For healthcare professionals only


Breast feeding and obesity in childhood: cross sectional study

BMJ 2003; 327 doi: (Published 16 October 2003) Cite this as: BMJ 2003;327:904
  1. L Li, statistical research fellow (L.Li{at},
  2. T J Parsons, lecturer1,
  3. C Power, professor1
  1. 1Centre for Paediatric Epidemiology and Biostatistics, Institute of Child Health, London WC1N 1EH
  1. Correspondence to: L Li
  • Accepted 18 August 2003


The evidence that breast feeding protects against obesity is inconclusive: some studies show a protective effect1 and others find no effect.2 3 Confounding factors may account for these inconsistencies. We used data from the offspring of the 1958 British birth cohort to assess whether breast feeding influences body mass index and obesity in childhood.

Methods and results

We used data from a randomly selected sample (n = 2584) of the members of the 1958 British birth cohort who had children by 1991.4 Information was collected on their offspring. Of 3077 children aged 4-18 years, we included 2631 children (1293 girls and 1338 boys from 1768 families; average age 8 years) for whom data on duration of breast feeding, body mass index, and confounding factors were available.

Body mass index (weight (kg)/(height (m)2)) was standardised relative to the 1990 British growth reference,5 and obesity was defined as a standard deviation score > 1.64 (95th centile). Duration of breast feeding had been reported by the mother in 1991 (see table for categories).

Relation of duration of breast feeding to mean body mass index (BMI) and odds ratio for obesity* in children aged 4-18 years in 1991

View this table:

Potential confounding factors, reported in 1991, were birth weight; mother's smoking during pregnancy (< 1 cigarette/day, 1-9/day, or ≥ 10/day); and social class, based on the 1991 occupation of the male head of household, classified as professional or managerial, skilled non-manual, skilled manual, or semiskilled or non-skilled. Parent's body mass index was derived from height and weight and standardised within the study by sex. The body mass index of the parent (only one parent was the cohort member) was available for each child.

The multilevel models that we used to estimate the relation between body mass index and duration of breast feeding (linear model) and between obesity and duration of breast feeding (logistic model) took into account the correlation between siblings. We used the iterative generalised least squares procedure (MLwiN statistical package) to calculate outcomes for ages 4-8 and 9-18 separately. We calculated odds ratios before and after adjusting for sex, parent's body mass index, maternal smoking during pregnancy, birth weight, and social class.

In our sample, 62.9% of children (1655) had been breast fed for ≥ 1 week. The mean score for body mass index in all children was higher than the growth reference sample by 0.18. A total of 207 children (7.9%) were obese. Mean body mass index and obesity were consistently lower in those breast fed for 2-3 months, though not significantly (table). We found no evidence that breast feeding influenced body mass index or obesity and no dose dependent trend in either age group; adjustment for confounding factors did not alter these findings. That there was no difference in the relation between the age groups suggests that recall bias was not an important factor.


As in the 1958 birth cohort,3 results from their offspring provide no support for a protective effect of breast feeding on obesity. In studies reporting a protective effect, it is weak and not always supported by a dose-response relation, which might be expected, at least up to a threshold duration. Any effect of breast feeding may be limited to a critical period or depend on other cofactors. Secular trends do not suggest a protective effect: in both Britain and the United States the incidence of breast feeding has increased since 1990, but so has obesity. Promoting breast feeding is important, but evidence for an important beneficial effect on obesity is still equivocal.

Data were obtained from Centre for Longitudinal Studies, Institute of Education; National Child Development Study Composite File including selected perinatal data and sweeps one to five [computer file]; National Birthday Trust Fund, National Children's Bureau, City University Social Statistics Research Unit [original data producers]; The Data Archive [distributor], Colchester, Essex: SN:3148. 1994.


  • Contributors All authors designed the study and wrote the paper. LL did the data analysis and is guarantor.

  • Funding TJP holds a Medical Research Council special training fellowship in health services and health of the public research.

  • Conflict of interest None declared.

  • Ethical approval Not needed.


View Abstract