Breast milk is acknowledged as the optimal food for the new-born infant. Foda et al (1) show that breast-milk composition is modified after massaging the breast using the Oketani method. These findings merit attention because the results suggest that the estimated increase in energy intake in the late lactating period (after 3 months) is clinically relevant. Using the observed average gross energy of 67.2 and 74.1 kcal/100 ml of pre- and post-massage milk respectively in the late lactating period, and using an average estimated breast-milk intake of 600 ml/24h in a breast-fed infant in the second and third trimester of life (2), the energy intake may be around 403 rather than 445 kcal/24h, a 10% increase in energy intake due to breast-massage. This supplement of energy may be of special interest at the age when weaning foods are introduced in populations where such weaning foods are of low energy density. It is likely that other quantitative or qualitative changes in the composition of breast-milk occur, which may also be of clinical interest.
With this intervention, although the first step is indeed to ensure that a substantial enrichment of breast-milk may be obtained, the story does not stop there. Key questions follow. What might the mechanism be? Is a standardisation of the massage technique possible to ensure reproducible changes in milk composition? Can there be a generalisation of the massage technique? Does such a change in breast-milk composition warrant a revision of feeding practices? To justify changing the recommendations on feeding practices, is it enough to know that an intervention changes breast-milk composition, or should other evidence be sought?
To test for the effectiveness of therapeutic interventions, the reference research tool, which has evolved over the last 50 years, is the randomised controlled trial. Randomisation is used to reduce bias in assigning patients to treatment/intervention. Controls are used to reduce observer bias (comparator treatment or placebo), measurement bias (use of matched control subjects), group membership bias (e.g., statistical methodology of blinding), and to avoid chance findings (e.g., statistical methodology of power calculation). The design of such trials is critical if credible results are to be obtained. The trials must adhere to a tightly designed protocol to ensure that subjects in all treatment groups are treated and observed similarly during the trial. International guidelines are available (3–6).
The aim of a controlled trial is allow judgments on the efficacy of the intervention that stands up to public and scientific scrutiny. Is it possible to apply these principles to show the efficacy of an intervention for breast feeding? There are many inherent difficulties in the scientific approach to studies in breast feeding. Such issues critical to study design are the methodology or type of intervention, the use of control groups and the reduction of bias, and the endpoints and outcomes of interest.
In Foda's study, the intervention is breast massage. The authors describe it as a connective-tissue massage, and indicate that special training and technique are essential to practice it. Therefore, if further research is to be done using this technique, it would have to be fully described and standardised in a detailed protocol so that it will be practised in a reproducible manner during a controlled trial. Health professionals undertaking the trial would need thorough training in the massage technique. A further hurdle may be the teaching of this technique to the breast-feeding mothers themselves.
Randomisation to an intervention is critical, but in the case of an intervention that cannot be blinded such as breast massage, how does one reduce observer bias or group membership bias? Such aspects will require careful study design, and unfortunately, are likely to remain a source of bias in the interpretation of results. Matched control groups will be required in an attempt to reduce confounding factors.
With regard to outcomes and endpoints, it is necessary to identify those that are likely to have implications for clinical practice. In Foda's study, changes in macronutrients were measured. More subtle changes, such as the content of essential fatty acids or changes in micronutrients, may also occur consequent to manipulation of the breast tissue, and these factors should be studied. As the authors mention, changes in non-nutritive substances such as opioids or small peptides may or do occur, and their physiological significance also warrants study. Fat concentration in breast milk is relatively easily measured and it is thus an attractive endpoint for a research study. Fat concentration however, varies during a feed (lower fat concentration in fore-milk than in hind-milk), and a well-defined sampling methodology must therefore be applied to ensure reproducibility of results both between investigators and between studies (7,8). Breast-milk fat content has a plausible biologic link to infant growth via the energy input. Is showing an increased fat content sufficient? Should we not also measure milk intake, for example by deuterium oxide dilution? The qualitative composition of the fat intake after massage will also be of importance and should be measured. More detailed endpoints such as these might more convincingly indicate an intervention has the potential to optimise growth or health in infancy. In the long term, since events in early life are thought to impact health later in life (see the recent controversy about the protective role of breast-feeding on later obesity (9)), other outcomes may need to be measured such as the infants' growth, body composition, cognitive development, or health status.
Should these difficulties in designing meaningful trials prevent research on breast feeding? Certainly not. The pursuit of ideas with the potential to enhance infant nutrition and health and advance the science of this field justifies the perseverance of scientists and an innovative approach to study design.
1. Foda MI, Kawashima T, Nakamura S, Kobayashi M, Oku T. Quantitative differences in the composition of milk obtained from un-massaged versus massaged breasts of lactating mothers. J Pediatr Gastroenterol Nutr
2. Dewit O, Dibba B, Prentice A. Breast-milk amylase activity in English and Gambian mothers: effects of prolonged lactation, maternal parity, and individual variations. Pediatr Res
3. International Conference on Harmonisation of Technical Requirements for registration of pharmaceuticals for human use: May 1996. ICH harmonised tripartite guidelines. E6: Good Clinical Practice consolidated guideline. Available at: http://www.ich.org
. Accessed March 5, 2004.
4. International Conference on Harmonisation of Technical Requirements for registration of pharmaceuticals for human use: July 1997. ICH harmonised tripartite guidelines. E8: General Considerations for Clinical Trials. Available at: http://www.ich.org
. Accessed March 5, 2004.
5. International Conference on Harmonisation of Technical Requirements for registration of pharmaceuticals for human use: February 1998. ICH harmonised tripartite guidelines. E9: Statistical Principles for Clinical Trials. Available at: http://www.ich.org
. Accessed March 5, 2004.
6. International Conference on Harmonisation of Technical Requirements for registration of pharmaceuticals for human use: July 2000. ICH harmonised tripartite guidelines. E10: Choice of control group and related issues in clinical trials. Available at: http://www.ich.org
. Accessed March 5, 2004.
7. Prentice A, Prentice AM, Whitehead RG. Breast-milk fat concentration of rural African women. 1. short term variations within individuals. Br J Nutr
8. Prentice A, Prentice AM, Whitehead RG. Breast-milk fat concentration of rural African women. 2. long-term variations within a community. Br J Nutr
9. Butte NF. The role of breastfeeding in obesity. Pediatr Clin North Am