Contents - Previous - Next
This is the old United Nations University website. Visit the new site at http://unu.edu
Could you comment on what may be one of the larger effects of breastfeeding on maternal health and well-being, that is, avoiding two pregnancies back-to-back and two children under one year of age?
That is covered in my considerations of the health of the subsequent child.
I understand that there is a protein in human milk that induces programmed cell death in many malignant cell lines. Could it be that in a mother who has been lactating several times, there would be such milk proteins remaining in the gland, which could induce apoptosis of cells that in time might turn into cancers?
An attractive hypothesis, but I don't know enough about it to answer your question. Does anyone else want to take it on?
No, I wouldn't like to take it on. I was going to ask a different question. One of the features of prolonged lactation is the reduction of the number of reproductive cycles that a woman has in her life. As I understand it, most circumstances that produce such a reduction also produce a reduction in ovarian carcinoma. Do you know of any direct evidence of that in women who have breastfed for a long time?
I am not aware of people having looked at that, but again, it is an attractive hypothesis.
You indicated that breastfeeding had an effect on maternal behaviour. Could you say something more about that? And a second question. The United States is going through major social changes. Many children of teenage mothers of lower socio-economic class are born out of wedlock, often without a father present, and the use of drugs (in particular, crack) has reached epidemic proportions. What is the benefit of breastfeeding for this group of mothers?
Dr. Garza has strong opinions on maternal behaviour, so I will ask him to respond to your first question. The issue here is that there are changes in the maternal brain as a result of breastfeeding that change the mother's responses to painful stimuli. Mothers become more placid.
The only thing I would add is that there are some very intriguing animal data suggesting that oxytocin, for example, may change maternal behaviour. It is very difficult to think that a hormone that has played such a basic role in animal physiology would not also have a similar impact on people. But of all the benefits-nutritional, immunologic, and bonding -the last is the least well documented and the least studied both in animals and in humans. A number of investigators around the world are taking a closer look at this. One symposium held in Europe about three years ago looked at the behavioural effects of hormones that we normally associate with lactation.
As a behavioural scientist myself, I think it is very attractive to think in terms of what breastfeeding does and how it could affect maternal behaviour, for example, through oxytocin acting on the brain. But I don't think there are any hard data that would actually support the notion that breastfeeding changes maternal behaviour. Just to carry out such a study would be very difficult. The data on bonding have been primarily related to the issue of immediate contact between the infant and the mother and have not really been related to breastfeeding, as such.
The point that you are making is correct. The hard data available are essentially nil. We are making an inference. I was very careful not to use the word bonding because of exactly the problems that you outlined. Bottle-feeding mothers do feel close to their infants, and what do we have as a measure of closeness? It's not easy.
On your second question, you posed the all-tooreal situation of the teenage mother in the United States. What do we recommend? Such young women are biologically capable of breastfeeding. The issue is whether we recommend it to them. Breastfeeding is done most effectively by women who are committed to it. Teenage pregnancy programmes around the country are giving these girls a lot of support, and some of the investigators are finding that these young women can successfully breastfeed with this kind of support. When you see that, you are seeing a whole lifestyle change, and they may be giving up some of the practices that got them pregnant in the first place. We would like to have a situation in which we could feel good about recommending breastfeeding.
Carol Bryant and colleagues did as controlled a study as you can on a teenage population under those circumstances. They randomly assigned young women to two groups. They educated the women in one group to breastfeed. In the other group, they supported the mothers in every other way but never mentioned breastfeeding. They measured the effects on the mother, and after one year, there was a clear difference. Those women who had breastfed had changed. They had established some self-esteem and had gotten hold of their lives, whereas the mothers who had not breastfed had not changed. They had not deteriorated, but they had not improved in their own self-image and some other behaviours.
So, I do not know whether we can make this a general recommendation. It requires a whole lot of support to happen well, and that is the kind of thing that Carol Bryant is doing.
I have been involved in some support for grassroots projects on mother-friendly workplaces. When we are thinking of policy, we should include schools as the teenage mothers' workplace. It is really amazing how much can be accomplished if teenage mothers can be encouraged to stay in school. Were you including sociocultural environment?
