Infant nutrition research difficulties

 What are some of the problems with studies on how to best feed babies?

Research on the differences between families that breastfeed and those that use infant formula can be challenging. To do the best kind of studies, researchers usually try to compare similar people who have been randomly assigned to one of two groups. The researchers treat one group differently and compare the results. Mothers, however, cannot be randomly assigned to breastfeeding and non-breastfeeding groups, so researchers have to find other ways to study the differences. Another complication is that some mothers combine breastfeeding and infant formula. The amounts of breast milk and infant formula that the babies of these mothers get can vary widely and affect research results. Research can be skewed by commercial interests, the characteristics of families who do or do not breastfeed, the difficulty of measuring long-term effects, the size of the studies, and the interpretation of the results.

A) Evidence for breastfeeding recommendations

While there are challenges to doing research about the benefits of breastfeeding, there is clear evidence that, other than in a very few specific circumstances, breastfeeding and breast milk feeding results in healthier babies and mothers and benefits families, societies, and our environment.

This evidence is the reason that health-care providers, healthcare systems, and other groups concerned with health and well-being universally recognize breastfeeding as the ideal way to feed babies.

Breastfeeding research can be challenging for a number of reasons.

B) Lack of randomized studies

1) Inability to randomize breastfeeding families

The best type of research is the randomized controlled trial (RCT) in which one group of people is divided into two equal groups (randomizing). One group has an intervention and the other does not. The two groups are then compared after the intervention.

Researchers cannot randomize mothers into a you-will-breastfeed and a you-will-not-breastfeed group. It is clearly not ethical. There are many accepted ways to work around this using different types of studies and mathematical analysis, but it remains a concern for researchers (Cebolla-Boado  2017; Huang 2016).

2) Describing randomized studies

Here is an example of an RCT. A diagram of this is found at the bottom of this FAQ.

Take a large group of similar people with the same disease. 

  • Assign them randomly to one of two groups so that both groups have similar people.
  • Give pills to the members of one group and not to the others.

At the end of the study, if the people in the pill group  are  less sick than the no-pill group, you can assume the pill works.

If the two groups are not identical at the start, the differences may not be due to the pill but due to differences between the groups. That is why randomizing is important.

For example, continuing with the above study, assume the following:

  • All of the people are told the pill will probably make them better but there is a small chance it can make them worse. 
  • Some people with more severe disease are afraid to use the pill and insist on being in the no-pill group.
  • Some people with less severe disease are willing to take the pill and more  of them are put into the pill group.
  • The pill is useless.
  • The disease does not get better or worse with time.

The starting group was not well randomized as the no-pill group has more sick people and the pill group has healthier people. If the two groups are examined at the end of the study, it would appear that the pill worked, because the pill-group people are less sick than the no-pill group.  However, the difference is actually because the two groups were not similar: they were not randomized.

C) Assigning mothers who combine breastfeeding and infant formula into “breastfeeding” groups 

Exclusive breastfeeding may show a benefit compared with infant formula use, but this benefit may be reduced or eliminated by any infant formula use. A study in which mothers who are partially breastfeeding are assigned to a “breastfeeding” group may not show a benefit of breastfeeding when there actually is one.

For example, it is possible that even small amounts of infant formula given shortly after birth to babies who are otherwise fully breastfed can increase the risk of cow’s milk protein allergy. Such babies may be put into the “breastfeeding group” of a study on allergy and minimize the difference between breastfeeding and infant-formula fed babies.

D) The definition of breastfeeding

Partial breastfeeding can mean as little as 30 millilitres (1  U.S. fluid ounce) of infant formula a day and breastfeeding the rest of the day, or as much as 700 ml  (24  oz) of infant formula and breastfeeding only once a day. Therefore, the amount of breast milk these two types of babies receive is very different. They may have different outcomes even though they are in the same partial-breastfeeding group.

E) The characteristics of the mother who chooses to breastfeed 

In higher-income countries, mothers who breastfeed are more likely to earn more, have more education, be married, live in optimal housing, and have lower rates of smoking and drug abuse. They may have more resources to support the child’s development and education and create a healthier environment. It can be argued that some of the benefits thought to come from breastfeeding actually come from living in a healthier environment (Raissian 2018).

Mothers who have health problems such as obesity and diabetes are more likely to have difficulty breastfeeding. Their children are also at risk of similar challenges.

Some or all of these differences may be present and, unknown to researchers, influence the outcome of the study.

F) Reversing cause and effect

Some relationships are clear: A causes B but B does not cause A.

Falling into mud (A) makes an individual more likely to have a bath (B). It is clear that bathing (B) does not cause a person to fall into the mud (A). Similarly, having sore nipples (A) may cause a mother to switch to infant formula (B) but infant formula-feeding (B) does not cause sore nipples (A).

Other relationships may be less clear. It is possible that A and B are related but not in the way you would expect. Instead of A causing B, it is possible that B is causing A. This is called reverse causation or reverse causality.

