The strongest breastfeeding evidence is more modest than the slogan suggests

Breastfeeding has some real, measurable benefits, but the strongest evidence shows them to be narrower and more modest than the phrase "breast is best" implies. The one large trial that could randomise the question found fewer gut infections and less eczema in infancy but no significant reduction in respiratory infection, and sibling-controlled studies find that most of the sweeping long-term advantages seen in ordinary observational research shrink or disappear once family background is held constant [s1] [s3].

This is an emotionally loaded subject, and the point of laying out the evidence is not to steer anyone toward or away from a feeding choice. It is that the confidence with which large lifelong benefits are asserted outruns what the best-designed studies actually show.

Why ordinary studies are hard to trust here

Almost all evidence that breastfeeding improves health comes from observational studies, and the PROBIT investigators state the problem plainly: mothers who breastfeed differ systematically from those who do not, and "potential sources of bias in such studies have led to doubts about the magnitude of these health benefits in industrialized countries" [s1]. In wealthy countries, breastfeeding is strongly patterned by income, education, and other advantages that independently improve children's outcomes. A study that compares breastfed with formula-fed children is therefore partly comparing more-advantaged families with less-advantaged ones, and the resulting "benefit" is inflated by everything else that travels with the choice.

Two research designs try to break that confounding: a randomised trial, and comparisons between siblings in the same family who were fed differently.

The randomised trial: real but selective benefits

The Promotion of Breastfeeding Intervention Trial (PROBIT) is the largest randomised study in the field. Conducted in Belarus with 17,046 mother-infant pairs, it randomly assigned maternity hospitals — not individual mothers — to either a breastfeeding-promotion programme modelled on the WHO/UNICEF Baby-Friendly Hospital Initiative or to usual care [s1]. The intervention worked at changing behaviour: exclusive breastfeeding at three months was 43.3% in the intervention group versus 6.4% in the control group (P<.001) [s1].

The health results were mixed. Infants at the intervention sites had a significant reduction in gastrointestinal infection (9.1% vs 13.2%; adjusted odds ratio, 0.60; 95% CI, 0.40 to 0.91) and in atopic eczema (3.3% vs 6.3%; adjusted OR, 0.54; 95% CI, 0.31 to 0.95) during the first year [s1]. But there was no significant reduction in respiratory infection — the rates were essentially identical, 39.2% in the intervention group and 39.4% in the control group (adjusted OR, 0.87; 95% CI, 0.59 to 1.28) [s1].

An important caveat is baked into the design: PROBIT randomised a promotion programme, not breastfeeding itself, so its effects reflect the difference more breastfeeding made on top of usual care, not a clean comparison of fed-this-way versus fed-that-way [s1].

The cognition question

PROBIT's most-discussed follow-up looked at intelligence at age 6.5 years, assessing 13,889 of the original children (81.5%) [s2]. The intervention group scored higher on all subtests, but the precision varied: the cluster-adjusted mean difference was +7.5 points for verbal IQ (95% CI, +0.8 to +14.3), which excludes zero, but +2.9 for performance IQ (95% CI, -3.3 to +9.1) and +5.9 for full-scale IQ (95% CI, -1.0 to +12.8), both of whose intervals cross zero [s2]. Teachers rated the intervention children higher in reading and writing [s2].

The authors read this as strong evidence for a cognitive benefit [s2]. A careful reader will note the narrower claim the numbers actually support: a statistically significant verbal-IQ difference in a single large cluster-randomised trial, with the full-scale estimate not reaching significance, and the usual difficulty that teachers and testers in a promotion trial are hard to keep fully blind to a village's breastfeeding culture.

The sibling studies: most of the long-term gap is selection

The most direct test of confounding compares siblings. A 2014 study in Social Science & Medicine used 25 years of US panel data and compared brothers and sisters within the same family who had been fed differently as infants [s3]. In standard regressions — the kind that produce the familiar large benefits — breastfed children aged 4 to 14 did significantly better on 10 of 11 outcomes studied, from BMI to behaviour to academic measures [s3].

Then the authors restricted the comparison to siblings and added within-family fixed effects, which cancels out everything the siblings share: household income, parenting, neighbourhood, genes. Once that was done, the estimated association between breastfeeding and all but one of the outcomes "dramatically decrease[d] and fail[ed] to maintain statistical significance" [s3]. Their conclusion is that much of the long-term benefit typically attributed to breastfeeding may instead reflect who breastfeeds — selection along lines of race and socioeconomic status — rather than the feeding itself [s3].

What this adds up to

The defensible summary is layered. Breastfeeding has real benefits that survive strong study designs, concentrated in infancy: fewer gastrointestinal infections and less eczema in the first year, from the one large trial that could test it [s1]. A cognitive edge is plausible but rests on a single trial with a significant verbal-IQ result and non-significant full-scale estimate [s2]. And the large, lifelong advantages routinely cited — for obesity, behaviour, long-term achievement — largely dissolve when siblings are compared, implying they were mostly confounding all along [s3].

None of that makes breastfeeding unimportant or formula harmful; it makes the honest benefit smaller and more specific than the messaging. Feeding decisions depend on a family's health, circumstances, and support, and the evidence above is not a verdict on any individual choice.

This article is informational and is not medical advice. Questions about infant feeding for a specific child are best discussed with a clinician or a qualified lactation professional.