Monday March 14 2011
Breastfeeding improved grades, but the effect was small
“Breastfeeding produces not only healthier babies but also brighter children,” reported The Guardian. It said a study has found that just four weeks of breastfeeding gives newborns a "positive and significant effect", which lasts up to secondary school and beyond.
The news story is based on a report published by the Institute for Social & Economic Research. The report has not yet been published in a peer-reviewed medical journal. The study used data from more than 12,000 children some of whom were breastfed and some of whom were not. The cognitive outcomes between the groups were compared based on their Standard Attainment Tests (SATs) scores at ages 7, 11 and 14, and their school entry test scores at the age of 5.
This is not a new area of research, but the authors have used a statistical method that should disentangle some of the complex relationships between the many different factors that can influence cognition (the ability to think, reason and plan). The study has some shortcomings, including the fact that it was not possible to take into account every factor that can affect cognition and school performance. However, it does demonstrate that breastfeeding has a beneficial effect, albeit a small one.
These findings do not change advice to mothers to breastfeed their child if they can, and makes no recommendations about the duration of breastfeeding or whether this should be exclusive or not. More information about breastfeeding can be found on our breastfeeding pages.
Where did the story come from?
The study was carried out by researchers from the University of Essex and the University of Oxford. Funding was provided by the Economic & Social Research Council (ESRC). The research was published by the Institute for Social & Economic Research in December 2010.
The newspapers have covered the research accurately. However, it is worth mentioning that the effect is small and equates to only about 3% improvement on scores in maths, English and science tests between children who were breastfed for at least four weeks compared with those who were not.
What kind of research was this?
This study used data from a larger study called the Avon Longitudinal Survey of Parents and Children (ALSPAC), which is funded by the UK Medical Research Council, the Wellcome Trust and the University of Bristol. ALSPAC is a large, long-term study following 12,000 children born in the Avon area in England in the early 1990s. Data on the children’s health, development and other factors is collected periodically. The study recruited the mothers of these children when they first notified their doctors that they were pregnant in 1991 and 1992.
In this study, researchers were interested in evaluating the link between breastfeeding and the children’s SATs scores at different ages, after making adjustments for other factors that may affect the relationship. This is a popular area of research and other studies have looked at this question. However, this study uses a particular type of statistical analysis that gives a better idea of the size of the effect that breastfeeding may have.
What did the research involve?
The researchers aimed to assess the effects of breastfeeding on children’s results from Standard Attainment Tests (SATs) at ages 7, 11 and 14 and school entry test scores at age 5.
The data were obtained from a large prospective cohort study and included information on physical and mental health, child development and socioeconomic status. Data were collected several times from birth directly from the 12,000 children who were enrolled and from their parents and teachers. Detailed information was also collected from parents about their attitudes to breastfeeding and how they were feeding their child. This enabled a calculation of the duration of exclusive breastfeeding and the duration of total breastfeeding (which included the time when breastfeeding overlapped with the introduction of solid foods).
Several studies have examined the link between breastfeeding and children’s outcomes, but many of them suffer from the same inherent difficulties and cannot show causality. There are complex relationships between maternal characteristics, the decision to breastfeed, the duration of breastfeeding and children’s outcomes. The authors say that regression analyses, which are the standard way of analysing data from cohort studies, cannot clarify these relationships perfectly.
To deal with this, the researchers applied a type of analysis called propensity score matching. This is a statistical method that can reverse engineer the data from a cohort study in such a way that the methods better approximate those used in a randomised controlled trial. Using this, the authors were able to report whether there were any differences in cognitive ability between children who were breastfed and those who were not, after the groups had been matched for several potential confounding variables (characteristics that may have influenced the relationship between breastfeeding and intelligence).
To do the propensity score matching the researchers needed three things:
- A large data sample. They had data from around 12,000 children born in the Avon area in the early 1990s. They excluded multiple births and did not have birth data for only 69 of the pregnancies.
- Data on SATs score results. Four Local Education Authorities in the former Avon area (Bristol, South Gloucestershire, Bath and North East Somerset, and North Somerset) used the same assessment scheme in 80% of the local state schools.
- Data on breastfeeding outcomes. From the data available, the researchers were able to determine the duration of exclusive breastfeeding and the duration of total breastfeeding, including mixed feeding.
Using these data, the researchers calculated the chance (probability) that each child had of being breastfed, based on a typical regression analysis. These children were then separated into two groups and matched so that those with similar background factors were grouped together. A large number of these background factors were matched for, including the child’s sex, the baby’s birth weight, mode of delivery (vaginal or caesarean section), the mother’s age, the parents’ marital status, the education levels of both parents, housing tenure, the size of the home, neighbourhood characteristics, features of the mother’s or father’s health and future labour market intentions. They were also matched for other aspects of parenting, including how frequently the parents read to the child, smacked the child or shouted at the child.
What were the basic results?
There were significant differences in average scores on maths, English and science tests between children who were breastfed and those who were not. The majority of this difference could be explained by differences between the mothers (such as the mother’s education or socioeconomic class), as in previous studies. However, after the matching that adjusted for potential biases in a number of maternal/paternal and social factors took place, there was still a significant beneficial effect of breastfeeding on school test scores. This effect persisted at least until the children were aged 14, specifically in terms of English, maths and science ability.
The effects were not large. For example, at the age of 14 (key stage three level) the average scores on maths and science tests for breastfed children were about 3% higher than those of the non-breastfed children (about 0.1 standard deviations).
How did the researchers interpret the results?
The researchers conclude that their study provides further evidence of the causal effect of breastfeeding on children’s cognition. They say they have also established the duration of the effect on children and whether some mothers and babies may benefit more than other groups.
As the authors discuss, breastfeeding is far more likely in women who are better off socioeconomically. For this reason, previous studies have found it difficult to conclude robustly that breastfeeding is the cause of improved outcomes in a child. In this study, the researchers analysed their data using standard accepted methods, but also applied a statistical technique (known as propensity score matching), which is another way of addressing the issue of confounding in cohort studies. The technique involves statistically pairing breastfed babies with those who were not breastfed, but who were similar in respect of many other factors. The technique effectively simulates an experiment because it creates two groups that are matched on all possible measurable factors except for the exposure of interest, in this case breastfeeding.
The study has achieved its aims, which were to “disentangle the effects of breastfeeding from the effects of mother’s characteristics” and other difficult to measure factors on children’s cognitive outcomes. The findings after the analyses were that children who were breastfed for four weeks or more did better at school (on the tests measured in this study) than children breastfed for fewer than four weeks or not at all. It is important to point out, however, that these differences were quite small.
Also, while the study goes some way to disentangling these complex effects, it does not account for everything that could affect cognition and school performance. While propensity score matching can balance groups on known and measured factors, there are still potentially unobserved differences between the groups that were not taken into account. The researchers cite maternal IQ as one example of such a factor.
These findings do not change advice to mothers to breastfeed their child if they can, and makes no recommendations about the duration of breastfeeding or whether this should be exclusive or not. More information can be found on our breastfeeding pages.