Executive
Summary
The purpose ofthis report is to examine primarily the epidemiologic evidence reporting the effects of the method of infant feeding on infant health and to determine recent trends in method of infant feeding as well as factors associated with the choice and duration of feeding method.
The report is divided into two broad sections: one dealing with the literature relevant to the United States (Domestic Review), the other with the lit-erature relevant to developing countries (Interna-tional Review). In this regard, the report does not address many other important areas of research including recent studies of immunologic, biochem-ical, growth-promoting, and nutrient composition of breast milk and formulas. Further, it concen-trates on the post- 1970 scientific literature.
In the subsections dealing with trends in infant-feeding practices, we describe the frequency, dura-tion, and rate of change in the United States and developing countries. Population subgroups whose trends of infant-feeding practices differed from overall trends were also noted. By examining fac-tors associated with infant-feeding choices, we hope to determine those factors potentially amenable to change by infant nutrition programs. Also, better understanding of these factors assists in the sorting out of actual health effects from spurious associa-tions due to relationship between these factors and risk of illness. Finally, in the sections dealing with infant health and growth, the primary goals were to determine those areas in which the relationships with infant-feeding practices and health are rela-tively well established and to determine the areas needing further research.
The conclusions stated here are based largely on studies with serious design limitations, especially if randomized clinical trials are held as the standard. The Task Force believed, however, that it was important to reach conclusions, tempered as they are by the quality of the evidence, rather than state that cause-effect relationships were not conclu-sively demonstrated. Relative criteria used to eval-uate individual studies included (1) the strength of the association or effect; (2) whether there was adjustment for important factors related to infant-feeding decision making or practice and health; and (3) the likely impact of uncontrolled biases on the crude effect or association. The representativeness
of the population studied was heavily considered during examination of trend and infant-feeding choice and duration data. In the reviews of health effects and infant-feeding method, studies were considered more strongly if they showed a gradation of effect as feeding method varied from exclusive breast-feeding to exclusive bottle-feeding.
DOMESTIC REVIEW Trends
Starting from high levels of practice in the 1940s, breast-feeding declined steadily to low levels in the early 1970s and then began an upward trend that apparently has continued to the present. Although in the 1940s breast-feeding was more common among disadvantaged women (black, poor, and un-educated), the subsequent decline was more rapid among this group. Thus, by the early 1970s, disad-vantaged women were more likely to bottle-feed than breast-feed and, among the breast-feeding women, disadvantaged women breast-fed for a shorter average duration.
Conversely, the increase that has occurred since the early 1970s has not been as pronounced among the disadvantaged. Thus, recent information mdi-cates that black and poorly educated women are less likely to breast-feed than are white women; within racial groups, poorly educated women are less likely to breast-feed than are highly educated women.
Factors Associated with Method of Infant Feeding
The factors associated with the decision to breast-feed and with early termination are only partially understood. The interplay of social, bio-logic, and attitudinal factors makes the design of well-controlled studies with effective question-naires especially difficult. Thus, most of the avail-able studies are cross-sectional surveys that con-centrate on demographic variables such as maternal age, parity, and income.
are also more likely to breast-feed. Studies regard-ing maternal employment have produced conflict-ing findings as to its effect on length of breast-feeding and early use of supplemental bottles.
The social taboo against breast exposure in public is stronger in women who use bottle-feeding than in women who breast-feed. Other studies of attitu-dinal factors have shown that breast-feeding women believe that breast-feeding is superior to bottle-feeding and that these women have husbands who are more in favor of breast-feeding. Conversely, women who use bottle-feeding are more likely to believe that bottle-feeding is more convenient and provides more freedom than breast-feeding. Most women decide on the method of feeding before delivery. Relatively few women list physicians as having an important role in the choice as opposed to nurses and midwives. Insufficient milk is fre-quently stated as the reason given for termination of breast-feeding.
Little has been done to examine the role of mar-keting practices for infant formula as a factor in infant-feeding decisions. However, one recently re-ported randomized clinical trial found that receipt of a sample package of infant formula on discharge from the hospital was associated with lower rates of breast-feeding at 1 month in women who were less well educated or who became ill postpartum.
Future research needs to focus on careful design of prospective studies that will rigorously test spe-cific hypotheses including the effects of well-de-fined interventions. Nonexperimental (observa-tional) studies are needed to estimate the independ-ent effect ofvarious factors on infant-feeding choice and duration through the use of careful study design and multivariate statistical techniques.
