By Philip R. Fischer, MD, DTM&H
Professor of Pediatrics, Department of Pediatric and Adolescent Medicine, Mayo Clinic, Rochester, MN
Dr. Fischer reports no financial relationships relevant to this field of study.
SYNOPSIS: Despite published concerns, there is no good evidence that macrolide use during pregnancy causes birth defects.
SOURCE: Fan H, Gilbert R, O’Callaghan F, Li L. Associations between macrolide antibiotics prescribing during pregnancy and adverse child outcomes in the UK: Population based cohort study. BMJ 2020;368:m331.
The macrolide antibiotics, erythromycin, clarithromycin, and azithromycin, are used commonly during pregnancy. However, safety concerns have been raised. Thus, Fan and colleagues reviewed data from a large cohort of primary care patients, including, in fact, approximately 7% of people in the United Kingdom. This database is thought to be representative of the diverse population of the United Kingdom. Children born from January 1990 through June 2016 were included if they were registered within six months of birth. Stillborn babies were excluded, as were babies with known genetic syndromes and offspring of mothers who received a “known” teratogen during pregnancy.
Fan and colleagues looked particularly at those mothers who received a prescription for a macrolide between the fifth and final weeks of their gestation. Comparison groups included those who had been treated with penicillin, those who had received a macrolide or a penicillin prescription between 50 and 10 weeks prior to the pregnancy, and siblings of the children included in the study cohort.
The primary outcome measures were malformations and neurodevelopmental disorders (cerebral palsy, epilepsy, attention deficit disorder, and autism spectrum disorder). There were 1,071,379 live births in the database during the study period, and 726,274 had adequate data for inclusion without meeting exclusion criteria. (Most of the excluded subjects were excluded because adequate pre-pregnancy data were not available to complete all aspects of the study.) A total of 621,669 were excluded since they did not receive antibiotics during pregnancy. For comparison, that left 104,605 who had received a macrolide or penicillin, and comparison groups of 82,314 who had received a macrolide or penicillin prior to pregnancy and 53,735 siblings of treated mothers. Of the treated cohort, 24,714 were treated in the first trimester, and 79,891 were treated during subsequent trimesters. Penicillin antibiotics were given approximately 10 times as often as macrolides.
Major malformations were seen in 27.7 per 1,000 children born to mothers who had received a macrolide in the first trimester and in 17.7 per 1,000 children of the mothers who received a penicillin during the first trimester. Corresponding prevalence rates for the subsequent two trimesters of pregnancy were 19.5 for macrolides and 17.3 for penicillins. Specifically, cardiovascular malformations were seen in 10.6 per 1,000 live births for macrolides vs. 6.6 for penicillins.
Genital malformations, such as hypospadias, were reportedly more common with macrolide use than with penicillin use in all trimesters. Based on these data, the authors proposed that, until further data are available, macrolide use in pregnancy be replaced by the use of other antibiotics whenever feasible.
COMMENTARY
Fan and colleagues did a rigorous retrospective study to determine whether macrolide use during pregnancy is associated with increased risks for birth defects and neurodevelopmental problems in the children born to macrolide-treated mothers. The study had a large enough population base to be adequately powered to identify relevant risks. The investigators wisely used several different comparison groups to help readers infer risks actually due to macrolide use, as opposed to other coincident pregnancy-related situations. They concluded that: “prescribing macrolide antibiotics during the first trimester of pregnancy was associated with an increased risk of any major malformation,” and, “macrolide prescribing in any trimester was associated with an increased risk of genital malformations.”
Their work has already prompted a published warning about macrolide use during pregnancy in the nursing literature.1 If true, these conclusions should prompt “caution” and a decision to use alternative antibiotics “if feasible,” as the authors suggested. However, several thought processes cast doubt on the appropriateness of the published conclusions.
As originally published, the paper suggested that the authors’ recommendation to restrict the use of macrolide antibiotics during pregnancy was in line with published British guidelines. However, as noted in a correction subsequently published online, those British guidelines actually had merely advised to use macrolides only when the benefit outweighs the risk,2 similar to guidelines for any medication. The authors concluded that macrolide use “in any trimester” was associated with the risk of malformations.
This conclusion runs counter to consideration of plausibility. The malformations evaluated, specifically genital malformations such as hypospadias, result from alterations in fetal development taking place early in pregnancy. Genital structures are already formed before the third trimester,3 and it is not plausible that a treatment in the third trimester could reverse aspects of formation/development that are completed already. Nonetheless, Fan and colleagues used their data to conclude that second- and third-trimester exposure to macrolides does increase the risk of hypospadias.