I was considering the support for breastfeeding in the mother's particular home environment. Did her husband or partner support it? Did her mother-inlaw, her mother, her family? I was thinking about various cultural environments. For example, upstate New York is pro-breastfeeding. Davis, California, is perhaps the classic pro-breastfeeding environment. In inner-city Baltimore, the breastfeeding rate is only 4%.
What are your feelings about osteoporosis in women who do not have such high levels of calcium intake as those in Janet King's studies? What do we advise women who are breastfeeding in the United States and in the United Kingdom who wish to give up dairy products because they are worried about colic in the baby? We have seen quite a few breastfeeding women give up drinking milk completely because of concern about fat in the diet. Do you feel that the 300 to 400 mg daily calcium intake in these women would be sufficient to sustain repletion?
The simple answer is that I don't know. There are a number of papers on women whose daily calcium intakes are 1,000 mg or more. They certainly have adequate calcium to make that repletion. I don't know of any study in which women with lower calcium intakes have been followed over a longer period. I am not sure how ethical it would be to randomly assign them to a calcium pill or no pill plus their usual intake. Certainly nobody has done a joint analysis of lactation history, usual calcium intake, and incidence of osteoporosis. You have hit a raw nerve, because I took calcium pills myself when breastfeeding, since I am a non-milk-drinker myself.
In the Gambia we found a small loss of bone calcium in the forearm. In women who were fully breastfeeding with no supplementation at all, there was a repletion before 12 months. There is an urgent need to follow at least some groups of women who are consuming low amounts of calcium. It doesn't have to be a trial, just an observational study.
Before closing this session on the impact of breastfeeding on fertility, I would like to ask Bishop McHugh to comment on a point that some of us were discussing earlier today: whether it is acceptable to the Catholic Church and to the Holy Father if a woman decides to breastfeed with the express intention of not getting pregnant, of delaying her next pregnancy, of spacing her family.
It is an easy answer. Yes, it is all right. There is no prohibition or inhibition on the part of the Church for a woman to use breastfeeding or the LAM (lactation amenorrhea method) as a method of spacing or delaying future births. In fact, it is our intent to encourage that. As Dr. Perez said, it very often correlates with the use of the natural methods of family planning, and indeed, many people in natural family planning today make a direct effort to combine breastfeeding with natural methods of family planning, as the early phase of deferring the next birth. There would be no prohibition by the Church to the use of breastfeeding to defer subsequent birth.
Thank you, Bishop McHugh. I think that is a very important point.
Could I follow up on that? I assume there would be no objection to the use of methods of detecting ovulation, if the people could develop better means. Natural family planning is based on one method, a rather crude method. Would there be any problems if there were better methods of detecting when ovulation occurs?
I would say no. As a matter of fact, at the earlier meeting on natural family planning, a lot of data on more technical methods were presented, but the more technical you get, the less useful it is to a population, especially to a third world population. There is a direct effort being made to find more accurate methods of predicting ovulation.
Peter W. Howie
There is strong evidence that breastfeeding protects infants against infection in environments where clean water cannot be guaranteed, leading to substantial reductions in morbidity and mortality. This is particularly evident in the protection against gastrointestinal disease, although there is also evidence for protection against respiratory infection and otitis media. The evidence for a protective effect of breastfeeding against infection in developed countries has been more controversial, with criticisms being levelled at the methodology of many studies. Evidence is presented from a study in a developed country that met key methodological criteria to show that breastfeeding for 13 weeks offers substantial and continuing protection against gastrointestinal illness. Smaller, but still potentially important, protective effects against respiratory illness also occur. The results add strong support to policies that promote breastfeeding in both developed and developing countries.
"Breast is best" is one of the most widely used and best-known slogans in the promotion of breastfeeding. There are several reasons to promote the support of breastfeeding for all mothers and babies, but one of the most important is the protection that breastmilk offers babies against infectious disease. There is very strong evidence from many studies that breastfeeding confers substantial protection against serious morbidity and mortality in those developing countries where access to clean water is not readily available [1-3]. This subject is discussed in more detail by Victora  in this issue.
Evidence both from epidemiologic studies in developing countries and from the physiological components of breastmilk strongly suggests that breastmilk has an important role in protecting babies against infection. In particular, lactoferrin, which binds iron necessary for coliform growth, and lysozymes, which have bactericidal activity against gram-positive and gram-negative organisms, are supportive of the antimicrobial effect of breastmilk. The influence of the immune system and the secretion of IgA in breastmilk is discussed in this issue by Hanson et al. . Taken together, this evidence indicates a clear and important role for breastfeeding as a means of protecting infants from infection.
