INFANT AND CHILD MORTALITY

Mortality affects the volume of a population. Deaths are not equally distributed among all groups; rather many unique patterns have been identified. For example, the probability of dying is high among both extremes of a population's age structure—the very young and the very old. As a general pattern, the death rate is relatively high at age zero, reaches a minimum in the range from ages ten to fifteen, and then begins to increase gradually with increasing age. This increase becomes marked after age forty-five or fifty (Coale 1965).

Infant and child mortality are important because the largest mortality risk differentials between a society with high mortality and one with low mortality are always found within infancy and childhood. Because infant and child mortality are often related to general levels of health and living conditions, they are often thought of as measures by which nations can gauge their current level of socioeconomic development and societal cohesiveness. Comparisons can then be made with past mortality estimates, along with those of neighboring nations, allowing a country to make projections concerning its future development and public policy programs. Assuming this general precept, researchers investigate historical mortality trends in an attempt to isolate general patterns of infant and child mortality. The goal is to provide information for policy makers, raise standards of living for residents of a nation, thereby reducing early childhood mortality.

MEASURES AND DATA SOURCES

The infant mortality rate (IMR) is defined by demographers as the probability that a newborn infant will die before it reaches its first birthday. This probability is generally computed by calculating the ratio of the number of infant deaths under one year of age in a given year per number of births in that given base year. This approach is termed a period IMR because it only utilizes data from one year to arrive at its estimate. This method is not entirely accurate in that it 1) utilizes infant deaths of those born in the preceding year and 2) misses those infant deaths that occur in the following year. If the number of births and deaths does not change drastically from year to year, then this method results in a close approximation of an infant's probability of survival to age one (Shryock and Siegel 1976).

A more accurate measure of infant mortality is a ratio that involves the use of infant birth and death data for two years. In this instance, a researcher is able to calculate more precisely the true probability of infant death by identifying all infant births in the base year of interest and linking them to their respective infants' deaths at less than one year of age in both the base year and the following year. This method is referred to as the cohort IMR for a given year and population. Infant mortality rates are expressed in terms of infant deaths per 1,000 live births (Shryock and Siegel 1976).

In order to capture different causes that are known to affect the timing of infant deaths within the first year of life, researchers often decompose infant mortality into various components. Neonatal mortality usually refers to deaths of infants under 28 days of age, whereas postneonatal mortality involves the deaths of infants aged 28 to 364 days. Due to the large proportion of pregnancies that result in fetal deaths, and the existing similarity between the causes of death for fetuses and for neonates, other complementary measures have been constructed. Fetal mortality refers to fetal deaths at a gestational age greater than 28 weeks, and perinatal mortality is defined as fetal mortality plus infant deaths at less than 7 days of age. All corresponding rates are constructed in much the same manner as the IMR, including the use of either the period or cohort method of estimation. For both the fetal and perinatal mortality rates, researchers usually include the number of fetal or perinatal deaths (whichever may be the rate of interest) within the denominator of the rate along with the number of births for the population of interest (Shryock and Siegel 1976).

Conventionally, the number of deaths that occur among children aged one to four is a general measure of child mortality. In some cases, though, the child mortality rate may include the IMR (in which case it takes into account all deaths under five years of age). As a rule, most researchers interested in child mortality separate out infant mortality in order to arrive at a clearer picture of the specific causes of death that affect children who survive infancy. Child mortality rates are generally calculated by dividing the number of deaths of children aged one to four by the total number of children aged one to four in the population and multiplying by a 1,000 (Shryock and Siegel 1976).

The most common source of data on infant and child mortality in more developed countries (MDCs) is national vital statistics registries, which were established in some countries as early as the beginning of the eighteenth century (e.g., Sweden). Researchers are able to access detailed birth and death records for both infants and children and analyze their trends and patterns. Throughout the world, the most stable data on infant and child mortality have come from those countries with established vital registration programs. For those MDCs that do not have adequate systems of vital registration and for the majority of less developed countries (LDCs) that do not have a vital statistics system in place, other techniques exist for estimating infant and child mortality (Palloni 1981). In most of these countries, population censuses and surveys of a sample of the total population are utilized to indirectly estimate mortality rates and trends (Hill 1991). Techniques developed in the late 1960s and early 1970s have been refined and improved upon in a variety of ways (United Nations 1990). Most techniques involve the use of birth histories collected from women in the population.

EPIDEMIOLOGIC TRANSITION AND INFANT AND CHILD MORTALITY

Early research on infant mortality revealed an array of factors that presumably led to infant mortality declines in certain countries, specifically certain Western European MDCs. The experiences of these nations are the basis for a set of general hypotheses regarding the decline of infant mortality. These nations serve as test cases because of the nature of the available data and the familiarity of researchers with them. Ideas have been developed concerning the relationship among certain economic, social, and technological developments and the reduction of adult, childhood, and infant mortality. Epidemiologic transition theory, as elucidated by Omran (1971,1977), gives researchers a theoretical framework for understanding these processes as they have occurred throughout different time periods and within different nations. This theory outlines several stages that specific nations have moved through on their way to their current mortality patterns. Many have used this theory to investigate current infant mortality levels among various countries and within specific ones (Frenk et al. 1989; Pampel and Pillai 1986).

In applying the precepts of epidemiologic transition theory to the case of infant mortality, researchers have used diverse historical sources. Epidemiologic transition theory outlines three stages through which various Western European nations moved through as they advanced toward a higher state of economic and technological development. The first two stages, termed the Era of Pestilence and Famine and the Era of Receding Pandemics, characterize most of human existence. Within these eras, infant and child mortality were relatively high and primarily due to the prevalence of infectious disease. In the third stage, the Age of Degenerative and Man-Made Diseases, infectious diseases receded and came under medical control. Infant and child mortality fell accordingly. Other causes of death now became important, and interest in them rose. In this sense, infectious infant mortality is currently associated with countries that are less developed and populations within countries that live at disadvantaged socioeconomic levels (Omran 1971, 1977).

