The impact of antibiotics on growth in children in low and middle income countries: systematic review and meta-analysis of randomised controlled trials

We thank Singh et al. (1) for their comments on our manuscript (2). However, we would like to highlight a few additional points.

Singh et al. suggest that the geographic and ethnic heterogeneity between the trials included in our analyses could have biased our findings. We disagree with this assertion. The World Health Organization Multicentre Growth Reference study (3) suggests that children from different ethnic backgrounds actually grow remarkably similarly to each other, provided they live in environmental conditions conducive to healthy growth; there is little evidence that ethnicity in and of itself is an important determinant of growth.

Singh et al. also suggest that other sources of heterogeneity could have biased our results. Specifically, they point to differences in the indication for treatment and use of concurrent nutritional interventions in some trials. We pointed out these potential sources of heterogeneity in our manuscript, and explored the role of both as moderators of the treatment effect in our analyses. Neither indication for treatment nor concurrent nutritional interventions was found to be a significant source of heterogeneity between trials. While we acknowledge that the small number of trials in our study limited our ability to fully elucidate the role of these and other moderators of the treatment effect, we also identified a subset of trials, with very little heterogeneity (I-squared=0.0%), in which a large treatment effect on weight was also observed (41.4g/month; 95%CI: 31.0 to 51.7), and reported the finding in our manuscript.

Singh et al. highlight that we excluded four trials from our analysis that reported no growth benefits from antibiotics. As noted in our manuscript, these trials were excluded because they reported the effect of antibiotics on the prevalence of stunting or wasting. We were interested in pooling trials that reported the impact of antibiotics on attained growth. We reiterate here that it is not possible to pool studies with differently defined outcomes in a meta-analysis without the risk of introducing bias (4). In addition, Egger’s test suggested that publication bias was not likely in our analysis. Therefore we conclude that inclusion of these studies, with malnutrition prevalence as the outcome, could have introduced a bias into our analyses; moreover, there was no evidence that the exclusion of these studies was a source of publication bias.

We agree with Singh et al. that the benefits of antibiotic use with respect to morbidity and mortality in children is already well characterised. In their comment, they pose the following question “[I]s it worth giving antibiotic[s] to children irrespective of their nutritional status…?” They suggest that our manuscript sheds no light on this issue. However, we disagree with this assertion. Our findings demonstrate that, in addition to the recognized benefits of antibiotics in saving the lives of children at high risk for early mortality (e.g. HIV-infected or exposed, and severely malnourished children), antibiotics can also improve child growth. This has positive implications for their future health and development.

The comments by Singh et al. do not influence our conclusions that: (i) on average, antibiotic use produces significant growth gains in children,for their sex and age, from low- and middle-income countries; (ii) weight was more responsive than length to such treatment, particularly in populations with a high prevalence of HIV infection or exposure, and a high prevalence of severe acute malnutrition; (iii) widespread use of antibiotics to improve growth is unlikely to be a viable public health strategy, due to concerns about antibiotic resistance and potential adverse events; and (iv) the biological mechanism(s) responsible for this antibiotic associated growth effect may involve resolution of clinical or subclinical infections and/or microbiota alteration; more research is needed to better understand these mechanisms.

Finally, we should clarify that our analysis included ten trials, not eight as reported by Singh et al.

1. Singh MM, Devi R, Wadhwa V. Can antibiotics provide hope for promoting growth in children? http://www.bmj.com/content/348/bmj.g2267?tab=responses
2. Gough EK, Moodie EEM, Prendergast AJ, Johnson SMA, Humphrey JH, Stoltzfus RA et al. The impact of antibiotics on growth in children in low and middle income countries: systematic review and meta-analysis of randomised controlled trials. BMJ 2014;348:g2267
3. WHO Multicentre Growth Reference Study Group. Assessment of differences in linear growth among populations in the WHO Multicentre Growth Reference Study. Acta Paediatr Suppl 2006;450:56-65.
4. Greenland S, O’Rourke K. Meta-analysis. In: Rothman KJ, Greenland S, Lash Tl, eds. Modern Epidemiology. Lippincott Williams and Wilkins, 2008:652-82