Breastfeeding and infant infection in developed countries
The central question in this article is the role that breastfeeding may have in protecting infants against infection in developed countries where clean water supplies are guaranteed. A number of authors have expressed skepticism that breastfeeding has such a role in developed countries .
In 1986 Bauchner et al.  reviewed all the articles that had appeared in the English-language literature since 1970 investigating the relation between infant feeding and health. They scrutinized all studies that contained at least 40 subjects in their sample and applied four key methodological criteria to evaluate the scientific quality and reliability of the studies. The criteria were the avoidance of detection bias, adjustment for confounding variables, a definition of outcome events, and the definition of infant feeding. They examined 14 cohort studies and 6 case-control studies. Eight of the cohort studies reported a protective effect of breastfeeding against infant infection, and six found no evidence of protection. Four of the casecontrol studies found evidence of a protective effect, and two did not. None of the studies found any evidence that breastfeeding might increase the vulnerability of infants to infectious disease.
The significant finding of the review by Bauchner et al. was that they identified significant methodological flaws in all but two of the studies, and even these two could be criticized for having inadequate sample sizes.
To avoid detection bias, it is necessary to ensure that comparative groups of bottle- and breastfeeding motherinfant groups are studied with equal vigour. For example, breastfeeding mothers may be more anxious and therefore more willing to report episodes of infection to their physicians. In addition, it is important to ensure that the interval between any possible infectious episode and the recording by the investigator is sufficiently short, because there is a high incidence of failure to recall significant episodes when the reporting interval is long.
In most developed countries, there is a strong social class bias between bottle- and breastfeeding mothers, with breastfeeding mothers coming from the higher socioeconomic groups. This and other differences between bottle and breastfeeding mothers can introduce important confounding variables that may explain differences in rates of infectious disease in children of the two groups that are not attributable to the method of feeding. Unless great care is taken to allow for confounding variables, it is not possible to attribute reliably any difference in infectious disease frequency to the effect of breastfeeding itself. Studies that do not take due account of confounding variables, therefore, must be regarded as scientifically flawed.
A third important factor is to make clear definitions of the outcome events and to apply them with equal rigour to both bottle- and breastfeeding groups of mothers. In particular, when studying the effect of infant feeding on diarrhoeal disease, it has to be borne in mind that the stool consistency differs between bottle- and formula-fed infants. Unless rigorously applied criteria are applied to the definition of infant disease, it is possible that spurious differences between bottle- and breastfed infants will be observed.
Finally, many mothers do not fall neatly into bottle- or breastfeeding categories, as early introduction of formula and solids, mixed feeding, and use of expressed breastmilk may complicate the definition of infant-feeding categories. It is important, therefore, that infant-feeding groups be defined precisely on the basis of carefully collected contempora neous information about the feeding pattern of the infant.
Applying these criteria, Bauchner et al. found methodological weaknesses in the great majority of the studies that had examined the potential protective effect of breastmilk against infection in young infants. They concluded that "breastfeeding has at most a minimal protective effect in industrialized countries." In my view it would have been more appropriate to conclude that a rigorous study of the potential protective effect of breastfeeding against infection should be carried out using a sample of sufficient size. Because of the methodological weaknesses in the previously published studies, it was equally as possible that an important protective effect of breastfeeding against infection had been concealed or minimized in some of the studies as that a protective effect had been reported that did not exist.
Dundee Infant Feeding and Health Project
Because of the unsatisfactory nature of the published reports, a study was performed in Dundee, Scotland , to investigate the relation of infant feeding and infectious disease using a methodology that took into account the requirements set out by Bauchner et al.
The specific aims of the study were (1) to compare the frequency of childhood illness in breast- and bottle-feeders after correcting for confounding variables and (2) to determine if the frequency of childhood illness is influenced by the duration of breastfeeding.
Details of the study sample and the methods of recruitment have already been published . A total of 750 mothers were recruited to the study, of whom 6 withdrew after delivering their babies. Because the objective of the study was to include only healthy babies of normal birthweight, a total of 70 babies were excluded because they either were delivered before 38 weeks with a birthweight of less than 2,500 g or spent more than 48 hours in the Special Care Baby Unit. This left 674 mother-infant pairs, who were followed up for two years with a 91.5% completed followup rate.
