The aim of this study was to examine the relationship between duration of breastfeeding and educational outcomes. We hypothesized that longer periods of breastfeeding would predict better educational outcomes in middle childhood.
The Western Australian Pregnancy Cohort (Raine) Study used a cohort of 2900 women who were enrolled at 18 weeks' gestation; with 2868 live-born children were followed prospectively. At ∼10 years of age, data from 1038 children were linked to standardized mathematics, reading, writing, and spelling scores. Associations between breastfeeding duration and educational outcomes were estimated by using linear models with adjustment for gender, family income, maternal factors, and early stimulation at home through reading.
Ten-year-old children who were predominantly breastfed for 6 months or longer in infancy had higher academic scores than children who were breastfed for less than 6 months. The effect of breastfeeding on educational outcomes differed according to gender; boys were particularly responsive (in mathematics, spelling, reading, and writing) to a longer duration of breastfeeding.
Predominant breastfeeding for 6 months or longer was positively associated with academic achievement in children at 10 years of age. However, the effectiveness of breastfeeding differed according to gender; the benefits were only evident for boys.
Comments
Breastfeeding on Child Cognitive Development: Still a Long Way to Go
The effect of breastfeeding on cognitive development has been long debated. A recent article by Wendy et al. (1) reported that 10-year-old children who were predominantly breastfed for 6 months or longer scored higher on standardized tests, and boys in particular, do better than girls. However, their study has several serious methodological issues that affect the validity of the study findings. I comment in more detail below.
First, the study used a highly selective subsample (n=980) of the original cohort of 2,868 live births in the Raine study. The study sample is restricted to those who attended government schools and whose parents gave consent to link their child's data to government education data- bases. As the authors themselves described, the study participants are more socioeconomically advantaged as compared to nonparticipants. Therefore, the higher academic achievement found in this sample may quite likely be due to the advantaged socio-economic status, not breastfeeding.
The authors addressed the above limitation by indicating that the original cohort includes an oversample of the socially disadvantaged women, and dropping some of them from their analyses may improve the generalizability of the study findings. Such a statement is unscientific and, in fact, irresponsible. It may mislead other researchers to overcome their sampling problems by selectively dropping out part of the sample to make it "representative". Sample weighting provides an appropriate remedy to the over-representation of some groups in a sample.
In addition, the analytical method used in the study also has serious problems. They used multivariate regression to assess the association of breastfeeding duration and the child's academic achievement 10 years later. There are at least three issues here.
One, association is not the same as causal effect. In the abstract and introduction, the authors mentioned that they want to assess the "association". However, in the discussion and conclusion, they changed the word to be "independent effect". Breastfeeding may well be associated with child's academic achievement because breastfeeding is known to be linked to a wide array of socioeconomic factors, which may have large effect on child's academic achievement. But encouraging breastfeeding does not necessarily improve a child's academic achievement. And the differential effect of breastfeeding by gender may well be due to the genetic difference of boys and girls on those academic achievements.
Second, the authors did try to isolate the biological effect of breastfeeding from the broader socioeconomic factors by adjusting the measured confounders using multivariate regression. But such a method may not be sufficient to establish causality. For example, regression in this context would assume linearity between family income and the child's academic achievement, as well as linearity between family income and breastfeeding duration. While neither may be the case, they do no check this assumption. If these assumptions are incorrect, their methodology may not fully adjust comparisons across breastfeeding categories for differences in family income.
Third, the authors acknowledged that they cannot adjust for maternal intelligence. However, maternal IQ is really the key in examining the effect of breastfeeding on child development, as two recent studies have indicated that maternal IQ and the home environment are better predictors of breastfeeding status than even poverty and education (2, 3).
Finally, in addition to statistical significance, one also needs to consider the practical importance of the findings. While not having access to the data directly, one can roughly calculate the effect size from Table 1 and 4. Take mathematics test score for example, the effect size for predominantly breastfeeding >=6 months is 15.79/106.1=0.15. Although the p value is 0.036, highly statistically significant, the effect size is less than one fifth of the standard deviation. Whether such an increase in test score is worth pursuing is entirely different from the statistical significance of the result.
The above result is consistent with the findings of our own study (4), in which we included maternal IQ and used propensity score matching to overcome some of the limitations the ordinary regression have. We found modest effect of breastfeeding on child cognitive outcomes.
In summary, the study does not demonstrate a convincingly positive effect of predominantly breastfeeding for 6 months or longer on educational achievement.
References
1. Oddy WH, Li J, Whitehouse AJO, Zubrick SR, Malacova E. Breastfeeding Duration and Academic Achievement at 10 Years. Pediatrics 2011;127(1):e137-145.
2. Der G, Batty GD, Deary IJ. Effect of breast feeding on intelligence in children: Prospec-tive study, sibling pairs analysis, and meta-analysis. British Medical Journal 2006;333(7575):945-950.
3. Jacobson SW, Chiodo LM, Jacobson JL. Breastfeeding effects on intelligence quotient in 4- and 11-year-old children. Pediatrics 1999;103:e71.
4. Jiang M, Foster EM, Gibson-Davis CM. Breastfeeding and the Child Cognitive Out-comes: A Propensity Score Matching Approach. Maternal and Child Health Journal 2010:1-12.
Conflict of Interest:
None declared