The debate concerning the merits of breast feeding rages still (unfortunately). Over the past fifty years, there has been a huge swing from a very strong pro-formula position to a pro-breastfeeding position both in general society and within the medical field. This swing has been partly driven by scientific findings about the advantages breast milk has for the infant: faster and more thorough immune system development, lower infection rates, greater resistance to many diseases, reduced chances and severity of colic and constipation, greater intelligence, etc1. However, very recently (21 September 2002) a fascinating and controversial study2 was published in The Lancet, one of the premiere international medical journals. The study found that breastfeeding a child for more than four weeks led to an increased risk of developing asthma or skin allergies later in life (ages 9 to 26). This finding is of particular interest to the medical community since previous studies have produced contradictory results and have suffered from a number of methodological shortcomings. It is also of great interest to the general public, as these findings fly in the face of current wisdom. For these reasons, this write-up will deal principally with the contents of this paper, given that most private citizens will have a very time obtaining a copy of the publication. Some popular press concerning this study can be found at the following sites:
  • c=Article&cid=1032473017919&call_pageid=10146562738273
There are other news stories around the web, as this study has already garnered a lot of attention. is your friend.

The purpose of this write-up is not to simply re-iterate the information found in the above articles, or that easily found on the web by interested parties. Instead, it will present an abridged and 'vulgarized' version of the paper itself, which will be followed by a brief commentary and critique.

The Study

Directly from the article (the Background, which is part of the abstract):
Breastfeeding is widely advocated to reduce risk of atopy and asthma, but the evidence for such an effect is conflicting. We aimed to assess the long-term outcomes of asthma and atopy related to breastfeeding in a New Zealand birth cohort.
While many in the medical community and much of the scientific literature suggests that breastfeeding of infants should be pursued in order to reduce the risks of later development of atopy and asthma, studies prior to this one had produced some contradictory results. As a result, a 1988 paper4 by M. S. Kramer proposed 12 criteria for future studies which wish to examine the relationship between breastfeeding and asthma, and based on these 12 criteria found that most previous studies were lacking in rigour (thus making their findings suspect). Despite this important paper, further studies continued to appear with contradictory results and again (possibly) improper methodologies. The authors of the Lancet study followed the Kramer protocols to try to ensure that their study would produce defensible results. These criteria are:


  • Non-reliance on late maternal recall of breastfeeding
  • Blind ascertainment of infant feeding history
  • Sufficient duration of breastfeeding
  • Sufficient exclusivity of breastfeeding
  • Strict diagnostic criteria
  • Blind ascertainment of outcomes
  • Consideration of severity of outcome
  • Consideration of age of onset of outcome
  • Control for confounding
  • Assessment of dose-response effects
  • Assessment of effects in children at high risk of outcome
  • Adequate statistical power

    The authors initiated their study with the hypothesis that breastfed children would display a lower incidence of asthma and atopy later in life.

    The cohort

    The authors of the Lancet study had the opportunity to test, rigorously, the relationship between breastfeeding, asthma and atopy thanks to a study iniated in 1972 at the Queen Mary Hospital in Dunedin, New Zealand. A cohort of children born between April of 1972 and March of 1973 were included in a neonatal study. Those children still residing in the province of Otago at 3 years of age (1975-76) were invited to participate in the longitudinal Dunedin Multidisciplinary Health and Development Research Study. In total, 1037 children were enrolled, and these children form the basis of the paper being discussed. The families of these children spanned the entire socio-economic range of the general New Zealand population, and were mostly caucasian. The socio-economic status of the families was, however, recorded on a semi-quantitative scale (Elley-Irving scale) for later use in statistical models.

    The 1037 children in the study were assessed for a number of conditions and their general health within a month of their date of birth at ages 3, 5, 7, 9, 11, 13, 15, 18, 21 and 26. They were accompanied by their parents during their assessments up to the age of 11, and were examined individually thereafter. Of course, not all children were present and tested for their respiratory condition at each year of investigation, but the representation was very good (from 71% for age 7 to 96% at age 26). Skin testing (for atopy) was conducted only at ages 13 and 21, again on a subset of the whole cohort (69% and 87% respectively).