For example, breastfeeding (A) may protect against eczema or asthma (B). However the mothers of babies with these diseases may try to minimize the effect of these diseases by breastfeeding for longer periods. So the baby’s disease (B) causes breastfeeding (A). However the results of a study looking at the relationship between breastfeeding and eczema and asthma may wrongly conclude that breastfeeding makes babies more prone to these diseases (Lamberti 2011; Lodge 2008; Bigman 2019).

G) Long-term effects

Studying long-term effects is complicated. For example, it is possible that adding a particular ingredient to infant formula increases the risk of health problems once the babies reach adulthood. The effects may take 20 or more years to develop.

A researcher trying to assess this effect will need to know what type of formula a mother used. However, after 20 years, memories can be vague and data can be lost. A better test would be to identify a group of babies who currently receive the infant formula in question and another who do not and follow both for 20 years. Such studies are very expensive and often involve several generations of researchers.

Long-term risks or benefits of recent changes to infant formula ingredients cannot be examined as they have not yet developed (Vandenplas 2016).

H) Commercial interests

Research is not done in a vacuum and commercial interests can influence individuals, institutions, and publishing journals (Godlee 2019; Institute of Medicine 2009; Lundh 2018). This also applies to research about nutrition and in particular to feeding babies and infant formula (Chartres 2016; Kunc 2020; Penders 2017).

I) The quality of the studies

Studies can be limited by a lack of funding or by only being able to study small groups of individuals. This can limit their quality.

For example, some studies are reports of what happened to one person (case reports). These usually include a description of the individual and the signs, diagnosis, treatment, and medical outcome of the illness in question. Case reports are very useful in pointing out new illnesses or adverse effects of treatments, help to generate new ways of thinking, and may be necessary for rare diseases. They are, however, considered the least reliable form of evidence.

J) The north-south research divide

“Northern” countries are mostly in Europe, North America, East Asia, and Australasia. “Southern” countries are mostly in the rest of Asia, Africa, and Latin America.

Northern countries dominate research. This has negative consequences:

  • The pool of knowledge we can draw from is more limited.
  • The perspectives are narrowed.
  • Healthcare has less research to draw upon.
  • Research neglects certain diseases and populations.

Northern countries also have lower breastfeeding rates, which may magnify the above effects.


Bigman G. The Relationship of Breastfeeding and Infant Eczema: The Role of Reverse Causation. Breastfeed Med. 2019 Dec 19;10.1089/bfm. 2019.0269
Cebolla-Boado  H, Jiménez-Buedo  M, Salazar L. Avoiding selection bias without random assignment?  The effect of breastfeeding on cognitive outcomes in China.  Soc  Sci Med. 2017 Oct 16;194:151-159
Chartres N, Fabbri A, Bero LA. Association of Industry Sponsorship With Outcomes of Nutrition Studies: A Systematic Review and Meta-analysis. JAMA Intern Med. 2016 Dec 1;176(12):1769-177 
Godlee F, Cook S, Coombes R, et al. Calling time on formula milk adverts. BMJ. 2019 Mar 17;364:l1200
Huang J, Vaughn MG, Kremer KP. Breastfeeding and child development outcomes: an investigation of the nurturing hypothesis.  Matern Child  Nutr.  2016 Oct:12(4):757-67
Institute of Medicine (US) Committee on Conflict of Interest in Medical Research, Education, and Practice; Lo B, Field MJ, editors. Conflict of Interest in Medical Research, Education, and Practice. Washington (DC): National Academies Press (US); 2009

Kunc M, Dewji F, Jobson J, et al. G446(P) A systematic review evaluating the methodological quality of infant formula trials. Archives of Disease Childhood 2020;105:A161

Lamberti LM, Fischer Walker CL, Noiman A, et al. Breastfeeding and the risk for diarrhea morbidity and mortality. BMC Public Health 11, S15 (2011)
Lodge CJ, Lowe AJ, Dharmage SC. Is reverse causation responsible for the link between duration of breastfeeding and childhood asthma? Am J Respir Crit Care Med. 2008 Nov 1;178(9):994; author reply 995
Lundh A, Lexchin J, Mintzes B, et al. Industry sponsorship and research outcome: systematic review with meta-analysis. Intensive Care Med. 2018 Oct;44(10):1603-1612
Penders B, Wolters A, Feskens EF, et al. Capable and credible? Challenging nutrition science. Eur J Nutr. 2017 Sep;56(6):2009-2012
Raissian KM, Su JH. The best of intentions: Prenatal breastfeeding intentions and infant health. SSM Popul Health. 2018 Jun 18;5:86-100
Vandenplas Y, Alarcon P, Fleischer D, et al. Should Partial Hydrolysates Be Used as Starter Infant Formula? A Working Group Consensus. J Pediatr Gastroenterol Nutr. 2016 Jan;62(1):22-35