Avoidance of detection bias
To overcome the problem of detection bias, follow-up was carried out by both "hot" and "cold" pursuit. Medical, social, and obstetric data were collected at birth and used in allowing for confounding variables. Each mother-child pair was visited at home at 0.5, 1, 2, 3, 4, 6, 9, 12, 15, 18, 21, and 24 months after birth, and detailed information was collected about the child's feeding and health. Standardized questionnaires were used, and the visitors were given instruction sessions to standardize the method of data collection. In addition to this hot pursuit of information, further data were acquired by scrutiny of the general practice records to identify episodes of illness that were reported to the general practitioner but not to the health visitor. This method of data collection was applied with equal vigour to all groups of mothers in the study.
Definition of outcome event
The intention of the study was to define the effect that infant feeding would have on episodes of serious illness. The definitions of illness were adapted from those used by Chandra et al.  in a previous study. The episodes of infectious disease had to last at least 48 hours before they were considered significant. Gastrointestinal infection was defined as vomiting and diarrhoea, respiratory infection as coryza and cough or wheeze, and ear infection as painful discharging ear, all lasting for at least 48 hours. Infections of mouth and skin were determined by the presence of inflammation, and eczema was diagnosed by a physician on the basis of typical signs. The episodes of illness were reported to the study coordinator, and if there was any doubt whether an episode qualified for inclusion under the study criteria, a decision was taken by a paediatrician who reviewed the information without knowing the feeding category of the subject.
Definition of infant feeding
During each visit detailed information was collected about the infant's feeding in the previous 24 hours. This was done to ensure that accuracy of recall would be maximized. Information was collected about the number of breastfeeds, the number of formula feeds, the number of solid feeds, and the number of juice and water feeds. This information was used to calculate the dates of the first introduction of formula, cow's milk, and solid feed as well as the last date of breastfeeding.
On the basis of this detailed contemporaneous information about breastfeeding, babies were allocated to one of four groups. (1) Bottle-feeders were exclusively bottle-fed from birth. (2) Early weaners were breastfed for less than 13 weeks. (3) Partial breastfeeders were breastfed for 13 weeks or more, but formula or solid-food supplements were introduced before the age of 13 weeks. (4) Full breas-feeders were breastfed for 13 weeks or more, and no formula or solid-food supplements were introduced before the age of 13 weeks. The comparative analyses were carried out using these four infant-feeding groups.
Adjustment for confounding variables
A total of 40 potential confounding variables were identified in the groups of maternal factors, infant factors, and paternal factors. Although a number of factors differed between bottle- and breastfeeding mothers, the three most significant confounding variables among the main groups were social class, parental age, and maternal smoking. A regression analysis technique was used, and after allowing for these three factors, all other differences were nonsignificant. Reanalysis of data introducing a greater number of confounding variables made no difference to the final conclusions.
Calculation of sample size
The principal outcome variable was determined as the incidence of gastrointestinal disease in the first 13 weeks of life. This was expressed as the proportion of babies having one or more episodes of gastrointestinal disease during that time. On the basis of the data of Fergusson et al. , it was hypothesized that breastfeeding would lead to a reduction in the incidence of gastrointestinal illness from 16% to 8% in the first 13 weeks of life. It was calculated that a sample size of 560 mother-baby pairs would be needed to achieve a 90% certainty of a significant result at the .05 level (two-tailed test). Thus, the actual sample size of 674 resulting from the original recruitment of 750 mothers was more than sufficient.
Incidence of gastrointestinal disease during the first 13 weeks
Figure 1 shows the percentage of babies with one or more episodes of gastrointestinal disease in the first 13 weeks of life, according to their infantfeeding category as defined above. The unshaded columns show the observed incidence of gastrointestinal illness, and the shaded columns show the adjusted incidence after allowing for the three confounding variables described above. The observed incidence of gastrointestinal illness in bottle-fed babies was 19.3%. This did not differ significantly from the observed incidence (18.3%) in babies who were weaned before 13 weeks, usually within a few days of birth. These incidences contrasted sharply with those in the partially and fully breastfed groups (4.8% and 2.2%, respectively). The latter two groups did not differ significantly in the frequency of gastrointestinal illness. When the incidence of gastrointestinal illness in the breastfed babies, whether partially or fully breastfed, was compared with that in the bottle-fed babies, there was a highly significant difference that persisted after adjustment for confounding variables (p < .001).
