This longitudinal study found that early-life antibiotic exposure was associated with subsequent increases in body mass. Of the three time windows analyzed, only exposure during the period before 6 months of age was consistently associated with increases in body mass. At 38 months, children who had been exposed to antibiotics during this earliest period had significantly higher standardized BMI scores, and were 22% more likely to be overweight than children who had not been exposed. In contrast, exposures after 6 months were not consistently associated with body mass increases. Those in the 6- to 14-month period showed no association, while those in the period 15–23 months were significantly associated only with elevated standardized BMI score at 7 years, but not with consistently elevated scores in the interim.
Our finding of an association of antibiotic exposure at <6 months with later-life body mass is consistent with a prior report. It adds important evidence that exposure timing matters. We also add a test of spuriousness, with the finding that exposure to non-antibiotic medications during the same early windows is not associated with elevations in body mass. This makes it less plausible that exposure to antibiotics reflects a greater receptivity to medications, which is also correlated with increases in body mass. Unlike the previous study, however, we did not find an association with antibiotic exposure at <6 months persisting to 7 years of age. Perhaps this reflects differences in the antibiotics used in the two samples, or different doses. Given that intravenous antibiotics are used in these first 6 months of life (often for neonatal sepsis), antibiotic type (that is, Gram-positive or Gram-negative/anaerobic coverage) and route of administration (intravenous or orally administered antibiotics) might have differential effects on gut microbiota composition and development. This is consistent with a recent analysis finding associations of intravenous vancomycin, but not amoxicillin, treatment in adults with the development of obesity.50
Alternatively, our failure to find a significant association may simply reflect our somewhat smaller sample size.
That early-life antibiotic exposure can lead to increased body mass is consistent with a evidence on the farm of antibiotic-induced weight gain,10,25,26
and with more recent studies in laboratory animals elucidating a link between early antibiotic exposure and changed development in controlled environments.51
Many of these studies also find that the earliest months of life are periods of unique vulnerability to antibiotic disruption.
There are important limitations to our work. The relationship between antibiotic exposure and body mass is potentially confounded by multiple social, behavioral and biological factors. For example, BMI in infancy and childhood is associated with biological factors, such as parental BMI and maternal smoking in utero
, social class and education; feeding (breastfeeding, later family food environment) and childhood ‘lifestyle’,42
including prevalence of sedentary behavior and regularity of sleep patterns. Early-life antibiotic use also is potentially correlated to each of these, as parental receptivity to antibiotics varies by social class and ethnicity,52,53
and children in larger families or those who attend daycare also have greater exposure to contagious diseases. Fortunately, the ALSPAC data are rich, and we were able to control for many confounders in multivariable analyses. We found remarkably weak associations between antibiotic exposure and the potential confounders that we examined. However, there may be other variables which, if included, would have attenuated the reported association. Exposure was measured by parental recall for events over a 3- to 9-month period. Although such memories may be imprecise, there is no a priori
reason why exposure recall bias would be associated with child body mass.
No method to assess adiposity (BMI, ponderal index, weight-for-height or the Quetelet index) is uniquely suited to represent adiposity and risk for subsequent cardiovascular disease,54
and we accept that ponderal index and BMI Z
-scores as outcomes of adiposity may be more clinically meaningful insofar (as Howe et al
have suggested) they represent greater fat mass and cardiovascular risk later in life. Dual-energy X-ray absorptiometry data, which better represent fat mass deposition,56
are later data points available for further study of the ALSPAC data.57
Also limiting generalizability is that we only examined effects among children born at ≥2500 g. Even so, our sample of infants with normal birth weight likely included children who had substantial illness and were treated with antibiotics. To the extent that children with such illnesses are likely to be thinner, the magnitude of the association between exposure and body mass we identified may underestimate effects that occur in healthy children. Given that high birth weight is also a possible risk factor for obesity in later life, our effect size may have been diminished by inclusion of a high-risk subpopulation who is more likely to develop obesity for other reasons.58
Future work can examine exclusion of this subpopulation. Finally, the ALSPAC sample dates from the early 1990s in the United Kingdom, when antibiotic exposure was probably less frequent than that found today, given the average increase in use of 4.3% annually since 2000 in the United Kingdom.59
The effects of today’s more frequent antibiotic exposure in early life may therefore be greater, and transgenerational impacts of continued antibiotic use could compound effects.13,14
The effect sizes that we report are modest. Translated into weight increments for the average 38-month-old, a +0.067 s.d. unit increase in BMI Z
-score corresponds to a +90 g increase in weight. Although such increases are likely to have small impacts on individuals, they may be important for population health: assuming a normal distribution, shifts of 0.067 s.d. units in BMI Z
-score (the change noted at 38 months in our final model) would increase overweight by 1.62% and obesity by 0.72%, in the population. As obesity is multifactorial and early exposure to antibiotics is common, such increases in overweight and obesity due to small shifts in means may be potentially important.60