To better understand changes in infant mortality according to epidemiologic transition theory, researchers must be able to access data that detail the cause of infant mortality death. But as has been noted, very few countries possess the data repositories, either current and past, to adequately examine infant mortality declines and their component causes. The use of neonatal and postneonatal IMRs were one way of moving forward with these types of studies without having the necessary data for detailed analyses. Neonatal and postneonatal mortality became associated with cause-of-death groupings according to general preconceived notions of etiologic mechanisms.

Bourgeois-Pichat (1951) presumed that infectious infant mortality was primarily due to environmental factors that were wholly amenable without medical intervention. This category was termed "exogenous" infant mortality (e.g., infectious, parasitic, and respiratory diseases), while "endogenous" infant mortality was defined as those causes of infant death that were primarily due to natal and antenatal factors (e.g., congenital malformations, prematurity). Neonatal mortality became a proxy for endogenous causes of death while postneonatal mortality was associated with exogenous mortality. All that was now minimally required to continue analyses of infant mortality transitions was information on the timing of infant death (Bogue 1969; Bouvier and Van der Tak 1976). Although these conceptual distinctions probably held true throughout most of human history, debates exist as to the current validity of these conceptualizations and the continued value of their use. The introduction of neonatal technologies and other factors in MDCs have led to further discussion and reconceptualization (Sowards 1997).

INFANT AND CHILD MORTALITY THROUGH HISTORY

Among historians such different academic spheres as demography, sociology of the family, and public health, there is an interesting debate concerning the timing and factors that have led to a decline in infant and childhood mortality. There are some interpretative similarities that may offer us a foundation with which to begin this discussion.

Historically, the size of the human population remained fairly constant until the first half of the eighteenth century. One hypothesis states that until then, although periods of low mortality were common, a population was still vulnerable to famines and epidemics, thus rendering population size stable. Before the Neolithic period (c. 9500 b.c.), it is estimated that infants had a 50 percent chance of surviving to adulthood (Stockwell and Groat 1984). It is thought that the first real decline in infant and child mortality came after the Neolithic revolution, when agriculture and animal domestication became more common, despite the increased spread of infectious diseases among now clustered and sedentary populations (Creighton-Zollar 1993). Within the framework of demographic transition theory, the staggering growth of the human population registered during the last 250 years is mainly explained by a combination of significant decrements in mortality and constant fertility rates (Kitagawa 1977; Stolnitz 1955).

Interestingly, some historians have found a link between the decline in infant mortality and the greater significance of children within the family organization. In the second half of the twentieth century, a discussion about the history of the family properly began with Philippe Ariès's L'Enfant en la Vie Familiale sous L'Ancient Régime (Van de Walle and Van de Walle 1990). Ariès suggests that the recognition of childhood as an independent stage of the life cycle occurred first during the sixteenth and seventeenth centuries. Other family historians have argued that infant mortality was so high that parents preferred not to "invest too much emotional capital in such ephemeral beings" (quoted by Van de Walle and Van de Walle 1990, p. 151).

For these authors, the British aristocracy's movement toward the increased domesticity of women redefined the role of the mother within the domestic economy. British aristocratic mothers now spent an increasing significant proportion of their time with their infants and children, forming strong parent–child bonds, which had previously been weak. These mothers were the first social group to conceive of a child as a treasure above and beyond the small utilitarian value children may have previously possessed (Van de Walle and Van de Walle 1990).

These authors suggest that from the seventeenth through the nineteenth centuries, the interest in children became increasingly mother-centered. Feelings of empowerment over the health and well-being of infants and children rose to such a level that women began demanding that the medical community and government pay more attention to children. Children thus passed from a "private" realm of familial responsibility to a "public" one, withe the community now expressing a vested interest in improving the health and well-being of children. These historical changes in the power structure of the family may have operated in conjunction with various other social and economic factors that led to the decreasing rate of infant and child mortality among Western developed countries (Van de Walle and Van de Walle 1990).

TWENTIETH-CENTURY MORTALITY TRENDS IN DEVELOPED COUNTRIES

Developed and developing countries are comprised of different population histories worthy of individual historical analysis, even though the only available and reliable documents that allow the reconstruction of demographic transitions are found in Europe. For some local Western European populations, studies have revealed that mortality levels decreased slightly between 1840 and 1860, but several scholars agree that the radical decline of mortality among the MDCs took place during the period 1880–1910 or even up to 1930 (Kitigawa 1977; Preston 1992; Stolnitz 1956). Although it is generally believed that this decrease in mortality levels was primarily due to a reduction of deaths during childhood, the causes of that change are still nonetheless a matter of controversy.

Epidemiologic transition theory's main argument is that the socioeconomic development achieved by those countries was fundamental in the reduction of mortality (Omran 1971). From this perspective, socioeconomic factors such as improved health and hygienic habits, diet and standards of living, were the main determinants of the transition from high to low mortality in Western nations. The sanitary revolution played an auxiliary role within this process of transition. Some other scholars suggest that although the improvement of economic conditions in Western nations was important, but this was more a permissive element than a precipitating factor. For them, the introduction of hygienic measures that helped to control communicable diseases was the leading cause of the decreased mortality rate (Preston 1992; Preston and Haines 1991; Spiegelman 1956; Stolnitz 1956).

According to Stolnitz (1956), in the mid-1800s reliable estimates for France and England show that real per capita income was rising rapidly but declines in mortality were still rather negligible. On the contrary, when mortality declined clearly after 1870, the West had already experienced a diverse set of economic circumstances. For instance, in France and Sweden the decrease in mortality appears to have occurred at a time when economic growth was slower than in earlier decades, while in Great Britain standards of living may have been slightly declining. Simultaneously, great changes occurred in the field of bacteriology, including the discoveries of Pasteur, and the large-scale public health programs were implemented. At the end of the nineteenth century in England, the main reason for the reduction in mortality levels was due to the lower prevalence of diseases resulting from these health innovations.