    Data Collection

    The breastfeeding history was documented by independent interviewers at age 3. The duration of breastfeeding and the age at which cow's milk was introduced were recorded by the interviewer and verified against the information gathered through the New Zealand Plunket Nurse programme, where nurses see mother and child both in clinics and at the home every week after birth and then less frequently until the age of 2 or 3. While those children who were breastfed did often receive formula during night feedings while at the hospital, this was only for 3 or at most 4 days, and the researchers considered this quantity to be insufficient to eliminate the benefits associated with breastfeeding. A participant was considered to be a member of the breastfed group if the child had been exclusively breastfed for at least four weeks. Using this classification, 45% of the children were not breastfed, 49% were breastfed and 7% were breastfed initially, but were moved to formula before the child was 4 weeks old (for a variety of reasons). This latter group was statistically indistinguishable from the non-breastfed cohort, and thus was included in the non-breastfed group. For the breast-fed cohort, the mean duration of breastfeeding was 21.1 weeks.

    At age 7, trained interviewers asked the parent whether the child suffered from asthma, hayfever, wheezing or other respiratory problems. During this same interview, the family history of asthma or respiratory problems was also recorded. At age 9 a more comprehensive questionnaire was introduced, wherein the frequency and occurrence of asthma or wheezing was recorded. At ages 18, 21 and 26 the participants completed questionnaires based on those from the American Thoracic Society and the International Union Against Tuberculosis and Lung Diseases. Those participants using bronchodilators were asked to withhold on the day of the study, and all individuals were tested using a spirometer. After an intial test in which the forced exhaled volume (FEV) and vital capacity were recorded, treatment with methacholine was performed and spirometry repeated. Presence of a respiratory problem or defect was recorded for the participant if their FEV was reduced by 20% or more following treatment with methacholine. In the case of children already showing signs of respiratory distress, the patients were treated instead with salbutamol and a problem or defect was recorded if their FEV improved by over 10%.

    To test for the presence of atopy, skin prick tests were performed at ages 13 and 21. Participants were excluded from this trial if they had used antihistamines within the 2 days prior to the investigation. The allergens tested were dust mites, rye grass pollen, cat dander, dog dander, horse hair, kapok, cladosporium, Aspergillus fumigatus, Penicillium spp., Althernaria spp. and wool, with both a positive and negative control. After 15 minutes, an positive test was recorded if the pin prick produced a weal at least 2 mm larger than the negative control.

    Statistical analyses were performed on these data to determine first whether there was any difference between the breastfed and non-breastfed children with respect to the occurrence of asthma and atopy. The initial tests used were simple chi-squared and likelihood ratio tests to compare the breastfed to non-breastfed groups. Later, multivariate methods were used in a step-forward manner in order to attempt to control for some confounding factors.

    The results

    Whenever you perform a mensurative study, some things are beyond your control. In the case of this study, the breastfed and non-breastfed cohorts were not identical. Children were more often breastfed if they were first born, if their parents were of higher socioecononmic status or if their mothers were non-smokers. Breastfed children were also more likely to have had sheepskin on their bed in infancy. Luckily for the researchers, these three factors would suggest that the breastfed cohort may have been biased to having a lower risk of asthma or allergies, while they found the opposite relationship to be true.

    At age 13, positive skin tests to cat, house dust mites, grass, alternaria or any other allergen were more likely to occur in the breastfed cohort. This same pattern was repeated in the tests conducted at age 21. In fact, the odds ratios (ie., the risk of showing a response in the breastfed group / risk in the non-breastfed group) ranged from 1.48 to 2.41 for these tests, meaning that breastfed children were as much as two and one-half times more likely to show an allergic response. The same pattern was repeated when examining the evidence of asthma or respiratory distress. At every age from 9 to 26, breastfed children were more likely to report having asthma than their counterparts. The odds ratios varied from 1.50 to 2.93 (three times!). The same pattern was uncovered when examining the respirometry results before and after treatment with methacholine or salbutamol: significant increases in the incidence of distress were observed in the breastfed cohort in every case, except for those children wheezing (not diagnosed as asthmatic) at age 15.

    When the data were examined in a more intensive manner, it was found that any breastfeeding beyond the age of three weeks resulted in an increase in the risk of atopy at age 13. For example, while those children never breastfed or breastfed for fewer than four weeks had a 37% and 34% risk of displaying atopy to at least one of the tested allergens, those breastfed for 4-7, 8-11, 12-15, 16-25 and more than 26 weeks had 58%, 49%, 68%, 48% and 46% chances of showing such a response. This same pattern was found when testing for the incidence of atopy at 21 years or asthma at 9 years of age.

