Antibiotics and risk for birth defects

We read with interest the article, ‘Use of antibiotics during pregnancy and the risk of major congenital malformations: a population‐based cohort study’ by Muanda, Sheehy and Bérard 1. We appreciate the authors for highlighting the importance of studying the safety of antibiotic use during pregnancy. However, we are concerned that the authors dismiss data from case–control studies as being inherently less valid than those from prospective cohort studies, without acknowledging the limitations of the latter.

The authors observed no association between nitrofurantoin exposure during pregnancy and the birth defect categories they considered. Data from case–control studies have reported associations between nitrofurantoin exposure and risk for specific birth defects 2, 3. Muanda et al. attribute these differences to recall bias, which can theoretically occur if mothers of infants born with a birth defect are more likely to recall an exposure than are mothers with unaffected infants. Maternal antibiotic use was assessed in two analyses using data from the National Birth Defects Prevention Study (NBDPS), a multisite, population‐based case–control study. In the first analysis, Crider et al. examined six different classes of antibiotics and reported associations primarily among only two of those classes 2. If recall bias were driving these associations, we would expect that most or all of the antibiotic classes considered would be associated with the outcomes, since it is unlikely that mothers of affected pregnancies would remember one type of antibiotic over another. A follow‐up analysis used mothers who had a urinary tract infection and took penicillin as the referent exposure group 3. This analytical design reduces recall bias because mothers of all infants had exposed pregnancies.

The authors did not acknowledge all weaknesses of their analysis, some of which may explain discordant results with previous studies. One limitation is lack of outcome specificity. Teratogens rarely increase risk for all congenital malformations; they tend to affect certain developmental processes, resulting in specific defects 4. Analysing all congenital malformations combined or large groups of heterogeneous defects as a single group can obscure associations with specific defects. We compared the results from the Crider et al. NBDPS analysis to those of Muanda et al., who reported an odds ratio (OR) of 1.3 (95% confidence interval [CI]: 0.9, 2.0) for the association between nitrofurantoin and any cardiac malformation. The Crider et al. analysis also showed an OR of 1.3, with a similar CI (0.8, 1.9), for combined heart defects. However, when Crider et al. assessed the specific cardiac defect hypoplastic left heart syndrome (HLHS), they observed a fourfold increased risk (OR: 4.2; 95% CI: 1.9, 9.1), which was obscured when HLHS was combined with other cardiac defects. Muanda et al. lacked the sample size to analyse specific defects, a result of the cohort study design. The prevalence of HLHS is approximately 2.3/10 000 live births 5, and we would therefore expect approximately 32 cases of HLHS among their cohort of 139 938 pregnancies. By contrast, in the Crider et al. analysis of case–control data, there were 243 cases of HLHS, which would require a cohort size of over 1 million infants.

The authors base their exposure on prescription records, which provide the precise medication and dose dispensed; these details are challenging to collect from interviews. However, while prescription records provide information on prescriptions filled, they do not indicate whether women took the medication as prescribed. Although Muanda et al. cite their previous work showing that 87% of women with prescriptions for antibiotics during pregnancy also reported taking them, that same analysis showed that during the first trimester, only 38% of women with antibiotic prescriptions reported actually taking these medications 6.

Given the importance of treating infections during pregnancy, transparency about limitations of research is critical, so pregnant women and their healthcare providers have the best information.

Competing Interests

There are no competing interests to declare.

Tinker, S. C. , Crider, K. S. , and Ailes, E. C. (2018) Antibiotics and risk for birth defects. Br J Clin Pharmacol, 84: 1626–1627. doi: 10.1111/bcp.13581.