A study of the importance of different causes of death during the transition to a lower mortality rate shows interesting findings (Preston and Nelson 1974). Among 165 local populations (from 43 countries around the world) from the period 1861–1964, approximately 60 percent of the total decline in death rates was attributed to declining mortality from infectious diseases—25 percent due to influenza, pneumonia, and bronchitis; 10 percent due to respiratory tuberculosis; 10 percent due to diarrheal diseases; 15 percent due to other infectious and parasitic diseases; another 20 percent was attributed to a decrease in vascular diseases.

Many Western countries experienced a steady decline in mortality rates after 1930. Spiegelman (1956) suggests that during the period 1930–1950, these nations were experiencing a reduction in mortality at all stages of life. According to the author, from 1946 to 1954 most countries of Western Europe achieved a reduction of up to 33 percent in their IMRs. In Table 1 we reconstruct infant mortality trends for some selected countries during a period of forty-five years. In 1948, Sweden already had a relatively low IMR (23.2 per 1,000); by 1993 it had decreased to 4.8

Japan is a striking case. Japan, Australia, and New Zealand are included in Table 1 among the developed countries following the suggestion of the World Bank (1983), although they do not share the historical experience and culture of the Western countries included in the same group. The reduction of infant mortality has been explained by specific health measures first taken during the U.S. occupation after World War II and continued later by the Japanese government. During the 1930s, life expectancy at birth (life expectancy at birth represents the average number of years to be lived for a newborn, given a specific pattern of mortality in a country in a given moment) in Japan was similar to that in Western countries at the end of the nineteenth century. However, between 1948–1950 and 1951–1953, the increase in life expectation was about 5.5 years at age 0, 4 to 5 years at age 15, and 2 years at age 45. This was far and away the greatest average annual increase in demographic history (Stolnitz 1956). Table 1 shows a Japanese decrease in infant mortality from 61.7 per 1,000 in 1948 to 30.4 in 1960. This accelerated decrease continued to 13.1, 7.4, and 4.3 in 1970, 1980 and 1993, respectively, giving Japan the lowest IMR in the world.

In the United States, child mortality had been declining in the two decades before 1900 for the total population, although the decline was greater among whites than among blacks. It is difficult to estimate the magnitude of this differential because of data problems (Preston and Haines 1991). Preston and Haines find that between 1899 and 1900, 88 percent of deaths among those under age 15 were children aged 0 to 4 and 59 percent of these were infant deaths. Regarding the three main causes of infant mortality, 25 percent of deaths were due to gastrointestinal diseases, 20 percent to respiratory diseases (e.g., influenza, pneumonia, and bronchitis), and 27 percent to malformations and a wide array of causes, such as premature birth, debility and atrophy, inanition, and hydrocephalus. The main causes of child mortality (deaths below age five) were basically the same, but with a slightly different distribution: 21 percent were due to gastrointestinal diseases, 23 percent respiratory diseases, and 20 percent to malformations.

In comparing eighteen developed countries, Pampel and Pillai (1986) found the United States to have a relatively high IMR despite its leading economic position. Indeed, the United States has the highest IMR among developed countries. Table 1 shows that in 1948 the American IMR was half that of Japan. By 1993, however, that ratio had reversed; for every Japanese infant death there were 2 in the United States. Although this 2:1 does not exist between the United States and other developed countries, most of the countries shown in the table had higher IMR rates than the United States before World War II and have achieved significant declines since that time.

To briefly highlight some other developed countries, we take Spain as an example. According to Spiegelman (1956), acute infectious diseases were not totally under control there before World War II. This country began the forty-five-year period covered in Table 1 with an IMR of 70 in 1948. Infant mortality has continually decreased, reaching 6.7 in 1993.

Another case that may be worthy of highlighting is that of Germany, which is included in the group of Western countries because of its reunification and the general adoption of a market economy. Interestingly, in 1948 the Democratic Republic (East Germany) had a considerably higher IMR than the Federal Republic (West Germany), 94.0 versus 70.1. East Germany's rate was slightly higher in 1960, but in 1970 the trend reversed. In later years, both countries experienced equal rates. In 1993, unified Germany's IMR was 5.8, the third lowest of this selected group of developed countries.

TWENTIETH-CENTURY MORTALITY TRENDS IN EASTERN EUROPE

It is difficult to obtain comparable data for Eastern Europe; Table 1 shows only a few examples. In Romania in 1948 the IMR was particularly high. From 1948 to 1960 the IMR was almost halved, decreasing from 142.7 to 75.7. That pace of decrease could not be sustained, and in 1993 it

reached 23.3, the highest IMR among the Eastern European countries shown.

In the cases of Poland and Yugoslavia, both countries had a high IMR in the immediate postwar period (110.7 and 102.1, respectively). Despite its lower starting level, Yugoslavia consistently experienced higher rates thereafter: 87.7 in 1960, 55.5 in 1970, 32.8 in 1980, and 21.9 in 1993, whereas in Poland such rates have been 56.1, 33.2, 21.2, and 16.2 for the same years.

In 1948, Czechoslovakia had the lowest IMR among the Eastern bloc countries shown in Table 1, but it was still higher than that of Western countries. The decrease observed from 1948 to 1960 is very significant: from 83.5 to 23.5 in twelve years. This pace decelerated but continued to decline. By 1993, it had achieved an IMR similar to that of the United States: 8.5 deaths 1,000/live births.