Health
Our review of the scientific literature provides some evidence of a positive association between the health of an infant born in the last decade in the United States and breast-feeding, but this evidence is not entirely convincing. If there are health bene-fits associated with breast-feeding in populations with good sanitation, nutrition, and medical care, the benefits are apparently modest. In middle and upper class populations in developed countries where rates of serious illness are already low, it would be difficult to unequivocally demonstrate effects of breast-feeding on health by the observa-tional methods most frequently used. In part, the lack of firm evidence for an association between infant health and feeding practice is due to poor study design; in only a few studies were confounding factors controlled.
In certain US subpopulations in which good
san-itation and nutrition are not universal and mothers and infants may not receive optimal medical care, infants have higher morbidity and mortality rates than those seen among the general population. Any health benefits associated with breast-feeding would more likely appear in these populations than in groups with higher socioeconomic status. Such is the case with the few studies on native North American populations, which suggest that substan-tial health benefits accrue to breast-fed infants, especially in terms of protection from gastroenter-itis. Infant-feeding practices and health have rarely been studied in poor urban populations; such stud-ies are clearly needed.
It is unlikely that a large proportion of postneo-natal deaths in the United States could be averted by universal breast-feeding. However, the lack of good epidemiologic studies in this area suggests that studies of specific causes of death and infant-feed-ing methods are warranted.
Of the infectious morbidities reviewed, the evi-dence for a protective effect attributable to breast-feeding appears strongest for mild gastrointestinal illness, for which an effect was consistently seen. Rates of hospitalization for gastroenteritis were not found to be significantly different between breast-fed and bottle-fed infants in two studies that con-trolled for several other factors. Others, however, have found that breast-feeding is associated with lower rates of hospitalization when rates for all illnesses are combined. Recently reported studies that controlled for several other factors suggest that previously reported reduction in rates of respiratory illness among breast-fed infants may be attributa-ble to socioeconomic and other differences (such as parental smoking) between feeding groups rather than the feeding method. Recent reports document the protective effect of breast-feeding against otitis media.
Evidence regarding infant feeding and allergies is inconsistent. Some reports suggest lower rates of asthma among breast-fed infants; others find no effect.
Most infants who are exclusively breast-fed show normal growth until 4 to 6 months of age. After that time, an increasing proportion require supple-mentation in order to maintain growth. The rela-tionship between infant feeding and subsequent obesity remains unresolved. There is some indica-tion that breast-fed infants are less likely to become obese, but confounding factors such as time and type of solids introduced have rarely been ad-dressed.
is little evidence, however, to determine whether this is an independent effect of breast-feeding on intellectual function.
Problems with Human Milk and Infant Formulas Breast milk is clearly a superb food for infants. Some trace minerals such as zinc and iron are considered to be more bioavailable from human milk than from formulas.
It is also clear, however, that deficiencies of vi-tamin K, vitamin D, and iron may develop in nor-mal breast-fed infants. Other nutrient deficiencies may occur as a consequence of special conditions existing in the infant, the mother, or both. Also, breast-feeding can be absolutely or relatively con-traindicated in unusual circumstances such as gal-actosemia or serious maternal emotional illness.
An area of necessary concern is the presence of chemical pollutants in human milk, especially or-ganochlorine pesticides and industrial chemicals such as polychlorinated biphenyls (PCBs). Assess-ment of risk and safety levels is difficult and sub-stantially more research is clearly needed in this area.
Similarly, almost any drug present in the moth-er’s circulation will also be detectable in her milk. An infant can become sensitized to drugs ingested via breast-feeding although this rarely warrants cessation of breast-feeding.
There have been relatively few problems with commercial formulas over the years. Several in-stances of infant illness from formulas were due to lack of fortification with what later was determined to be necessary nutrients, eg, vitamin K, linoleic acid. There have also been occasional errors in the manufacturing process (eg, chloride-deficient for-mula that led to metabolic alkalosis in some in-fants).
More research is needed on the influence of milk composition on optimal development during in-fancy and the effects of the total infant diet pattern on later development of obesity and hypertension. Finally, information is also needed concerning op-timal levels of trace minerals in infant formulas.
INTERNATIONAL REVIEW Trends
Few countries have information on the rate of initiation and duration of breast-feeding based on internally comparable national data collected over time. Seven countries with such information are: Taiwan, Thailand, Malaysia, Korea, Singapore, Mexico, and Panama. Common findings in these countries include: (1) a decline in the rate of
initi-ation and the duration of breast-feeding; this de-dine was sharp to moderate in six of these coun-tries; (2) the decline in the rate of breast-feeding occurred more rapidly in urban than rural areas; and (3) the greatest decline appeared among the most modern (nontraditional) subpopulations.
