of their reproductive systems to cope with the burden of pregnancy has declined with age. Evidence of decline in the function of the reproductive system with maternal age includes an increased incidence of fetal chromosomal abnor- malities and spontaneous abortion. Pregnancies that begin shortly after a previous birth may also pose higher risks for women. Short interbirth internals, especially if accompanied by intensive breastfeeding, may prevent a woman from rebuilding depleted nutritional stores before He next pregnancy begins. This problem is likely to be more serious among women who are malnourished to begin with and may be exacerbated by a sequence of closely spaced pregnancies. A pregnancy that occurs when a woman's health is already jeopardized is likely to pose a greater risk than a pregnancy for a healthy woman. Women who are malnourished, who are seriously ill, or who have chronic health conditions are clearly at higher risk than healthier women. By avoiding pregnancy, women with health problems may substantially improve their own chances for survival and good health. Third, in addition to reducing the total number of pregnancies and avoiding potentially higher-risk pregnancies, women in many developing countries can also substantially reduce their risk of reproductive morbidity and mortality by using contraception to avoid unwanted pregnancies rather than resort to unsafe abortion to terminate such pregnancies. In many developing countries, abortion is illegal and is often performed by untrained personnel in unhygienic conditions. Abortions attempted by women themselves or performed by abortionists under septic conditions substantially increase a woman's risk of infection, injury or hemorrhage, and death. The three hypothesized mechanisms just described suggest that changes in reproductive patterns may improve health by decreasing exposure to infection, injury, and other reproductive complications. A fourth possible mechanism is the use of contraception itself, which may affect women's health. There may also be a fifth and more general effect of changes in reproductive patterns on women's health. Most families in developing countries have limited resources that must be allocated to a variety of family needs. Families with fewer young children to care for are likely to have more resources (including time, food, and money) to devote to the health of each family member. For example, women in smaller families may have more time to go to a clinic for treatment of an illness or for a prenatal . . VlSlt.
6 CO=RACE=ION AND PRODUCTION Reproductive Patterns and Children's Health A woman's reproductive pattern may also have important effects on the health and survival chances of her children. Children's well-being, especially in the first year of life, is highly dependent on their mothers' health during and after preg- nancy. For this reason, some of the hypothesized effects of reproductive patterns on children's health are closely related to the effects of reproductive patterns on women's health. Specifically, children who are firstborn or are of high birth order, children born into larger families, children born to very young or older mothers, children born after short previous interbirth intervals or before a short subsequent interbirth interval, and children who were unwanted at the time they were conceived may be at higher risk of poor health and mortality than other children. It is also possible that a mother's use of contraception may affect her child's health directly, for example through effects on lactation. These hypothe- ses are summarized in Table 2.2. Birth Order Since nulliparous women experience more problems during pregnancy, b~rst- born children may be less healthy at birth, may weigh less at birth (because of poorer intrauterine growth or shorter gestation), and may experience more trauma during birth. The parents of firstborn children may also be less experienced in child care, although this explanation for poorer health among firstborns seems less plausible in societies in which new parents frequently live with older, more experienced relatives. Reducing the number of first births that women have is obviously not a sensible policy objective, since families choosing to have chil- dren must have a first birth, but delaying first births could be an important policy objective, particularly for very young women. Children born of higher-order pregnancies may experience higher risks of morbidity and mortality for at least two reasons. First, as discussed above, because of the cumulative toll of numerous previous pregnancies and associated breastfeeding on maternal nutritional stores (described as the "maternal depletion syndrome"), mothers of higher-order children may be in poorer health prior to and during pregnancy, as well as after birth. Women who have reached high parity (fifth and higher parity) are also more likely to have experienced injures during childbirth, which may complicate a higher-order pregnancy and birth. Thus, higher-order children may be at greater risk of poor intrauterine growth, greater trauma during birth, and, more generally, poorer health than children born at orders 2, 3, and 4. Second, children born at higher orders may be in poorer health because their families have fixed resources (such as time, money, food, and shelter) and more children to care for with these resources. On one hand, the sixth or seventh child in a poor family may receive less time and attention from parents than the first or second child did at a comparable age because there are now many children who need attention. On the other hand, older children may help to care