TWENTIETH-CENTURY MORTALITY TRENDS IN DEVELOPING COUNTRIES

In the developing countries of Latin America, Asia, and Africa, reduction in mortality levels began around the beginning of the twentieth century. Scholars have found only fragmented information to reconstruct the demographic history of such countries. They can affirm, however, that the pace of declining mortality was slow until 1940 and accelerated after World War II (Kitagawa 1977; Meslé and Vallin 1996; Preston and Nelson 1974). According to Stolnitz (1956), the decline in mortality in developing countries during the postwar period, which was highly accelerated, has no parallels in the Western experience. The common explanation, found in numerous studies (Arriaga 1970; Kitigawa 1977; Preston 1976; Stolnitz 1956), is that access to medical innovations coming from developed countries and the implementation of specific programs of disease control were the main cause of the mortality reduction, especially in infant mortality.

Since the changes in those regions have not been uniform, currently there is a convergence of the mortality patterns in developing countries, whereas in the developed areas there is a marked divergence (Meslé and Vallin 1996). Indeed, trends of mortality just described for developed countries show variations, but among developing countries such heterogeneity is even greater. It would be impossible in this space to describe the mosaic of infant mortality experiences in developing countries; however, we can point out some major tendencies and outstanding cases during the forty-five years period covered by Table 1.

In order to understand the scope of differences among developing countries, it is important to consider a general characteristic of populations with high mortality: The younger the age, the larger the absolute change in mortality rates (Arriaga 1970). At the beginning of this transition toward a low mortality pattern, drastic changes are observed in infant and child mortality among different populations. For many developing countries, infant and child mortality trends throughout history cannot be known because there are no available data. Although this issue will be discussed below, usually the more socioeconomically disadvantaged countries are the ones for whom the data are lacking.

Latin America. During the 1930s Latin America had a high mortality pattern. For instance, a newborn baby had less than a 50 percent chance of living to the age of thirty; consequently, no country had a life expectancy more than forty years at birth (Arriaga 1970). The main decline was registered between 1940 and 1960. The largest decline could be found during the 1940s; there was a slower pace of decrease in the 1950s (Arriaga 1989; Camposortega Cruz 1989a).

Brazil and Mexico are the two most highly populated countries in Latin America, and they experienced similar processes of industrialization up until the 1970s. Brazil had a military government and Mexico did not; however, both have high economic inequality, possibly Brazil the more so. Unfortunately there are no data available for Brazil prior to the 1980s, but Table 1 shows that its IMR was higher than that of Mexico in both 1983 and 1993. Mexico's IMR decreased by 25 percent between 1948 and 1960; it continued to decline even in the midst of a drastic economic recession that began in the 1980s. By 1993, the persistently high rate had been reduced to 34.0 infant deaths per 1,000 live-births; the Brazilian IMR was 47.0 in that year.

Argentina provides an example of early industrialization as well as early demographic transition in the region. During the immediate postwar period, this country had an IMR similar to those of several Western countries (69.5 in 1948); considering this fact, the rate in 1993 (22.9) was relatively high, even higher than the IMR (22.0) of Venezuela, a country that had a rate 30 percent higher than Argentina in 1948.

El Salvador, Bolivia, and Peru, some of the poorest countries in the region, had high IMRs in 1948. For example, in El Salvador a tenth of all infants died in 1948, but this rate had decreased 25 percent by 1960 to 76.3 infant deaths per 1,000 live births. This pace of decline slowed in subsequent years, reaching 42.0 in 1980. However, by 1993 this indicator of general well-being had reversed, and the IMR had increased slightly to 44.0, perhaps because of severe economic conditions during the 1980s.

In Bolivia, if the data in Table 1 are correct, the decrease in infant mortality between 1948 and 1960 was spectacular: from 131.3 deaths to 74.3 deaths per 1,000 within twelve years. Information for the 1970s is unavailable. In 1983 the rate was again high (124.4), but ten years later it had droped to 75.0. These significant shifts give rise to concerns about the quality of the data. In any case, the IMR in 1993 was higher than the Salvadorian rate and similar only to the Peruvian one.

Bolivia and Peru have the highest IMRs among the selected countries (75.0 and 74.9, respectively). In this region, such a rate might be surpassed only by Haiti, for which information is not available. Interestingly, Peru had a decrease in its IMR from 1948 (109.0) to 1970 (64.6). But in 1983 the IMR was even higher that it had been twenty years earlier: In 1983 it was 98.6, up from 92.1 in 1960. Although the IMR had decreased to 74.9 in 1993, Peru still has one of the highest rates of infant mortality in all of Latin America.

In contrast, Cuba and Costa Rica have the lowest and the second lowest IMRs, respectively. The public health systems in both countries have good reputations throughout the region. Costa Rica shows a steady decrease from 1948, when its IMR was 90.4. By 1970, it had declined 30 percent, reaching 61.5. Contradicting Arriaga (1989) and Camposortega Cruz (1989a), this country achieved a steeper decrease in infant mortality during the 1970s: In 1970 the rate was 61.5 and in 1980, 20.2, showing a decrease of 33 percent. In 1993, the Costa Rican IMR was 13.7.

The Cuban IMR was already the lowest in the region in 1948 (40.1) among those countries shown in Table 1. Up until 1970 there was only a small and unsteady decrease to (38.7). During the 1970s and afterward, the country achieved significant decreases: The rate reached 19.6 in 1980 and 9.4 in 1993. The decrease in infant mortality has been one of the main goals and achievements of the Cuban health system, despite the country's unfavorable economic situation. Cuba now has the lowest IMR in Latin America, being only slightly higher than that of Czechoslovakia and the United States.

Asia. Infant mortality rates in Asia are quite diverse. As regards of India, Jain and Visaira (1988) argue that even though several studies suggest that the infant mortality rate in 1920 was 240 per 1,000 live births, this might have been understated, and the IMR could have been between 200 and 225 during the 1940s. This finding is in contrast to the estimations of Arriaga (1989) and other researchers. Using different sources of information and techniques, Jain and Visaira construct a trend of infant mortality with higher rates than those shown in Table 1, with the rates coinciding only since the 1970s. Jain and Visaira (1988) and Arriaga (1989) found that as a result of implementing a specific program of hygiene among mothers, infant mortality did decline after 1970. Between that year and 1993, the IMR decreased from 122.0 to 74.0.