The cross-sectional data already available for many countries are not adequate to measure trends in breast-feeding. However, within a few years, ongoing national studies will permit examination of trends from many other developing nations.
Factors Associated with Method of Infant Feeding
The factors associated with infant-feeding choice and duration vary within and between communities in the less developed world. Moreover, the factors associated with infant-feeding choice at birth may differ from those related to duration of breast-feeding as well as from those associated with time of introduction of liquid breast milk substitutes and solids.
Factors associated with breast-feeding at birth included having a “normal” delivery, birth weight 2.5 kg, and parity 4. Variables frequently asso-ciated with a longer duration of breast-feeding in-cluded having breast-fed previous children, rural residence, low income, and low level of education. Other associations tend to be specific for national or ethnic groups.
Studies of the duration of breast-feeding do not analyze separately the factors associated with the introduction ofliquid breast milk substitutes, intro-duction of solids, or termination of breast-feeding. In general, shorter durations of full or partial breast-feeding appear to be associated with high socioeconomic status, maternal education, urban residence, use of oral contraceptives, and maternal employment away from home. The most common reasons given for supplementation are breast milk insufficiency and maternal employment. The most common reasons for stopping breast-feeding in-dude milk insufficiency, pregnancy, maternal or infant illness, and the infant was too old and re-jected the breast.
Little information exists concerning the influ-ence of marketing of infant formula on infant-feeding practices. The difficulty in framing ques-tions to examine the impact of advertising and the difficulty in separating the effects of high socioec-onomic status and urbanization from exposure to advertising have hindered the study of these factors.
Mortality and Infectious Morbidity
studies of overall mortality that used statistical techniques to adjust for factors other than feeding practice that might affect infant mortality. These studies still found a significant beneficial effect of breast-feeding on infant survival after the factors were controlled.
The majority of studies of infectious illness dem-onstrated lower rates of gastroenteritis among breast-fed infants compared with bottle-fed infants. This protective association persists in those studies that statistically adjusted for potential covariates. The findings with regard to respiratory infection were less clear; several studies found little or no effect.
In one randomized clinical trial of high-risk neo-nates, infants given breast milk had lower rates of infection than infants receiving formula. Because all bacterial cultures of the formula used were neg-ative, a direct protective effect of breast milk (rather than lack of exposure to contaminated for-mula) was suggested.
Results of investigations with relatively strong study designs suggest that protection operates spe-cifically against Shigella, Salmonella, and Vibrio cholerae. Studies of other organisms were too pre-liminary, failed to account for other important fac-tors, or showed no effect.
Interpregnancy Intervals
Breast-feeding is associated with increased length of the postpartum anovulatory period. This has the effect of lengthening the interval between pregnancies and may be of major importance in populations in which other forms of contraception are not widely available or used. Short pregnancy intervals have been widely held to be detrimental to the health of the second born of two infants; however, most studies of this question have been poorly controlled. The well-controlled studies from the developed world show only small, if any, effects of pregnancy intervals on infant mortality. One well-controlled study from Ecuador suggests that a sizable effect on infant mortality does exist in the developing populations in which overall infant mor-tality is high and the interval between pregnancies is very short (1 year). This effect is evident for both the firstborn and second born of the pair of interest.
Because there are presently high rates of breast-feeding in most of the populations in which preg-nancy interval could affect infant mortality, few pregnancies occur at very short intervals. There-fore, the net effect of very short interpregnancy intervals on rates of infant mortality is probably small. The potential effect, however, could be much larger if any large declines in breast-feeding rates
are not accompanied by increases in use of contra-ception.
Growth
In the developing world, methodologic problems in studies of the effect of infant-feeding practice on growth during the first year of life have led to critical gaps in existing knowledge. Available growth references are based on predominantly hot-tie-fed infants who received early supplementation. These standards may be inappropriate for evalua-tions of growth of breast fed infants. Furthermore, studies from developing countries rarely take into consideration the high percentage of small-for-ges-tationai-age infants and their diminished growth potential.
Theoretical calculations have been used to esti-mate how long a period of exclusive breast-feeding should be sufficient for normal growth. These cal-cuiations involve assumptions about breast milk volume, energy and content, and energy and protein requirements for growth in infancy. Generally, the estimates indicate that breast-feeding becomes in-adequate as the sole source of nutrition after 3 to 4 months of age. However, direct observation of the growth performance of breast-fed infants is a more reliable method than theoretical algorithms for de-termining how long infants can grow normally on an exclusive breast milk diet.
