In the 1958 British birth cohort (n = 12 857 at age 7), breast feeding and BMI were unrelated in childhood. Breast feeding was protective against increased BMI at ages 16 and 33 years in females, and at 33 years in males, but this effect was markedly reduced and no longer significant after adjustment for confounding factors.
- body mass index
- breast feeding
- cohort study
Statistics from Altmetric.com
A possible causal link between method of infant feeding and subsequent obesity has been debated for a long time, stimulated by studies suggesting that bottle fed infants were fatter than breast fed infants. Several large, recent studies present arguments both for and against a protective effect of breast feeding on childhood obesity.1–3 To our knowledge only two previous studies have follow up data in adulthood, and suggest no effect,4 or even an adverse effect5 of breast feeding on adult obesity.
PATIENTS AND METHODS
We examined the relation between breast feeding and body mass index (BMI) in the 1958 British birth cohort, which includes all children born in England, Scotland, and Wales, 3–9 March 1958.6 From a target population of 17 733 births, information was obtained on 98%. Surviving children were measured at ages 7, 11, 16, and 33 years. At age 33 years, 11 407 subjects from a target sample of 15 667 provided information. We excluded multiple births (n = 446). Despite sample attrition, the remaining sample is generally representative of the original sample.6 To illustrate, comparing the sample with data at 33 years (n = 9287) with that at age 7 (n = 12 857), 45% and 44% respectively were breast fed for more than a month.
Body mass index (BMI, kg/m2) was calculated from heights and weights. Heights (to the nearest inch) and weights (in pounds) were measured in school by trained medical personnel at 7, 11, and 16 years.6 At 33 years, height was measured to the nearest centimetre, and weight to the nearest 0.1 kg. Pregnant women at 33 years (n=256) were excluded. Infant feeding was reported when the subject was 7 years, in four categories; breast fed wholly or partially for (i) >1 month, (ii) <1 month, (iii) not at all, and (iv) unknown. The latter group was small (0.9%), and excluded from analyses. Social class, maternal BMI, maternal smoking during pregnancy, parity, birth weight, and birth order were identified from the literature as potential confounding factors.
The association between breast feeding and BMI was investigated separately for each age and sex, using linear models with contrasts to test differences between breast feeding categories. The effect of potential confounding factors on the relation was examined for BMI at 33 years, and interactions between breast feeding and confounding variables were tested. Repeating analyses using sex specific standard deviation scores for BMI (since the variance of BMI changes with age) or after logarithmic transformation (to correct the slight positive skew in BMI distribution) did not change the findings.
Of those with BMI data at 7 years, the proportions breast fed for >1 month, <1 month, or not at all were respectively 43.5%, 24.5%, and 32.0% of males and 44.7%, 24.9%, and 30.4% of females. Breast feeding was unrelated to BMI in childhood, to age 16 years in males and 11 years in females (table 1). In adulthood, a dose-response relation was evident, with mean BMI increasing with decreasing duration of breast feeding. For example, compared with males breast fed for >1 month, those breast fed <1 month and those not breast fed at all had higher BMIs at 33 years, by 0.17 kg/m2 (p > 0.05) and 0.35 kg/m2 (p < 0.05) respectively. The relation showed a similar pattern with age if obesity (BMI ⩾30 kg/m2 in adulthood, BMI ⩾95th centile in childhood) was used as the outcome rather than BMI (data presented only for 33 years).
Birth weight, parity, and birth order did not explain the relation between breast feeding and BMI (or obesity) at 33 years, and thus were not examined further. Social class, mother’s BMI, and mother’s smoking in pregnancy, each reduced the magnitude of the relation (table 2). After adjusting for all three factors simultaneously, the relation became non-existent in males, and reduced to non-significance in females (p > 0.05). We therefore did not consider further potential confounding factors in later life. There were no interactions between breast feeding and potential confounding variables.
Our data on breast feeding had the limitation of being recalled over seven years and lacked detailed information on duration. Problems associated with recall are likely to be minimal for the three broad categories used here, which are comparable to many other studies, for example, Poulton and Williams.4 The composition of infant feeds has changed considerably over time, and consequently the extent to which we can generalise from our study remains uncertain. Nonetheless, this is to our knowledge, by far the largest study of breast feeding and BMI which reports outcome data through to mid-adulthood, and also takes account of a wide range of confounding factors. It is unclear why we see an association only in adulthood and not in childhood, although this might be due to increasing fatness of the cohort over time. However, after adjusting for social class, mother’s BMI, and mother’s smoking in pregnancy, the adulthood association disappeared, despite the limitation that mother’s BMI was self reported. Our findings, which are based on a large dataset, suggest that the method of infant feeding itself is unlikely to influence obesity. Others suggest that breast feeding protects against obesity,1,3 and inconsistencies between studies may be due to different definitions of breast feeding. Breast feeding undoubtedly has many benefits for child health which justify its promotion as the feeding method of choice, but to date, the literature on its role as a preventive measure for obesity is very inconclusive.
Contributors: TP and CP developed the hypotheses for the study. TP, CP, and OM discussed core ideas, and participated in analyses and in writing this paper. TP and CP are guarantors for the study.
Data access: 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), and the Data Archive distributor, Colchester, Essex (SN: 3148. 1994).
Funding: These analyses were funded by the Department of Health (England); the views expressed in this publication are those of the authors and not necessarily those of the sponsors.