Although there is no complete direct information for China that allows us to construct a complete infant mortality trend since the postwar period, some estimations are provided in a study by Banister (1986). From the mid-1960s to 1977–1978, the author and her colleagues found a constant decrease in infant mortality (from 70 to 40 deaths), although the trend was later reversed. Their calculation for 1981–1982 was 61 deaths per 1,000 live births, which does not match the information from the United Nations: 39.3 in 1983 and 44.0 a decade later (Table 1). Banister and her team examined the possibility of measurement errors in order to discern whether the estimation was fatally biased and changed the direction of the trend. They confirmed that such was not the case (Banister 1986). There may have been an over- or underestimation between both calculations, but either way, the direction of the tendency toward an increasing IMR seem to be correct.

According to Banister (1986), there are three reasons that may explain this inversion. First, in 1978 a general economic reform began and the availability of preventive and curative health services was somehow affected, which may have been detrimental to the survival of some infants. Second, at the same time China's severe family-planning reached rural couples. In 1981, 53 percent of all births corresponded to second and higher-order births; therefore, if mothers of these unauthorized babies lacked prenatal care, not surprisingly their infants would experience higher risk of early death. Additionally, the provision of medical care had been tilted intentionally toward the only child, which may have implied the deterioration of the health care available for children in multichild families. Third, a decrease in the proportion of breast-feeding combined with the use of contaminated water used to mix formula might have adversely affected infant survival. Although these researchers did not have precise information on the proportion of breast-feeding, based on fragmentary data that proportion seems to have decreased (in 1982 22 percent of the babies in urban areas and 60 percent in rural settlements were breast-fed). A possible fourth factor is infanticide, but the distribution of infant deaths by sex did not show a bias toward more female deaths (it is said that there exists in China a preference for male babies). Hence, if infanticide played an important role in increasing the IMR, Banister's evidence suggest either that the incidence of female infanticide is not statistically significant or that respondents reported neither the birth nor the death of female victims of infanticide.

Sri Lanka is a rare case of a developing country's retaining records from nearly the beginning of the twentieth century. Stolnitz (1956) refers to Sri Lanka (formerly Ceylon) as one example of rapidly declining infant mortality; more recently, Meegama (1986) offers an overview of mortality trends since the beginning of the twentieth century. According to this latter author, during the second half of the nineteenth century, the IMR started to decrease because of a reduction in the number of famines and the control of cholera epidemics.

During 1911–1915, the IMR was 201 deaths per 1,000 live births, fifteen years later it had decreased to 175 deaths. In 1945, at the end of the World War II, the IMR was recorded at 140 infant deaths per 1,000 live births. Prior to the war, there was a decrease in the mortality rate for both sexes and all age groups in both endemic and nonendemic malarial zones. One of the main reasons for this decline was the expansion of maternity and child welfare services, which decreased both infant and maternal mortality (Meegama 1986). It might have been the combination of that preexisting decline and the introduction of a wider range of medical measures that caused a steady decrease in mortality after the war. As shown in Table 1, in 1948 Sri Lanka had an IMR of 92.1. This was halved by the beginning of the 1970s (45.1) and was again reduced to a third of the 1970 rat by 1993 (16.5).

South Korea and Singapore are two cases of recent industrialization and fast decline in mortality patterns. Information is not available to reconstruct the South Korean mortality history; however, what can be seen in Table 1 is a decrease of more than 50 percent from 1983 to 1993. In 1983 the country had an IMR of 29.7, which had declined to 12.3 ten years later.

Singapore is another impressive case of accelerated reduction in infant mortality. In 1948, it had a relatively low IMR among Asian countries (80.8), but its rate was greater than that of Japan. By 1960, the rate had been reduced by more than 50 percent, reaching 34.8. The trend continued, and by 1993 Singapore's IMR was 4.7, similar to the Swedish and very close to the Japanese rate.

Africa. Mortality in Africa is the highest of any major region in the world and historically has had the least reliable information, with certain exceptions. Through the application of special techniques, mortality trends have been reconstructed. The pattern of mortality in Africa is characteristic of populations with a high mortality schedule: It is dominated by mortality among children under five years of age, and infectious, parasitic, and respiratory diseases are the major causes of death. Concerning infant mortality in Africa, we examine some cases from this region displayed in Table 1.

Kenya lacks a usable vital registration system, but the application of indirect techniques has helped in reconstructing mortality trends there. In 1948, when the country began its transition toward lower mortality, it might have had an IMR of 184 per 1,000 live births, according to some estimates (Ewbank et al. 1986). In 1970 that rate had dropped to 81.2, but in the following decades the pace of change was considerably slower; in 1993 the rate was 71.0. For an African country, this is a low mortality pattern, but it is comparable to some Asian countries and to the poorest in Latin America.

Mauritius has a mortality pattern that is gradually approaching that of Western Europe (United Nations 1984). It has the lowest IMR of African countries shown in Table 1. In the period immediately after the World War II, Mauritius had an IMR of 186.2 deaths per 1,000 live births; this decreased 62 percent in twelve years to 70.9 in 1960. That steady rate of decrease was maintained, with the IMR falling to 19.9 by 1993.