HYPOTHESES 1 7 TABLE 2.2 Hypothesized Direct Effects of Reproductive Patterns on Children's Health Reproductive Pattern Hypothesized Effect Firstbom children Higher-order children Large families Children born to very young mothers Children ham to older mothers Short interbirth intervals Unwantedness Maternal death or illness (e.g., chronic infection such as ADS) Contraceptive use More frequent maternal problems during pregnancy and childbirth; parents have less experience wad child care; poorer intrauterine growth Matemal depletion; cumulative effect of earlier maternal reproductive injury; poorer intrauterine growth limited family resources allocated to more children; spread of infection among family members Inadequate development of maternal reproductive system and incomplete matemal growth; young mothers less likely to receive adequate prenatal and intrapartum care or to provide good child care Greater risk of birth trauma; greater risk of genetic abnormalities Inadequate maternal recovery time (matemal depletion); competition among similar aged siblings for limited family resources; early termination of breastfeeding; low biIthweight; increased exposure to infection from children of similar ages (Conscious or unconscious) neglect; child born into a stressful situation Early termination of breastfeeding; no maternal care; disease may be passed to child Hommonal contraception may interrupt breastfeeding for younger children in large families and may contribute to the economic well- being of the family, thus increasing both total family income and possibly per capita income. Another hypothesis is that children who have a large number of siblings, regardless of their own birth order, are more likely to be in poor health. When there are a large number of children in a household with limited resources, there is
~ ~ CO=RACE=ION ED PRODUCTION increasing competition among children, so each child not just children of higher birth ordermay receive less time, attention, and care. Moreover, a child who has a larger number of siblings, especially if they live and sleep in crowded quarters, will be at increased risk of contracting infectious diseases. Maternal Age Children born to very young mothers and to older mothers may also be at higher risk of poor health and mortality. In the case of the children of very young mothers, as argued above, the reason may be that pregnancy is more stressful physiologically for adolescents because their reproductive systems are not yet fully mature, and they may not yet have completed their growth. As a conse- quence, adolescent girls may be less able to produce healthy babies and may experience more trauma during childbirth. A second possible reason that the children of very young mothers may be in poorer health is that these mothers may be less likely to seek and receive adequate prenatal care and may be less ready psychologically and materially to care for their children. Children born to older mothers may also experience greater risks of mortality and morbidity. As argued above, women at the older end of the reproductive span may encounter more frequent problems during pregnancy and birth because the capability of their reproductive systems to cope with the burden of pregnancy has declined. Children born to older women may have poorer health at the time of birth because of the greater likelihood of birth trauma or genetic abnormalities. Birth Spacing Children born either after or before short interbirth intervals may also be at higher risk of morbidity and mortality, for several reasons. The first is related directly to the hypothesized effect of close pregnancy spacing on maternal health. For women living in poverty who are predisposed to malnutrition or poor health, a very short interval between one pregnancy and the next may not provide adequate time for rebuilding nutritional stores and for physiological recuperation. The consequences for children born after short interbirth intervals may be poorer intrauterine growth as well as a higher risk of preterm birth. We noted above that competition among siblings in large families for scarce resources may mean that higher-order children, or possibly all children, may be at greater risk of poor health. Competition among children for family resources may be even more of a problem among children of similar ages, especially when they are young, because they have similar needs. When two births are spaced very closely, each child may not receive as much care and attention as he would if he did not have a sibling of roughly the same age. Close birth spacing can also create even more direct competition among siblings in the case of breastfeeding.