Similar comparisons were made among these four feeding groups for the periods 14 to 26 weeks, 27 to 39 weeks, and 40 to 52 weeks. The results are shown in table 1. Although the relation between breastfeeding and protection against infection became less clear-cut as time progressed, there was a persistent reduction in the incidence of gastrointestinal illness in those babies who had been fully breastfed for 13 weeks compared with those who were bottle-fed from birth. This difference persisted despite the fact that many of the mothers who fully breastfed their babies during the first 13 weeks had completely weaned them by the age of one year.
The relation between infant feeding and respiratory infection was also analysed (table 1). In general, the incidence of respiratory infection during the first year of life was less among partially and fully breastfed babies, but the differences were much smaller than those for gastrointestinal illness, and they were only observed during the periods from 0 to 13 weeks and from 40 to 52 weeks.
Effect of breastfeeding on hospital admissions for infectious disease
The effect of breastfeeding upon the incidence of hospital admissions of babies for infectious disease was studied, using babies who were bottle-fed from birth as the comparative group (fig. 2). When the results were expressed as an odds ratio, the incidence of admission was similar for babies who were breastfed for less than 13 weeks and for those who were bottle-fed from birth. By contrast, the incidence of admission was significantly lower for all the groups of babies who were breastfed for more than 13 weeks than for those who were bottlefed from birth. No significant protective effect of breastfeeding was observed against ear, mouth, eye, or skin infection, colic, eczema, or nappy rash.
Breastfeeding for 13 weeks reduces the rate of gastrointestinal illness in infants during the first year of life after correcting for confounding variables. The protective effect of breastfeeding is not dependent upon the avoidance of early supplements and persists beyond the period of breastfeeding itself. There is a similar but smaller reduction in the rate of respiratory illness in babies who are breastfed, for 13 weeks. Breastfeeding for 13 weeks also reduces the incidence of hospital admission owing to serious episodes of infectious disease
FIG. 1. Percentage of babies with one or more episodes of gastrointestinal illness during the first 13 weeks of life according to early infant-feeding group. Unshaded columns show observed percentages, and shaded columns show percentages adjusted for social class, parental age, and maternal smoking
TABLE 1. Numbers (percentages and percentages adjusted for social class, parental age, and maternal smoking) of babies with infections up to one year of age according to method of feeding in the first 13 weeks
|Type of infection||Weeks||Bottle-fed (n=426)||Early weanears (n=161)||Breasfed||% for adjusted rates||95% CI for diference between bottle- and breadfed|
|Partially (n=121)||Fully (n=89)||Bottle-fed vs early wearnes||Partially vs fully breastfed||Bottle-vs breasfed|
|Gastrointestinal||14-26||47||(19.1, 17.9)||33||(20.5, 20.2)||8||(6.6, 6.9)||7||(7.9, 8.7)||0.33||0.21||8.06a||4.0 to 16.2|
|27-39||55||(22.3, 21.8)||35||(21.7, 21.1)||20||(16.5, 16.5)||7||(7.9, 8.5)||0.03||2.79||4.03b||2.5 to 16.1|
|40-52||55||(22.4, 22.7)||21||(13.0, 12.8)||19||(15.7, 15.2)||6||(6.7, 6.6)||6.23b||3.85b||6.70a||5.1 to 18.5|
|Respiratory||0-13||100||(38.9, 37.0)||54||(32.0, 31.7)||29||(23.0, 24.2)||22||(23.2, 25.6)||1.22||0.05||6.35b||3.9 to 20.3|
|14-26||116||(47.1, 46.0)||76||(47.2, 47.2)||42||(34.7, 35.5)||34||(38.2, 39.9)||0.06||0.40||2.80||- 0.8 to 17.4|
|27-39||112||(45.5, 42.9)||85||(52.8, 52.6)||57||(47.1, 49.5)||32||(35.9, 39.6)||3.48||1.89||0.31||- 10.8 to 7.5|
|40-52||133||(54.1, 53.3)||72||(44.7, 44.6)||46||(38.0, 38.6)||38||(42.7, 44.2 )||2.79||0.63||5.57b||2.7 to 21.1|
Sourece: ref 8.