The South African case presented some different data analysis challenges because prior to the 1980s information on mortality was not available for the country as a whole but rather by ethnic group, showing the effect of apartheid on vital statistics but also allowing us to see ethnic differentials in the levels of infant mortality. As shown in Table 1, the white population had always been better off than their fellow South Africans; the IMR of the black population held relatively constant until 1970 and the IMR of the Asian population was located between that of the white and black populaces. The ratio of deaths between ethnic groups illustrates social differences: In 1948 there were approximately 4 black infant deaths for each white one and 1.7 black infant deaths for each Asian one. The ratio for Asian to white deaths was approximately 2-to-1 for babies under the age of one year. In 1970 the gap between blacks and the other races widened: There were 6.1 black infant deaths for each white death, 3.6 black infant deaths for each Asian death, and 1.7 Asian infant deaths for each white infant death. By the 1980s the information from the Demographic Yearbook was no longer classified by ethnic group; it shows that the general IMR decreased between 1983 and 1993, from 83.3 to 53.0, but it does not show the ethnic differential during those years.

DIFFERENTIALS IN INFANT AND CHILD MORTALITY

Numerous studies have documented and examined various differences in infant and child mortality among different groups along various socioeconomic and ecological spectrums. For most of the earliest periods under investigation, ecological and geographic factors seemed to account for most of the observed differences in infant and child mortality rates (Preston and Haines 1991). Current research on this topic is guided by a theoretical framework put forth by Mosley and Chen (1984). Their original intent was to outline an analytical framework for studying child survival in developing countries, but many researchers have found this framework to be extremely suitable for studying infant and mortality differentials within various contexts.

Mosley and Chen (1984) propose that all socioeconomic determinants, or exogenous factors, of child mortality operate through a set of biological mechanisms that they call "proximate determinants" or "intermediate variables." That is, social and economic factors influence infant and child mortality through a set of biological factors that an infant is born with. Many studies have identified numerous variables and factors that influence both the biological viability of an infant at birth and the health and well-being of the infant ex utero, thereby increasing or decreasing the probability that an infant will die before its first birthday or that a child will die before its fifth. These social and economic factors range from individual-level characteristics to household- and community-level factors.

Research has identified three main infant birth outcomes that exert major influences on infant survival chances and through which many exogenous factors operate: infant birthweight, gestational age, and the interaction of the two, sometimes referred to as intrauterine growth retardation (IUGR). Infants weighing less than 2,500 grams at birth (low birthweight, LBW) and infants of short gestation (less than thirty-seven weeks' gestation=premature) have been consistently found to be at dramatically higher risk of infant mortality (Hummer et al. 1999). Interactions between these two outcomes have also been found to affect infant mortality rates significantly (Frisbie et al. 1996). These relationships hold for both MDCs and LDCs (UN 1984).

Of the set of proximate and exogenous determinants originally outlined by Mosley and Chen (1984), most have been identified as having significant effects on infant mortality. Smoking, alcohol, and illicit drug use have all been shown to adversely affect infant and child mortality (Chomitz et al. 1995; Petitti and Coleman 1990). Maternal weight gain has been found to be negatively correlated with infant mortality risk; that is, the more weight a woman gains during her pregnancy, the lower the infant mortality risk (Hummer et al. 1999). The birth order of the infant is associated with differential infant mortality risks, while shorter interpregnancy intervals are linked to higher mortality risks (Kallan 1993; Miller 1991). Male infants have always had higher mortality risks (Rumbaut and Weeks 1996). The role of prenatal care has been debated in recent times, but most researchers agree that adequate prenatal care produces beneficial effects for the health and well-being of mothers, infants, and ultimately young children (Fiscella 1995). The effect on infant mortality risk of maternal age had previously been thought to be constantly negative in direction, but more recent analyses indicate that many interaction effects may work specifically through maternal age, making the conceptual picture much more complex (Geronimus 1987).

Exogenous factors have also been investigated and analyzed. One factor that has captured the continued interest of researchers is the role that race and ethnicity play in explaining infant and child mortality differentials. In both MDCs and LDCs consisting of multiracial and/or multiethnic populations, racial and ethnic infant and child mortality differentials often exist. For example, in 1996 the black population in the United States experienced an IMR of 14.1 while the white population experienced an IMR of 6.1 (MacDorman and Atkinson 1998). In 1990, the indigenous population of Mexico experienced a higher risk of infant mortality 1.5 times higher than the nonindigenous population (Fernández-Ham 1993). Although the role that race and ethnicity may play in IMR differentials is complex, there is agreement that such differences are based on social, economic and historical inequalities (Cooper and David 1986; David 1990). In LDCs those historical inequalities are also related to place of residence, which in most of these countries imply a more disadvantaged position for rural populations (Camposortega Cruz 1989b).

More direct socioeconomic indicators are currently being tested to identify their effects on mortality risks. Increases in maternal education have consistently been shown to decrease both infant and child mortality risk at both the individual and national levels (Hummer et al. 1992; Pampel and Pillai 1986). Marital status has also been linked to differential infant mortality risks; unmarried mothers have a higher risk for infant mortality. This may reflect the fact that single mothers often must bear the costs of both child-rearing and dayto-day living expenses with limited assistance (Frisbie et al. 1997). Studies have also identified the important role that income plays in influencing infant and child mortality risk (Geronimus and Korenman 1988). Cramer (1995) finds evidence of this association even when controlling for a wide array of other sociodemographic risk factors. Many others have explored the effects of various public health assistance and intervention programs, either by governmental or nonprofit agencies. In both MDCs and LDCs, some evidence exists to support the creation and implementation of such intervention programs, such as the Women, Infants and Children (WIC) program in the United States and governmental health programs in Guatemala (Moss and Carver 1998; United Nations 1984). According to Frenk and associates (1989), in certain developed countries there may exist two populations divided by their differential infant and child mortality risk. These researchers point to existing differences between insured and uninsured populations in Mexico as an example of the divergent character of national infant mortality profiles in both high- and middle-income countries (Frenk et al. 1989).