    As should be expected, those children from families with a parental history of hayfever or asthma developed atopy or asthma at a greater rate than their counterparts, irrespective of whether they were breastfed. However, despite this family link the increased incidence in the breastfed group was still clearly visible and significant. Finally, in their multifactor analysis where they controlled for socioeconomic status, birth order, sheepskin use in infancy and the presence of asthma, atopy or smoking in/by the mother, all outcomes remained significant with the breastfed cohort again showing higher risk.


    The authors were clearly surprised by the outcome of their study. They had anticipated exactly the opposite. To the best of their knowledge and understanding, their study is the most comprehensive and well-designed of all those examining the link between breastfeeding, asthma and atopy. They opine that the reason some prior studies have found the opposite result is due to the fact that the outcomes (asthma and atopy) are assessed at very early ages. For example, those which test for asthma at age 2 or less found evidence of protection, while theirs and others find increased risk at greater age.

    The authors cite studies of Swedish, Californian, and Australian children who were breastfed had lower incidences of atopy at a young age, and in Kenya breastfeeding was related to a reduced severity of asthma, rather than incidence. A massive Brazilian study found lower rates of asthma in children breastfed for more than six months than those who were never breastfed. On the other hand, Tuscon, Italian and British children show that breastfeeding lead to increased risk of atopy and asthma.

    It is important to note that while some the studies mentioned in the previous two paragraphs were undeniably massive in their undertaking, none can hold a torch to the Lancet study discussed here. All previous studies confined themselves to cases where the children were, or were not, genetically disposed to asthma and/or atopy. They did not attempt to control for smoking, the date of breast-feeding cessation, socioeconomic status and other factors. This study was superbly designed and produced results which are hard to deny.

    To date the mechanism relating breastfeeding to asthma and atopy in children is poorly understood, which limits ability to make sense of these contradictory results. Without a causative mechanism, it is difficult to understand why one study produces one result while another produces something quite different. A study5 examining infants in Estonia and Sweden, countries with low and high prevalences of atopy, respectivley, found that one year old Estonian children had high densities of Lactobacilli and eubacteria in their guts, while Swedish children had high densities of clostridia instead. Those childern with lower lactobacilli densities and higher coliform densities were more likely to develop allergies. Thus, breastfeeding might affect the balance of the bacteria in a child's gut, giving way to future susceptibility to allergy.

    Commentary and critique

    When I first read about this story in the local newspaper, my reaction was two fold: first, this must have been paid for by a formula company; second, their methodologies must be totally skewed. To my surprise, upon reading the paper I found that this is not just solid science, but really good science. This study was not paid for by any multinational that I can find, and the authors seem to convey a real sense of discouragement (and excitement; they are scientists, after all!) at these findings. They state, as the last paragraph of their discussion:
    Breastfeeding could be promoted for many reasons, including optimum nutrition and reduction of risk of infant infections. However, the role of breastfeeding in protection of children against atopy and asthma cannot be supported on the basis of the present balance of evidence.
    For those in the scientific community, the previous two sentences are as close to an emotional lament as you get.

    I still think that breastfeeding should be encouraged and even pursued aggressively by the medical community, support groups and federal/state policies. It is cheaper, easier and better for the baby. However, it may not be the best nutritional course in every conceivable way. In the balance, by far better than formula. However, be aware that you might be increasing your child's risk of developing ashtma and/or allergies later in life.

    1 Note that all of these findings, as well as the ones discussed in this write-up, are correlative and not causative. This means we are implying or assuming that the existence of a pattern (A and B covary) is evidence that one causes the other (A causes B). See correlation is not causation for a good explanation of this principle; in particular, the write-up by bitter_engineer.
    2 Sears, Malcolm R., Justina M. Greene, Andrew R. Willan, D. Robin Taylor, Erin M. Flannery, Jan O. Cowan, G. Peter Herbison and Richie Poulton. "Long-term relation between breastfeeding and development of atopy and asthma in children and young adults: a longitudinal study." 2002. The Lancet. 360: 901-907.
    3 Please be aware that I've added a space (" ") between "Type1&" and "c=Article" in this address so that the page will format normally for anyone not blessed with a 23" CRT or an Apple TiTanium Powerbook.
    4 Kramer, M. S. "Does breast feeding help protect against atopic disease? Biology, methodology and a golden jubilee of controversy." 1988. The Journal of Pediatrics. 112: 181-190.
    5 Sepp, E., K. Julge, M. Vasar, P. Naaber, B. Bjorksten and M. Mikkelsaar. "Intestinal microflora of Estonian and Swedish infants." 1997. Acta Pediatrica. 86: 956-61.