There is a 2007 article in PNAS (Proceedings of the National Academy of Sciences), “Moderation of breastfeeding effects on the IQ by genetic variation in fatty acid metabolism,” by Avshalom Caspi et al., claiming to have discovered a genetic variation in the way that breastfeeding effects IQ. However, this is in complete contradiction with a 2006 article published in BMJ (British Medical Journal), “Effect of breast feeding on intelligence in children: prospective study, sibling pairs analysis, and meta-analysis,” by Geoff Der, G. David Batty, and Ian J. Deary, claiming that breastfeeding has no effect on IQ after other confounding factors are accounted for.
The results are completely contradictory. What’s going on here?
Der, Batty and Deary study
The Der, Batty and Deary study uses data from the 1979 National Longitudinal Survey of Youth. It looks at the children of participants in that study, who were tested between the years of 1986 and 2002, so these children were fed formula before supplementation of infant formula with DHA and ARA fatty acids (which are supposed to better duplicate the effect of human breast milk) were available in the United States.
The Der, Batty and Deary study notes the major confounding factor in breast-feeding analyses is that women who breastfeed are usually of higher intelligence and social class than those who don’t breastfeed. In the NLSY79 data, breastfeeding mothers scored about one standard deviation higher on the Armed Forces Qualifying Test (AFQT).
A multiple regression analysis of the NLSY79 data, which includes 5475 children, shows that breast feeding has no statistically significant impact on children’s cognitive assessments. Factors having the strongest correlations with children’s cognitive assessments are the mother’s AFQT score, the HOME cognitive stimulation score (which is probably a good surrogate for the father’s IQ and a host of other socioeconomic factors), and strangely enough, birth order.
The study also did an analysis of 332 sibling pairs where one was breastfed and the other was not. The results show that the formula-fed sibling had an IQ that was 0.63 points higher than the breastfed sibling. The authors didn’t bother to point out that this result was the opposite of the expected pattern. The P value is 0.506, so the authors probably figured this is just a statistically insignificant fluke and not worth mentioning.
There is additional evidence in the study, beyond what I’ve written about here, confirming the conclusion. Based on the Der, Batty and Deary study, I am convinced that the hypothesis that breast feeding causes higher IQ has been proven false. The cause and effect go the other way. Breast feeding predicts having parents with higher IQs and higher socioeconomic status.
The Caspi study claims that breast feeding increases IQ by a whopping 6.8 points (with IQ having a standard deviation of 15) when the child has one or two copies of the C allele of the rs174575 SNP of the FADS2 gene.
Such a large IQ increase attributed to breast feeding is an impossible result given the conclusion of the Der, Batty and Deary study, unless you think that Der, Batty and Deary falsified data.
Looking at Table 1 of the Caspi study, which is apparently what the conclusion of the study is based on, it’s apparent that the Caspi study did not properly take into account confounding factors. The Caspi study does not contain a regression analysis like the Der, Batty and Deary study, nor does it include a sibling analysis. Thus, the Caspi study is useless crap.
I’ve created the following four tables from the data in Table 1 of the Caspi study:
|New Zealand cohort, not breastfed|
|Children's IQ||98.4 (15.2)||95.8 (12.4)||100.3 (11.2)|
|Maternal IQ||97.1 (15.2)||96.0 (12.9)||100.2 (13.4)|
|New Zealand cohort, breastfed|
|Children's IQ||103.2 (13.9)||104.0 (13.4)||98.9 (13.8)|
|Maternal IQ||102.4 (15.2)||103.6 (14.2)||103.3 (14.9)|
|British cohort, not breastfed|
|Children's IQ||97.3 (14.1)||97.2 (13.9)||99.9 (15.3)|
|Maternal IQ||95.1 (14.9)||98.5 (12.7)||91.8 (14.5)|
|British cohort, breastfed|
|Children's IQ||104.0 (15.0)||104.6 (15.3)||100.7 (17.3)|
|Maternal IQ||105.0 (12.9)||105.2 (13.8)||102.6 (14.4)|
“Entries in the table are means and standard deviations. IQ scores were standardized to M = 100 and SD = 15 in each cohort.”
For children with the CC and CG alleles (91.7% of the samples), we observe results completely consistent with the conclusion of the Der, Batty and Deary study, which is that parental IQ and other factors determine the child’s IQ, and not breastfeeding.
Considering only the 91.7% children with CC and CG alleles, I added the numbers together to get the following results:
|New Zealand cohort|
Note that the differences in the tables above may appear off by 0.1 because of rounding error.
The British cohort, which is the larger sample, shows the average breastfed child has an IQ slightly lower than the average breastfeeding mother. The average formula-fed child has an IQ slightly higher than the average formula-feeding mother. These results are as expected. Maternal IQ predicts the child's IQ, plus there is a small amount of regression towards the mean.
In the New Zealand cohort, both breastfed and non-breastfed children have an average IQ that’s 0.6 higher than their mothers’ IQ. I don’t know why there’s no regression towards the mean in this sample. But the sample size is smaller, and the IQ difference between breastfeeding and non-breastfeeding mothers is also smaller.
The weird results occur with the 8.7 % of children with the GG alleles.
In the British cohort, non-breastfed GG children have an average IQ that’s 7.4 points higher than predicted by maternal IQ. Breastfed GG children have an average IQ that’s 3.6 points lower than predicted by maternal IQ. Thus there’s a net 10.9 IQ point benefit from formula feeding over breastfeeding.
In the New Zealand cohort, non-breastfed GG children have an average IQ that’s 0.5 points lower than predicted by maternal IQ, and breastfed GG children have an average IQ that’s 5.0 points lower than predicted by maternal IQ. Thus there’s a net 4.5 IQ point benefit from formula feeding over breastfeeding.
If we average the British and New Zealand results together, taking into account the larger British sample size, there seems to be an 8.3 IQ point benefit to formula feeding when a child has the GG alleles.
This result seems consistent with the finding of the Der, Batty and Deary study where breastfeeding resulted in a 0.63 IQ point boost when sibling pairs were looked at. 8.3% of 8.3 is 0.69. The numbers are pretty close.
The Caspi study is a completely crap study, and the people involved with it ought to be ashamed of themselves for promoting junk science, and for scaring mothers into thinking that they are harming their children by not breastfeeding them. And everyone who was involved with publishing the study ought to be ashamed of themselves. There should be accountability. These hacks should be fired.
The mystery of why formula feeding seems to boost IQ when a child has GG alleles obviously needs further investigation. The medical community talks about the alleged benefits of breastfeeding, but it seems that once again the medical community has it all backwards. Based on this data, formula seems like a better idea than breastfeeding.
Why do people want to think that formula is bad for infants? It obviously has to do with the notion we see quite often, especially from the political left, that technology is bad and nature is good. Thus natural breast milk has to be better than formula made in a factory. Furthermore, feeding your baby formula is too easy. Everyone knows that better results require effort. People want to believe that they have more control over their baby’s welfare than they actually have. But it turns out that when it comes to formula vs. breastfeeding, the easy way is actually the best way.
More junk science debunked by Half Sigma.
If you would like an e-copy of both of these studies, send me an email.