With the development of new theoretical frameworks and statistical methodologies, infant and child mortality research has traveled down new paths. One interesting development has been analyses based upon characteristics of both the individual and the neighborhood or community in which individuals live. Determinants of infant and child mortality are hypothesized to emerge from conflation of personal, household, and neighborhood factors. Collins and David (1997) find that violent neighborhoods may adversely effect an expectant mother's pregnancy outcome, in spite of the measured personal characteristics of the mother. Guest (1998) finds that the neighborhood unemployment level has an important effect on infant mortality beyond effects due to other socioeconomic indicators and race.

PROSPECTS

The search for solutions to reduce levels of infant and child mortality will require the joint participation of multiple actors: governments, international agencies, nongovernmental organizations, and individuals within the scientific and policy communities. Predicting future trends in infant and child mortality remains a rather difficult task given the complexity of this phenomenon. However, scientific knowledge gathered at both the aggregate and individual levels should make reduction of the significant gap between the developed and developing world possible. The multiple differences that exist in mortality patterns within groups in both developed and developing countries are illustrated by the current age structure of mortality. In 1990-1995, life expectancy at birth was 74.4 years in developed countries and 62.3 years in the less developed world. This indicator is consistent with the corresponding IMRs: 10 versus 70 infant deaths per 1,000 live births for MDCs and LDCs, respectively.

It is noteworthy that infant and child mortality rates fell so drastically in the twentieth century. But even so, there remains the need to determine the next course of action in order to combat the mortality differentials between less and more developed nations and among various socioeconomic groups within nations. Continued successes will depend on identifying problem areas and creating innovative solutions to these health, and ultimately, social problems.

references

Arriaga, E. E. 1970 Mortality Decline and its DemographicEffects in Latin America (Population Monograph Series No. 6). Berkeley: University of California Press.

——1989 "Changing Trends in Mortality Decline during the Last Decades." In L. Ruzicka, G. Wunsch, and P. Kane, eds., Differential Mortality: Methodological Issues and Biosocial Factors. Oxford: Clarendon Press.

Banister, J. 1986 China: Recent Trends in Health andMortality. Washington, D.C.: Center for International Research, U.S. Bureau of the Census.

Bogue, D. J. 1969 Principles of Demography. New York: John Wiley and Sons.

Bourgeois-Pichat, J. 1951 "La Mesure de la Mortalité Infantile. I. Princepes et Methodes." Population 2:1–17.

Bouvier, L. F., and J. Van der Tak 1976 "Infant Mortality: Progress and Problems." Population Bulletin 31:3–33.

Camposortega Cruz, S. 1989a "La evolución de la mortalidad en México, 1940–1980." Estudios Demográficosy Urbanos 4(2):229–264.

——1989b "Mortalidad en México. Algunas Consideraciones Sobre Los Diferenciales Urbano-Rurales." Estudios Demográficos y Urbanos 4(3):573–593.

Chomitz, V. R., L. W. Cheung, and E. Lieberman 1995 "The Role of Lifestyle in Preventing Low Birth Weight." Future of Children 5(1):121–138.

Coale, A. J. 1965 "Birth Rates, Death Rates, and Rates of Growth in Human Population." In Mindel C. Sheps and Jeanne C. Ridley, eds., Public Health and Population Change. Pittsburgh, Pa.: University of Pittsburgh Press.

Collins, J. W., and R. J. David 1997 "Urban Violence and African-American Pregnancy Outcomes." Ethnicityand Disease 7:184–190.

Cooper, R., and R. David 1986 "The Biological Concept of Race and its Application to Public Health and Epidemiology." Journal of Health Politics, Policy andLaw 11:97–116.

Cramer, J. C. 1995 "Racial and Ethnic Differences in Birthweight: The Role of Income and Family Assistance." Demography 27:413–430.

Creighton-Zollar, A. 1993 "Infant Mortality by Socioeconomic Status and Race: Richmond, VA 1979–1981." Sociological Spectrum 10:133–142.

David, R. 1990 "Race, Birthweight, and Mortality Rates [editorial]." Journal of Pediatrics 116:101–102.

Ewbank, D., R. Henin, and J. Kekovole 1986 "An Integration of Demographic and Epidemiologic Research on Mortality in Kenya." In United Nations, Determinants of Mortality Change and Differentials in Developing Countries (Population Studies No. 94). New York: United Nations.

Fernández-Ham, P. 1993 "La Mortalidad Infantil en la Población Indígena: Atrasos y Contrastes." Demos 6:12–13.

Fiscella, K. 1995 "Does Prenatal Care Improve Birth Outcomes? A Critical Review." Obstetrics & Gynecology 85(3):468–479.

Frenk, J., J. Bobadilla, J. Sepulveda, and M. Lopez 1989 "Health Transition in Middle-Income Countries: New Challenges for Health Care." Health Policy and Planning 4(1):29–39.

Frisbie, W. P., M. Biegler, P. deTurk, D. Forbes, and S. G. Pullum 1997 "Determinants of Intrauterine Growth Retardation and Other Compromised Birth Outcomes: A Comparison of Mexican Americans, African Americans, and Non-Hispanic Whites." American Journal of Public Health 87(12):1977–1983.

Frisbie, W. P., D. Forbes, and S. G. Pullum 1996 "Compromised Birth Outcomes and Infant Mortality Among Racial and Ethnic Groups." Demography 33(4):469–481.

Geronimus, A. T. 1987 "On Teenage Childbearing and Neonatal Mortality in the United States." Populationand Development Review 13(2):245–279.

——, and S. D. Korenman 1988 "Comment on Pampel and Pillai's 'Patterns and Determinants of Infant Mortality in Developed Nations, 1950–1975."' Demography 25(1):155–158.

Guest, A. M. 1998 "The Ecology of Race and Socioeconomic Distress: Infant and Working-Age Mortality in Chicago." Demography 35(1):23–34.

Hill, K. H. 1991 "Approaches to the Measurement of Childhood Mortality: A Comparative Approach." Population Index 57(3):368–382.