FIG. 2. Babies admitted to hospitals in their first year: odds ratio adjusted for social class, parental age, and maternal smoking according to duration of breastfeeding. breastfed is given an odds ratio of 1.0
The conclusions of this article are entirely dependent upon the rigour of the methodology. Every precaution was taken to meet the methodological criteria set out by Bauchner et al. . Assuming that the study is regarded as methodologically sound, it provides very strong evidence that breastfeeding offers substantial protection against gastrointestinal illness in developed countries. The 19% incidence of vomiting and diarrhoea during the first 13 weeks of life was very similar to the 16% incidence reported by Fergusson et al.  during the first four months of life in their study in Christchurch, New Zealand. Bearing in mind that the episodes of illness that were reported had to last for at least 48 hours, it is clear that babies who are bottle-fed from birth are exposed to a substantial risk of significant episodes of gastrointestinal illness not only during the first 13 weeks but throughout the first year of life, as shown by the much lower incidence of hospital admission among babies breastfed for 13 weeks or more.
A further analysis was carried out to examine the possibility that breastfeeding for less than 13 weeks would offer significant protection. The results showed that breastfeeding for 6 weeks did not confer the same level of protection as breastfeeding for 13 weeks, and confirmed the view that a minimum period of 13 weeks should be recommended to ensure a substantial degree of protection. Another important finding was that exposing infants to breastmilk for only the first few days of life is insufficient to offer any significant protection against infection, suggesting that exposure to colostrum alone is not enough.
This study provides strong evidence to support the promotion of breastfeeding for all mothers in developed countries [3,11,12]. It reinforces the need to provide mothers who wish to breastfeed adequate support and facilities in the form of health-service support, creche facilities, and adequate maternity leave. It is also possible that the benefits gained from breastfeeding in the early weeks of life may continue into later stages of infancy and childhood, and this possibility is being examined through further follow-up studies of this cohort of mother-baby pairs.
This study was supported by the Scottish Office Home and Health Department.
1. Plank SJ, Milanesi ML. Infant feeding and infant mortality in rural Chile. WHO Bull 1973;48:203-10.
2. Victora CG, Smith PG, Vaughan JP, Nobre LC, Combardi C, Teixeira AMB, Fuchs SMC, Moreira LB, Gigante LP, Barros FC. Evidence for protection by breastfeeding against infant deaths from infectious diseases in Brazil. Lancet 1987;ii:319-22.
3. Cunningham AS, Jelliffe DB, Jelliffe EF. Breastfeeding and health in the 1980s: a global epidemiological review. I Pediatr 1991;118:659-66.
4. Victora CG. Infection and disease: the impact of early weaning. Food Nutr Bull 1996:17:390-96.
5. Hanson LA' Wiedermann U. Ashraf R. Zaman S. Adlerberth 1, Dahlgren U. Wold A, Jalil F. Effects of breastfeeding on the baby and on its immune system. Food Nutr Bull 1996:17:384-9.
6. Kovar MG, Serdula MK, Marks JS, Fraser DW. Review of the epidemiologic evidence for an association between infant feeding and infant health. Pediatrics 1984;71(suppl):615-38.
7. Bauchner H. Leventhal JM, Shapiro ED. Studies of breastfeeding and infections. How good is the evidence? JAMA 1986;256:887-92.
8. Howie PW, Forsyth JS, Ogston SA, Clark A, Florey CD. Protective effect of breastfeeding against infection. BMJ 1990;300:11-6.
9. Chandra RK. Prospective studies of the effect of breastfeeding on incidence of infection and allergy. Acta Paediatr Scand 1979;68:691-4.
10. Fergusson DM, Horwood J. Shannon FT, Taylor B. Breastfeeding, gastrointestinal and lower respiratory illness in the first two years. Aust Paediatr J 1981; 17:191-5.
11. Habicht JP, Da Vanzo J. Butz WP. Does breast-feeding really save lives, or are apparent benefits due to biases? Am J Epidemiol 1986;123:279-90.
12. Standing Committee on Nutrition of the British Paediatric Association. Is breast feeding beneficial in the UK? Arch Dis Child 1994;71:376-80.
Contents - Previous - Next