All topics › Feeding breastfeeding formula
What does the evidence say about long-term differences between breastfeeding and formula feeding?
Does breastfeeding (versus formula feeding) produce meaningfully better health outcomes for the child — especially in the long run — and for the mother? This is one of the most emotionally loaded questions in parenting, so this topic sticks strictly to what the data show, and is equally strict about what they don't.
Breastfeeding has real, well-supported benefits in the first year: fewer tummy bugs, fewer chest infections severe enough to need hospital care, and fewer ear infections [2, 26, 10]. For premature babies, human milk substantially reduces the risk of necrotizing enterocolitis, a dangerous gut condition [8]. What the evidence does not clearly support is that breastfeeding makes children smarter, thinner, or allergy-free in the long term: when researchers compare siblings — which washes out differences in income, education, and home environment — most of those long-term advantages shrink to nothing [5], and the one randomized trial's early cognitive advantage had largely faded by age 16 [27]. The honest picture is strong short-term protection, modest benefits for the mother, and much weaker evidence for the big long-term claims.
This field has a built-in bias problem. In high-income countries, mothers who breastfeed tend to be wealthier, more educated, less likely to smoke, and have better access to healthcare [1, 4]. Those same factors independently produce healthier, higher-achieving children. Most breastfeeding studies compare different families, so they can't fully separate the milk from the circumstances. Keep this in mind for every observational finding below.
There are two designs that get around this. The first is PROBIT, the largest randomized trial ever run on breastfeeding: 17,046 mothers across 31 Belarusian hospitals, where hospitals (not mothers) were randomized to a breastfeeding-promotion program or usual care [3]. The second is sibling-comparison studies, which compare siblings fed differently within the same family, holding family background constant [5]. Where these two designs agree, we can be fairly confident. Where they disagree with ordinary observational studies, believe the better design.
Stomach bugs. A large US evidence review of cohort studies found breastfed babies had a lower risk of gastrointestinal infection, including in analyses adjusted for family background — but the adjusted studies couldn't be combined into a single pooled estimate, so the exact size of the effect is uncertain [2, 26]. PROBIT, the randomized trial, also found fewer GI infections in the first year of life in the breastfeeding-promotion group [3].
Chest infections. The same review found that exclusive breastfeeding for at least 4 months was associated with fewer hospitalizations for lower respiratory tract infections in the first year, in adjusted observational studies — again with the exact size uncertain because of confounding [2, 26].
Ear infections. A 2015 meta-analysis found breastfeeding protected against acute ear infections in the first two years, with the strongest effect from exclusive breastfeeding for 6 months [10]. The exact size is hard to pin down — baseline ear-infection rates vary widely between populations — but the direction is consistent across studies [10, 26].
Rating: B — moderate. Consistent across many observational studies and supported by the randomized PROBIT trial for GI infections; the exact size of the effect is uncertain because of confounding.
Necrotizing enterocolitis (NEC) is a serious intestinal disease that mainly affects very premature babies. A Cochrane review of 8 randomized trials (1,605 preterm or low-birthweight infants) found that formula feeding, compared with donor human milk, raised the risk of NEC (risk ratio 1.87) with an absolute risk difference of 3 in 100 — meaning roughly 1 fewer case of NEC for every 33 preterm babies fed donor milk instead of formula [8]. This is moderate-quality evidence from randomized trials, and it applies specifically to preterm infants, not full-term babies [8].
Rating: B — moderate (for preterm infants; not generalizable to full-term babies).
A 2011 meta-analysis of 18 case-control studies found that any breastfeeding was associated with roughly half the odds of sudden infant death syndrome (adjusted odds ratio 0.55), with an even stronger association for exclusive breastfeeding [7]. SIDS is rare, so even a real halving would mean a small absolute difference — and these are case-control studies, so residual confounding is possible; some well-adjusted studies found no effect [7]. The association persisted after adjusting for smoking and socioeconomic factors [7].
Rating: B — moderate, downgraded because the evidence is observational (case-control) with some conflicting studies.
PROBIT's follow-up at age 16 found the breastfeeding-promotion group had less flexural eczema — 3 in 1,000 versus 7 in 1,000 in the control group [4]. But the same trial found no reduction in asthma at age 6 or age 16, and no effect on lung function [3, 4]. The Lancet's 2016 series likewise concluded there was no association between breastfeeding and asthma, blood pressure, or cholesterol [1].
Rating: A — strong for the trial-based findings (eczema reduced; asthma unaffected).
Breastfeeding is usually framed as a baby decision. The evidence on parents is thinner and more conflicted than the evidence on babies — here's what exists, with the confounds stated up front.
A warning that applies to everything below. Women who breastfeed differ from those who don't: they tend to be wealthier, better educated, and less likely to smoke, drink heavily, or be obese. They are healthier before they breastfeed. Every study below compares these groups, so the apparent benefits of breastfeeding partly reflect who breastfeeds, not the milk itself. Nothing below is from a randomized trial of breastfeeding versus formula; there is no strong-design evidence here at all.