Hummer, R. A., M. Biegler, P. deTurk, D. Forbes, W. P. Frisbie, Y. Hong, and S. G. Pullum 1999 "Race/Ethnicity, Nativity and Infant Mortality." Social Forces 77(3):1083–1118.

Hummer, R. A., I. A. Eberstein, and C. B. Nam 1992 "Infant Mortality Differentials Among Hispanic Groups in Florida." Social Forces 70(4):1055–1075.

Jain, A. K., and P. Visaira (eds.) 1988 Infant Mortality inIndia. Differentials and Determinants. New Delhi: Sage Publications.

Kallan, J. E. 1993 "Race, Intervening Variables, and Two Forms of Low Birthweight." Demography 30:489–506.

Kitagawa, E. M. 1977 "On Mortality." Demography 14(4):381–389.

MacDorman, M. F., and J. O. Atkinson 1998 "Infant Mortality Statistics from the 1995 Period Linked Birth/Infant Death Data Set." Monthly Vital StatisticsReport 46(12s):1–24.

Meegama, S. A. 1986 "The Mortality Transition in Sri Lanka." In United Nations, Determinants of MortalityChange and Differentials in Developing Countries (Population Studies No. 94). New York: United Nations.

Meslé, F., and J. Vallin 1996 Mortality in the World: Trendsand Prospects. Paris: Centré Francais sur la Population et le Développement.

Miller, J. E. 1991 "Birth Intervals and Perinatal Health: An Investigation of Three Hypothesis." Family Planning Perspectives 23:62–70.

Mosley, W. H., and L. C. Chen 1984 "An Analytic Framework for the Study of Child Survival in Developing Countries." Population Development and Review 10(supplement):25–48.

Moss, N. E., and K. Carver 1998 "The Effect of WIC and Medicaid on Infant Mortality in the United States." American Journal of Public Health 88(9):1354–1361.

Omran, A. R. 1971 "The Epidemiologic Transition. A Theory of Population Change. Part 1." The MilbankFund Quarterly 49(4):509–538.

Omran, A. R. 1971 "Epidemiologic Transition: A Theory of the Epidemiology of Population Change." Milbank Memorial Fund Quarterly 49(4):509–538.

Omran, A. R. 1977 "Epidemiologic Transition in the U.S.: The Health Factor in Population Change." Population Bulletin 32(2):3–42.

Palloni, A. 1998 "A Review of Infant Mortality Trends in Selected Underdeveloped Countries: Some New Estimates." Population Studies 35(1):100–119.

Pampel, F. C., and V. K. Pillai 1986, "Patterns and Determinants of Infant Mortality in Developed Nations, 1950–1975." Demography 23(4):525–542.

Petitti, D. B., and C. Coleman 1990 "Cocaine Use and the Risk of Low Birthweight." American Journal ofPublic Health 80:25–27.

Preston, S. H. 1976 Mortality Patterns in National Populations. New York: Academic Press.

——1992 "Infant and Child Mortality." In E. F. Borgatta and M. L. Borgatta, eds., Encyclopedia ofSociology, vol. 2. New York: Macmillan.

——, and M. R. Haines 1991 Fatal Years: Child Mortality in Late Nineteenth-Century America. Princeton, N.J.: Princeton University Press.

——, and V. E. Nelson 1974 "Structure and Change in Causes of Death: An International Summary." Population Studies 28(1):19–51.

Rumbaut, R. G., and J. R. Weeks 1996 "Unraveling a Public Health Enigma: Why Do Immigrants Experience Superior Perinatal Health Outcomes." Researchin the Sociology of Health Care 13B:337–391.

Shryock, Henry S., and J. Siegel 1976 The Methods andMaterials of Demography. Washington, D.C.: U.S. Government Printing Office.

Sowards, K. A. 1997 "Premature Birth and the Changing Composition of Newborn Infectious Disease Mortality: Reconsidering 'Exogenous' Mortality." Demography 34(3):399–409.

Spiegelman, M. 1956 "Recent Trends and Determinants of Mortality in Highly Developed Countries." In F. G. Boudreau and C. V. Kiser, eds., Trends andDifferentials in Mortality. New York: Milbank Memorial Fund, pp. 26–34.

Stockwell, E. G., and H. T. Groat 1984 World Population:An Introduction to Demography. New York: Franklin Watts.

Stolnitz, G. J. 1955 "A Century of International Mortality Trends: I." Population Studies 9(1):24–55.

——1956 "Comparison Between Some Recent Mortality Trends in Underdeveloped Areas and Historical Trends in the West." In F. G. Boudreau and C. V. Kiser, eds., Trends and Differentials in Mortality. New York: Milbank Memorial Fund, pp. 51–60.

United Nations 1979 Demographic Yearbook. HistoricalSupplement. New York: United Nations.

——1982 Demographic Yearbook. New York: United Nations.

——1984 "The Influence of Developmental-Related Change on Infant and Child Mortality in Africa." In United Nations, Mortality and Health Policy: Proceedings of the Expert Group on Mortality and Health Policy,Rome, 30 May to June 1983. New York: United Nations.

——1987 Demographic Yearbook. New York: United Nations.

——1990 Step-by-Step Guide to the Estimation of ChildMortality. New York: United Nations.

——1996 Demographic Yearbook. New York: United Nations.

Van de Walle, E., and F. Van de Walle 1990 "The Private and the Public Child." In J. Caldwell, S. Findley, P. Caldwell, G Santow, W. Cosford, J. Braid, and D. Broers-Freeman, eds., What We Know About HealthTransition: The Cultural, Social and Behavioural Determinants of Health, Vol. 1. Australia: The Australian National University, pp. 150–164.

World Bank 1983 World Development Report. New York: Oxford University Press.

Georgina Rojas-GarcÍa
Samuel EchevarrÍa