In addition, Fan and colleagues reached their “any trimester” conclusions when combining data from each trimester. When there was a strongly significant statistical association during one part of pregnancy, statistical significance (P < 0.05) still was present when including data from trimesters during which there was no significant difference. Such was the case with genital malformations, mostly hypospadias, when the researchers found a “significant” P value in one trimester and then combined data from the other trimesters to conclude that the “risk” related to all trimesters.
Statisticians also would remind us that “statistical significance” actually just means that there is more than a 5% chance of a meaningful difference between groups. When 20 different unassociated variables are evaluated, one would expect, statistically, that at least one of those variables would rank in that “P < 0.05” significance range. Fan and colleagues made scores of comparisons, without using a Bonferroni or other “correction factor” in testing significance, and found only a very small number that “reached significance.” It is possible that the identified risk of 10.6 per 1,000 incidence of cardiac malformations with macrolide use during the first trimester, compared to a 6.6 per 1,000 incidence in penicillin-treated mothers (P = 0.03), with a total of 172 incident cases among the whole database, was the result of the “random chance” of finding some variables with low P values when testing scores of factors.
Similarly, it is not surprising that an apparently random “risk” was found (with a P value of 0.018) of urinary malformations in siblings of children whose mothers received macrolides vs. penicillin during pregnancy.
Of course, one could wonder if the illness prompting antimicrobial use, rather than the antimicrobial itself, might have triggered the risk of poor fetal outcomes. Wisely, Fan and colleagues provided indication-based data about their findings. In a supplemental table, they showed that macrolide vs. penicillin use for respiratory infections during the first trimester was associated only with a risk of cardiac malformations (95% confidence interval for risk, 1.05-2.51). However, there was no increased risk specifically for ventricular septal defects, atrial septal defects, or patent ductus arteriosus. Also, there was no difference in risk between these groups for other malformations or other poor neurodevelopmental outcomes.
In fact, the clearly negative findings of this study are valuable. Macrolide use in pregnancy was not associated with any risk of cerebral palsy, autism spectrum disorder, attention deficit hyperactivity disorder, or epilepsy. In a recent meta-analysis, macrolide use during pregnancy had only a “weak association” with any risk of congenital malformations, and that was only with first-trimester exposure and with either gastrointestinal or musculoskeletal malformations.4 A separate systematic review published last year by Fan and colleagues found hints only of macrolide-related risk of miscarriage, cerebral palsy, epilepsy, and gastrointestinal malformations — but not other malformations, stillbirths, or neonatal deaths.5
Interestingly, 25% of the 726,274 children in this study were exposed to prenatal macrolides or penicillins, and an additional 7% were exposed to other antibiotics. Thus, approximately one-third of women received antibiotic treatment during pregnancy. For those for whom an indication for antimicrobial therapy was identified in the medical record, 75% were treated for respiratory infections. One reasonably could wonder if bacterial respiratory infections truly required antimicrobial treatment in such a large proportion of pregnant women. Doubt about all the treatment actually being necessary certainly adds support to the notion that antimicrobials should be used judiciously and only when truly indicated, whether or not specific risks of treatment are identified. Thus, these new data from Fan and colleagues provide significant reassurance about problems that are likely not due to macrolide use during pregnancy. They do not provide evidence of much significant risk for even cardiac and genital malformations being due to macrolide use. At the same time, these new data do provide good reminders that antimicrobials should be used judiciously during pregnancy.
REFERENCES
- [No authors listed]. NewsCAP. Use of macrolide antibiotics during pregnancy increases risk of birth defects. Am J Nursing 2020;120:16.
- Correction: Fan H, Gilbert R, O’Callaghan F, Li L. Associations between macrolide antibiotics prescribing during pregnancy and adverse child outcomes in the UK: Population based cohort study. BMJ 2020;368:m766.
- van der Horst HJ, de Wall LL. Hypospadias, all there is to know. Eur J Pediatr 2017;176:435-441.
- Mallah N, Tohidinik HR, Etminan M, et al. Prenatal exposure to macrolides and risk of congenital malformations: A meta-analysis. Drug Saf 2020;43:211-221.
- Fan H, Li L, Wijlaars L, Gilbert RE. Associations between use of macrolide antibiotics during pregnancy and adverse child outcomes: A systematic review and meta-analysis. PLoS One 2019;14:e0212212.