Based on these growth studies, in developing countries breast-fed infants have simiiar incremen-tal weight gains from 0 to 6 months as compared with weight gains in breast-fed infants in developed countries. However, as many of the infants from developing countries have low birth weight, they do not recover from their birth weight disadvantage. Breast-fed infants from both developed and devel-oping countries have slightly lower weight gain from 3 to 6 months than existing growth reference populations. However, the health significance of this is unknown, especially in the developing world where the possibility for later catch-up growth is limited by nutritionally inadequate supplementa-tion and increased burden of infection. Many in-vestigators have considered this normal or optimal growth for breast-fed infants to the point of rec-ommending that growth patterns of breast-fed in-fants be used for reference purposes for all infants from birth to 6 months. Others have not and have described this slower growth as mild growth falter-ing.
In the developing countries, with only one excep-tion, all of the studies reviewed showed a clear growth benefit for breast-feeding v bottle-feeding
to less than 150 g at 20 weeks. Similarly, the use of complementary foods (liquid or solid) in the diet of breast-fed infants during the first 3 months is as-sociated with faltering or no advantage. Between 3 and 6 months, the results are conflicting.
After 6 months, significant growth faltering oc-curs in most infants of the developing world regard-less of method of milk feeding. In general, the nutritional adequacy of supplementary foods that should provide an increasing proportion of nutri-ents in the second half of infancy is poor. The addition of solid foods and the effect of infections make it increasingly difficult to distinguish the effects of method of milk feeding on growth in older infants, although most studies showed a definite advantage of prolonged breast-feeding after 6 months of age. In some communities, breast milk continues to be the only source of food in the second year of life, and in communities where supplemen-tary foods are poor in quality and quantity, breast milk is an extremely important high-quality food during the second half of infancy and beyond.
SUMMARY Domestic
Although the rate of breast-feeding is increasing in the United States, it appears that the rate of increase has been much slower among less well educated and economically disadvantaged women. Relatively little is known about the behavioral and attitudinal factors that affect the decisions to breast-feed or to stop if already breast-feeding. Breast-feeding does appear to decrease an infant’s risk of gastrointestinal infection and otitis media. The effect of method of infant feeding on risk of other infections and allergic illness is less certain.
International
The rate of breast-feeding in developing coun-tries appears to have declined, especially among urban women. Although some sociodemographic correlates of infant-feeding choice have been ex-amined, little is known about the behavioral and attitudinal factors that influence choice and dura-tion of infant-feeding practices. Milk insufficiency, maternal employment, and pregnancy frequently are given as reasons for terminating breast-feeding.
Rates of gastrointestinal illness are lower among breast-fed infants and when such illness is an im-portant cause of death, infant mortality from this
cause appears to be reduced. A randomized clinical trial carried out among high-risk infants found a significantly lower rate of infections among those given breast milk than those fed with infant for-mula. The evidence of the effect of breast-feeding on respiratory tract and other infections from other studies was less clear.
Direct comparison of the growth of predomi-nately breast-fed v artificially fed infants in the same populations from developing countries gen-erally show faster growth for the breast-fed infants for the first 6 months of life. After 6 months, severe growth faltering occurs regardless of the method of feeding. In communities where the nutritional ad-equacy of supplementary foods is poor, breast milk is an extremely important, high-quality food during the second half of infancy and beyond.
RESEARCH NEEDS
Some of the major research recommendations include those listed below. Although these are phrased to reflect domestic needs they also reflect some of the corresponding needs in the developing world.
1. Few of the studies of health effects related to infant feeding practices in the United States have examined disadvantaged populations. These popu-lations, with relatively high rates of infant mortal-ity and morbidity, could be expected to benefit the most from the positive health effects associated with breast-feeding. Studies of the nature and ex-tent of these health benefits are needed. As these and other epidemiologic studies are initiated, inves-tigators need to consider the recent findings of immunologic, growth factor, and nutrient compo-sition studies to refine the end points examined.
2. Research is needed on the normal growth of breast-fed infants. Because existing growth refer-ences are based on studies of bottle-fed infants, it is unclear whether small deviations from these ref-erence curves by breast-fed infants actually repre-sent normal growth.
3. There is little regularly collected data avail-able on prevalences and trends in infant-feeding practices in the United States. Such information is needed not only for breast-feeding but for other practices as well.