Breast cancer. A huge pooled analysis of 47 studies in 30 countries found each 12 months of lifetime breastfeeding linked to about 4.3% lower breast cancer risk [11]. In real terms: about 1 in 7 UK women (~143 in 1,000) develop breast cancer in their lifetime [12], so 12 cumulative months of breastfeeding is associated with about 6 fewer cases per 1,000 [11, 12]. But the healthy-user bias applies in full force here.
Ovarian cancer. The largest study to date (nearly 10,000 women with ovarian cancer from 13 studies) found ever-breastfeeding linked to about 24% lower risk, and even a few months seemed to help [14]. In the UK about 1 in 50 women (~20 in 1,000) develop ovarian cancer [13], so that relative reduction is roughly 5 fewer cases per 1,000. A separate pooled analysis of 40 studies found a similar-sized association [15].
Other long-term disease links. Observational studies also report lower rates of type 2 diabetes and cardiovascular disease in women who breastfed, but the estimates are relative-only, vary between studies, and are likely confounded [1]. There is no verified absolute framing for these, so this topic doesn't quote them.
Fertility. Exclusive breastfeeding usually delays the return of periods — and with perfect use in the first 6 months (exclusive breastfeeding, no periods, baby under 6 months), the pregnancy risk is about 0.7–2% [16, 17]. As contraception goes, that's effective but not foolproof, and "perfect use" is doing a lot of work.
Mood. The relationship between feeding and maternal mood is mostly about expectations, not milk: mothers who intended to breastfeed but didn't reported worse well-being than mothers who intended and did — the mismatch, not the method, predicted how they felt [18].
Sleep, weight, bones. Studies comparing maternal sleep by feeding method are mixed and all observational — roughly similar total sleep, with breastfeeding mothers sometimes waking more or falling asleep faster [19, 20, 21]. Mothers who breastfeed retain slightly less pregnancy weight, but they also started leaner — modest and confounded [22, 23]. Bone density drops during lactation but recovers after weaning, with no difference by midlife [24, 25].
Partners. Almost nothing is known about how feeding method affects fathers or partners — no good studies found.
Overall rating: C for maternal outcomes — the long-term disease associations are all observational with likely healthy-user confounding; the sleep, weight, mood, and bone findings are modest and mixed.
This is the claim most parents have heard: breastfed babies end up smarter. Here's what the best studies actually show.
PROBIT — the randomized trial — did find higher cognitive scores: children in the breastfeeding-promotion hospitals scored about 7.5 points higher on verbal IQ at age 6.5 [3]. But the trial randomized hospitals to a promotion program, not babies to feeding methods, and the groups may have differed in ways the analysis couldn't fully capture (maternal IQ, for instance, was not directly measured). And by age 16, the picture had changed: the overall neurocognitive benefit was gone, with only a much smaller advantage in verbal function remaining — small compared with other family and birth factors [27].
Sibling studies tell a different story. Colen and Ramey (2014) compared siblings fed differently within the same US families and found the apparent benefits for math, reading, vocabulary, and intelligence shrank by 69–92% and became statistically indistinguishable from zero [5]. Der, Batty, and Deary (2006) similarly found the breastfeeding–IQ link disappeared once maternal IQ was properly accounted for [6].
Rating: C — weak. One cluster-randomized trial with a fading cognitive signal, multiple sibling studies null. It is genuinely unclear whether breastfeeding itself raises IQ, or whether smarter, better-resourced mothers are simply more likely to breastfeed.
A WHO-commissioned 2015 meta-analysis found breastfed children had about 26% lower odds of overweight and obesity [9]. But the authors themselves flagged likely residual confounding and publication bias. And again, sibling studies — including Colen and Ramey (2014) — find no effect on BMI or obesity once family background is held constant [5].
Rating: C — weak. The observational association exists, but the best confounding-controlled studies find nothing.
Putting the numbers together for a full-term baby in a high-income country:
| Outcome | Breastfed | Formula-fed | Difference | Rating |
|---|---|---|---|---|
| GI infections, first year | Fewer (also seen in PROBIT) | — | Size uncertain; no defensible pooled absolute | B |
| Hospitalized for chest infection, first year | Fewer | — | Size uncertain; no defensible absolute | B |
| Ear infections, first 2 years | Fewer | — | Size uncertain (baselines vary widely) | B |
| SIDS | Associated with ~half the odds | — | Small absolute (rare event); observational | B |
| Eczema at 16 (PROBIT) | 3 in 1,000 | 7 in 1,000 | 4 fewer in 1,000 | A |
| NEC (preterm babies only) | — | — | 1 fewer case per 33 preterm babies fed donor milk vs formula | B |
| Asthma | no difference found | no difference found | 0 | A |
| IQ / obesity / diabetes | unclear | unclear | not established | C |
And the costs of breastfeeding, which are real and rarely quantified in studies: pain and difficulty in the early weeks, major time demands, disrupted sleep, incompatibility with some work situations, and the fact that only one parent can do it. These trade-offs are legitimate inputs to the decision, not failures of willpower.
Formula, for its part, is a regulated product: UK-regulated first infant formula is designed to meet babies' nutritional requirements [28]. There is no good evidence that formula feeding harms growth or development; the evidence simply shows breastfeeding adds protection against certain infections on top of an adequate baseline [1, 2] — and safe preparation matters more than which brand you buy.