HYPOTHESES 19 A mother who becomes pregnant soon after a child is born is likely to wean that child sooner than she would had she not become pregnant again. Since breastfeed- ing is an important determinant of child health in many societies, premature termination of breastfeeding often substantially increases a child's exposure to infection and increases the risk of malnutrition. Close birth spacing may also increase children's exposure to infectious dis- eases by fostering transmission of infections among household members who are of similar ages. Many infectious childhood diseases affect a relatively narrow age range. If there is more than one child in the household in that age range, the chances of introducing the disease to the household and transmission of higher or repeated doses of the infectious organism may be dramatically increased, thus increasing the likelihood of multiple or more severe illnesses. This is especially a problem with diarrhea! diseases, for which repeated incidence may result in malnutrition, and with measles, for which transmission in the household may be associated with more severe and fatal infection. Unwanted Births Finally, children who were unwanted at the time they were conceived may be at greater risk of poor health and mortality than other children. In households with limited resources, parents may, consciously or unconsciously, discriminate against unwanted children in the allocation of food parental time and attention, or preventive and therapeutic health care. An alternative hypothesis is that children who are unwanted often are conceived when the family or the mother is under economic, social, or psychological stress, and the child is at greater risk simply by being born into a stressful situation. Maternal Illness arid Death and Effects on Child Health Reproductive patterns may have another type of effect on the-well-being of families and especially children, through their association with maternal morbid- ity and mortality. The death of a mother, whether due to reproductive or other causes, is likely to cause major disruption in the lives of her children, as well as a breakup of the household in which she and her children lived. In addition to the serious emotional consequences for children, the disruption following their mother's death may be extremely detrimental to their physical health, particularly if they are very young and breastEeeding has ended. Serious illness or reproduc- tive injury may also prevent a woman from caring adequately for her children, with consequent negative effects on their health and survival chances. Maternal incapacity and death may be a growing burden on societies in which AIDS affects substantial numbers of women of reproductive age. Furthermore, it is possible that certain infectious diseases can be passed from mother to child.
20 CONTRACEPTION kD REPRODUCTION INDIRECT EFFECTS OF REPRODUCTIVE PATTERNS ON HEALTH Changes in reproductive patterns through control of fertility are also hypothe- sized to have important indirect effects on the health of women and children. Although some means of controlling fertility (including withdrawal and absti- nence) are theoretically always available to couples, the availability of modern methods of contraception brings the process of fertility regulation more firmly into the control of couples and of women themselves. Successful intervention in what was formerly seen as a natural process may change couples' or families' attitudes about their ability to make changes in other traditional practices. These practices may include those related to child care, prenatal diet and care for women, and the use of modern health services. The ability to regulate fertility may also increase women's autonomy and give them greater authority to make decisions concerning their own health and the health of their children. In some settings, increased control over fertility and the increased predictabil- ity of pregnancy timing that comes from contraceptive use may also make it easier for women to finish their education, to participate in the labor force, or to hold better-paying jobs. Higher educational attainment, work outside the home, or a better job are all likely to increase family income, which can then be spent on a more nutritious diet, better clothing and shelter, improvements in sanitation and water supply, and health services for all family members. In some societies, the fact that women make financial contributions to the household budget may also give them additional decision-making power in allocating household resources to themselves and their children, thus potentially improving their health. Women who have fewer children or fewer young children to care for may be under substantially less physical and psychological stress than women with very large families, especially women in poor families, in which the resources to care for children are often scarce or inadequate. Furthermore, the ability to control fertility may also change a woman's outlook on life and may contribute to her psychological well-being (Dixon-Mueller, 1989~. In other settings, the ability to control fertility may create new tensions in the family, at least in the short run. The process of making explicit decisions about reproductive matters may lead to disagreement between spouses, conflicts be- tween parents and their adult children about family size, and anxiety about violating traditional, often religious ideals surrounding sexual practices and child- bearing. Another hypothesized indirect effect of family planning on health relates to the use of health services. In countries where the program is strong and well-organ- ized, family planning services may serve as an introduction for women to mater- nal and child health care services. Contact with family planning clinics may provide these women with information about how the health care system worked, referrals to other types of care, and often the confidence to deal with other types of health care workers. In other areas the opposite case may occur, with women being introduced to family planning through contact with the health care services.
