Background: Several observational studies have linked breastfeeding with intelligence in children. However, these studies are limited by potential confounding variables, particularly those related to socioeconomic status, maternal education, and degree of stimulation provided to the child. Der and colleagues examined the role of maternal intelligence as a confounder in links between breastfeeding and childhood intelligence.

The Study: The researchers used data from the U.S. National Longitudinal Survey of Youth, which has followed a population-based sample of 12,686 persons who were 14 to 22 years of age in 1979. Those in the sample were surveyed annually until 1994 and biennially afterward; children born to the more than 6,000 women in the study have been assessed every two years since 1986.

The study excluded children who were born before 35 weeks' gestation or had birth weights of less than 5.5 lb (2,500 g). Data gathered for eligible children included whether and for how long they were breastfed; the mother's smoking status during pregnancy; maternal age at the birth of each child; and maternal educational level, poverty status, and race. Demographic data such as sex, age, birth order, birth weight, and gestational age at birth also were gathered for each child.

Maternal cognitive ability was assessed as an IQ using the standard Armed Forces Qualification Test (AFQT). Information about the home environment was gathered from the cognitive stimulation and emotional support subscales of the short form of the Home Observation for Measurement of T the Environment scale (HOME-SF). Children five to 14 years of age were tested with the Peabody Individual Achievement Test (PIAT), providing subscores in mathematics, reading comprehension, and reading recognition as well as total scores.

The researchers also examined the literature on breastfeeding and children's intelligence to clarify significant variables in the differences between mothers who breastfed and those who did not, as well as in the reported relationship between breastfeeding and intelligence in children. For this meta-analysis, they considered 73 studies with relevant original data.

The statistical models used included a comparison of 332 pairs of siblings of which one child was breastfed and one was not.

Results: Data were examined on 5,475 children born to 3,161 mothers. Children who were breastfed had mothers who had higher intelligence and greater education and who provided a more stimulating home environment. These mothers were also more likely to be older, white, and nonsmokers. The children were more likely to have heavier birth weights and higher birth orders than children who were not breastfed.

Maternal IQ was significantly associated with breastfeeding. A rise of one standard deviation in maternal IQ more than doubled the likelihood of breastfeeding. Maternal education level had a smaller but similar effect.

In the unadjusted analysis, breastfeeding had a highly significant positive effect on children's total PIAT scores and all subscores. Adjusting for maternal IQ reduced this effect by 71 to 75 percent, and adjusting for maternal education reduced the effect by 34 to 42 percent. No evidence of a dose-response relationship was found for duration of breast-feeding. Analysis of the data from the discordant sibling pairs showed no significant differences in intelligence scores between breastfed and non-breastfed siblings.

In the mutually adjusted analysis, the most powerful relationship with childhood intelligence was maternal AFQT score (see accompanying table). Maternal education, age, and cognitive stimulation at home had weaker positive effects. Later birth order and poverty were among the negative effects.

Conclusion: The authors conclude that breastfeeding has little or no direct effect on childhood intelligence. The intelligence of the mother appears to be a major determining factor on the intelligence of the children, whether or not they are breastfed. Nevertheless, the authors stress that there are many other advantages of breastfeeding for children and caution that their results may not be generalizable to premature infants or to children in developing countries.