HYPOTHESES 2 ~ OTHER POSSIBLE EXPLANATIONS Reproductive patterns and women's and children's health may be associated with one another, without the former causing the latter, either directly or indi- rectly. A third factor may cause both. For example, a baby born after a pregnancy of short gestation is more likely to be in poor health and to be born within a short interval after the preceding birth. Although the short interval is not the cause of the child's poor health, both the short interval and the child's poorer health are due to the short gestation of the pregnancy. Or it is possible that the direction of causation runs from health to the reproductive pattern. For example, if a child dies shortly after birth, breastfeeding will be shorter than it otherwise would have been. As a consequence, postpartum amenorrhea (the infertile period following a birth, which is related to the duration and intensity of breastfeeding) will be shorter. Unless the couple compensates by practicing contraception longer than it would have had the child not died, the result will be a shorter birth interval. In this case, the birth interval is short because the child died. All couples make implicit or explicit choices about whether or not to control their fertility, and, if they do decide to intervene, about how many children to have and when to have them. Decisions related to fertility control, family size, and birth spacing are unlikely to be entirely independent of women's and chil- dren's health or of other decisions people make which have consequences for their health. For example, a couple that experiences achild's death may choose to have another child to "replace" the one who died. Or when child mortality rates are high, couples may wish to "hedge" against the possibility of a child death by having additional births to increase the chances that a certain number of children will survive until maturity. Either of these mechanisms could result in a positive relationship between a large family and high child mortality, but, in these cases, fertility is high because mortality is high. A similar relationship could arise if some couples choose to have fewer children so that they can "invest,' more in the health and education and material well-being of each child. In this case again, the relationship runs from (desired) health to fertility. Another hypothesis about the observed association between fertility patterns and the health of women and children is that couples who use contraception to limit family size, space their children, and avoid unwanted pregnancies may simply be different from other couples in ways that affect both health and childbearing. An example of a couple's choices that affect both fertility and health involves the use of health services. As we noted above, in settings in which family planning programs are very strong, contact with family planning services may introduce families to other health services that they were previously unaware of. It is also possible that families who have previous experience with health services and are accustomed to using them are more likely to be aware of available family planning services and are also more likely to feel comfortable using them. Thus, it may be a familiarity with He health system that causes increased contraceptive use rather than vice versa
22 CONTRACEPTION ED PRODUCTION Parents who take action to prevent illness and who seek medical care when illness occurs may also be more willing to attempt to control their fertility, to try using contraception, and to have the persistence to seek out contraceptive services when they are not readily available. Undertaking both health-related behavior and control of fertility may require a nonfatalistic view of life, in which it is possible and socially acceptable to try to intervene in natural processes such as illness and conception. For many couples in developing countries, obtaining effective health care and fertility control methods may require substantial persis- tence and knowledge of how to obtain information and deal with an ineffective delivery system. Adults who have these skills, abilities, and beliefs are likely to use health services, to carry out health-improving practices at home, and to use contraceptive services. Finally, families with more financial resources and education are likely to be in better healthbecause they live in a better physical environment, because their diet is better, and because they receive better health care. These same families may also have fewer children because they prefer smaller families or because they have better access to contraceptive facilities. In short, the observed association between reproductive patterns and women's and children's health may result from the fact that families who take measures to protect their members' health are also more likely to control their fertility, as well as from causal effects of reproductive patterns on health. This possibility is important to consider if we want to determine the likely effects of changes in reproductive patterns on health, and we return to it in subsequent chapters. AVAILABLE EVIDENCE Probably the most direct and convincing way to distinguish among the hy- potheses discussed above would be to conduct controlled randomized double- blind experiments. For example, some women would be randomly assigned to have their fast child at age 16 and others would wait till age 20. Or two otherwise identical sets of communities could be chosen for study and contraception would be provided in one group and not the other. Such experiments, however, are difficult to perform, and they would raise ethical problems even if they were feasible. Consequently, it is not surprising that the extant evidence is based not on experimental studies, but rather on observational data collected through sur- veys or longitudinal data from nonexperimental settings. For policy purposes, we would like to be able to distinguish among the alternative hypotheses discussed above. To illustrate, take the example of the relationship between young maternal age and child health. If the reason that the children of teenagers are less healthy than the children of older women is that teenagers are less likely to seek prenatal care, further research could seek to understand the reasons and to try to remedy the situation. If the relationship between young maternal age and child health is physiological, due to incomplete
HYPOTHESES 23 maternal growth, a policy that helps women postpone childbearing until later ages should result in improved child health. If, alternatively, teenage mothers have poorer child health outcomes because girls who become pregnant as teenagers are poorer mothers regardless of the ages at which they bear their children, then postponing childbearing for these women will not necessarily have beneficial effects for the children's health. A key question is whether teenagers outgrow- physiologically, psychologically, or economicall~whatever causes their chil- dren to have poorer health outcomes, or whether it is a persistent characteristic of the types of women who become pregnant as teenagers. The evidence regarding the hypothesized effect of maternal age on children's health typically comes from analyses of data on maternal ages at children's births and the children's health (typically survival measures) that show health outcomes to be poorer for children born to young mothers under age 20 than to older mothers ages 2~29. Earlier studies tended to consider the simple correlation between maternal age and child health. Such a correlation, however, says nothing about which among the possible mechanisms discussed in this chapter might account for this association. For example, births to teenage mothers are more likely than births to older mothers to be first-order births, which have a greater risk of low birthweight and other problems, as discussed above. Hence, to assess the effect of being born to a teenage mother, it is important to hold constant the effect of parity and, in essence, consider age differentials within parity categories. As another example, teenage mothers may have less education or lower incomes than older mothers. In such a case, it is desirable to control for education and income, so as not to attribute to age what is really a result of low education or income. More recent studies have used multivanate statistical methods in an attempt to deal with these issues. The remaining possibilities are more difficult to distinguish, especially in large-scale population surveys, which have provided much of the evidence on the relationships between reproductive patterns and child health. To assess the possibility of a physiological mechanism, one would ideally like to consider gynecological age (years since menarche) and assess its relation to pelvic size and the women's nutritional status, and how these, in turn, affect her baby's health. Psychological maturity is even more difficult to assess; for example, teenagers tend to be more egocentric than older women and less likely to respond to the needs of others. One could look at the health-related behaviors that psychological development could be expected to affect, to see, for example, if pregnant teen- agers are less likely to use prenatal care (appropriately). However, assessing the effects of health care on health outcomes is complicated by the fact that unobser- ved factors that affect decisions to seek health care may also affect health outcomes. For example, as noted above, women with a greater concern about health may be both more likely to seek health care and more likely to engage in other behaviors that promote good health. In such a case, a simple correlation between health care and health would overstate the direct effect of the former on
24 CONTRACEPTION ID REPRODUCTION the latter. In other cases, women in poor health may be more likely to seek health care because of specific needs. The last example illustrates a problem generic to all research that uses nonex- perimental observational data: no matter how many known correlates and poten- tial confounding variables are controlled (e.g., parity, education, and income in assessing the effect of young maternal age on child health), it is always possible that there may be unobserved factors that may be correlated with both the reproductive patterns of interest and the health outcome being considered that contribute to their relationship; hence, even when other observed factors are controlled, it is possible that the estimated relationship is not entirely causal. Some analysts have used structural models to deal with these issues, but to date such methods have been used infrequently to study the effects of reproductive patterns on women's and children's health, particularly in developing countries. An example of using structural models is the simultaneous equation frame- work as commonly used in economics, for example, when fertility is modified in response to the expectation and occurrence of child mortality, while child mortal- ity may be affected by reproductive patterns. To disentangle the two effects, which operate in opposite directions, and to statistically distinguish only the latter effect, which is one focus of this report, the scientist requires information on an independent variable that directly affects fertility and does not directly affect child mortality. This variable could provide the information needed to identify statistically and thus estimate the one-way effect of independent changes in reproductive patterns on child (or maternal) mortality. The problem is to find such an identifying variable that can be plausibly excluded from entering the mortality determining process. This is difficult when studying complex, jointly determined household demographic processes such as fertility and child mortal- ity, especially when the assumptions made in choosing identifying variables are often unverifiable. The evidence reviewed in this report comes from many different sources: large-scale population-based surveys that ask women about their pregnancies and children's mortality, smaller-scale longitudinal studies and hospital or clinic samples that often include physiological information, and data on births and deaths in historical populations. The characteristics and advantages and disad- vantages of these various types of studies are reviewed in the chapters ahead. However, it is important to note that none of the studies to date has simultane- ously addressed all the different types of issues just discussed. This report focuses principally on the evidence from previous research con- cerning the hypotheses about the direct effects of family planning or reproductive control on the health of women and children. Although it is possible that indirect effects may be equally or more important than the direct effects under considera- tion, investigation of these hypotheses is outside the scope of this report. We return to the subject of the indirect effects in Chapter 7.
