Associations between macrolide antibiotics prescribing during pregnancy and adverse child outcomes in the UK: population based cohort studyBMJ 2020; 368 doi: https://doi.org/10.1136/bmj.m331 (Published 19 February 2020) Cite this as: BMJ 2020;368:m331
All rapid responses
We agree with Profs Damkier and Hopayian that recommendations about macrolide prescribing should take into account the strength of evidence from all relevant studies. We welcome this opportunity to discuss the findings from their study. However, we note that Damkier et al did not reference our recent systematic review of comparative studies of macrolides vs penicillins in pregnancy on a range of adverse fetal and child outcomes.
Both the study by Damkier et al and our study were based on large population-based databases of around one million live births. However, important differences in the way the exposure groups were specified may partly account for the different findings.
Firstly, we tried to minimise two potential biases. The effect of infection itself on adverse child outcomes (known as indication bias) and the benefit of antibiotic treatment on adverse outcomes of infection in both the macrolide and penicillin groups, which could mask a harmful effect of the treatment. We restricted the comparison groups to women most likely to have mild infections and to benefit least from antibiotic treatment. Our analyses included women who received a single monotherapy of macrolide or penicillin antibiotics during pregnancy (42% of all macrolides in pregnancy). In this way, we removed women with severe or recurrent infections, who might be more likely to benefit from antibiotics. We also conducted a sensitivity analysis restricted to women treated with monotherapy for respiratory tract infections (RTIs). As RTIs are predominantly caused by viral infections, these women would be unlikely to benefit from antibiotic treatment.
The study by Damkier et al did not restrict to monotherapy and compared women who ever filled a prescription for macrolides with women who filled only safe penicillins during the first trimester (mostly Phenoxymethylpenicillin). Thus women in the macrolides group could have had more complex or severe infections than those in the penicillin group (also mentioned by the authors). Given these comparison groups, a small risk of harm from macrolides could have been masked by the beneficial effect of macrolides on infections.
Secondly, to minimise potential dilution bias, we removed 36% of women treated in the first trimester because their treatment started before four weeks of gestation, which is before the onset of organogenesis and when congenital malformations cannot be induced by teratogenic agents. The study by Damkier et al included all treatment since the last menstrual period (the first day of gestation), thereby diluting the relative effect of macrolides vs penicillins by including women who were not susceptible to an adverse effect of treatment.
The letter from Saripanidis refers to previous research showing adverse cardiac effects of macrolides in patients with heart disease. These examples emphasize the need to draw on evidence from human and animal studies to understand potential mechanisms for an adverse effect of macrolides. One potential mechanism is through the Ikr blocking potential of macrolides in cardiomyocytes. Such effects on embryonic hearts have been reported in animal studies for other Ikr blocking agents. 
We agree that uncertainty remains about the association of macrolides with harm in the fetus. Therefore, where alternative, safe antibiotics can be used, we recommend avoiding macrolides in pregnancy. We need further research to compare macrolide with penicillin treatment, based on mapping specific sensitive exposure periods to adverse fetal and child outcomes. To achieve a sufficient sample size, such research would need to bring together existing datasets from a number of high quality cohorts that have accurate measurements of treatments, underlying infections and child outcomes.
 Damkier P, Brønniche LMS, Korch-Frandsen JFB, Broe A. In utero exposure to antibiotics and risk of congenital malformations: a population-based study. Am J Obstet Gynecol. 2019 Dec;221(6):648.e1-648.e15. doi: 10.1016/j.ajog.2019.06.050. Epub 2019 Jun 28.
 Danielsson BR, Danielsson C, Nilsson MF. Embryonic cardiac arrhythmia and generation of reactive oxygen species: common teratogenic mechanism for IKr blocking drugs. Reprod Toxicol 2007;24:42-56. doi:10.1016/j.reprotox.2007.04.005
Competing interests: First author of the original article.
With respect to Dr Saripanidis, the second reference is to a news website reporting on a poster presentation so not amenable to critical appraisal. The fourth reference is to a meta-analysis  by the same authors as this observational study and on reading it, I could not see a mention of the large study that Prof Damkier mentions . So we have one new observational study  that finds a link between macrolide prescribing in 1st trimester and a previous, larger one that does not. We have a meta-analysis that finds some links but misses out the said large observational study. What should we conclude? Well, we could look at the Risk Ratio. Many epidemiologists would expect a minimum RR of 2 before accepting causality. In both Fan's BMJ article and the meta-analysis, the RR is below 2.
 Damkier P, Brønniche LMS, Korch-Frandsen JFB, Broe A. In utero exposure to antibiotics and risk of congenital malformations: a population-based study. Am J
Obstet Gynecol. 2019 Dec;221(6):648.e1-648.e15.
 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.
Competing interests: No competing interests
Not Dark Yet: Associations between macrolide antibiotics prescribing during pregnancy and congenital malformations.
I read this paper with great interest. It appears to be a well-designed, conducted and reported study that adds information to a subject of some controversy.
However, I am surprised and disappointed by the superficial and inadequate discussion of their findings as related to other recent publications on the very subject of macrolides in pregnancy and risk of major congenital malformations. Specifically (and yes, I hold an academic conflict of interest here) I would appreciate a justification for not including the results from our 2019 paper in the discussion and perspectives of the authors' findings. We report null-results (major and cardiovascular congenital malformations) for erythromycin, azithromycin and roxythromycin in a population-wide registry study comprising more than 13.000 first-trimester exposed live-born children (1).
For comparison, the authors findings and conclusions - as pertaining to congenital malformations - are based on 2170 first trimester exposed live-born children. I am confident that the authors have been meticulous throughout and the data-set analysed produces the results presented for the population studied. But I am really at a loss as to why - among 45 references - they could not fit in the largest amount of data reported and discuss our null-results against their findings.
Signals of potential harm are picked up and promoted by media, academia, patients and all parties with interest in health-care exponentially more so than null-results. And twice that within the sensitive and often fairly controversial field on the use of drugs during pregnancy, where it is nigh on impossible to put the lid back on, regardless of the quantity and quality of data to the contrary.
The authors call for caution and warnings in drug safety leaflets based on their results. To suggest this with no real discussion of other results - which do not suggest such risk - is not justifiable. The way the results from this study are being framed and disseminated will likely (my word) result in substantial and insufficiently substantiated doubt and uncertainties among pregnant women receiving these drugs.
Pregnant women deserve better.
Professor, MD PhD
1) Damkier P, Brønniche LMS, Korch-Frandsen JFB, Broe A. In utero exposure to antibiotics and risk of congenital malformations: a population-based study. Am J
Obstet Gynecol. 2019 Dec;221(6):648.e1-648.e15. doi: 10.1016/j.ajog.2019.06.050. Epub 2019 Jun 28.
Competing interests: First author of a paper ( PMID: 31260651; DOI: 10.1016/j.ajog.2019.06.050) in which the results are not compatible with those reported here.
Α systematic review and meta-analysis of 21 published clinical studies on 228,556 participants revealed that macrolide prescribing during pregnancy was associated with increased risks for miscarriage, cerebral palsy, epilepsy, and gastrointestinal malformations. 
The US Food and Drug Administration (FDA) issued a safety communication informing that clarithromycin prescriptions to patients with heart disease can increase heart problems and mortality many years later. 
Furthermore, antibiotics during pregnancy could also increase mother's tardive mortality.
A long clinical trial on 36,429 women who had received antibiotic treatment for more than 2 months at some point in their lives revealed that they were more likely to suffer a myocardial infarction or a stroke. 
Women who took antibiotics for 2 months or longer in late adulthood were 27% more likely to die from all causes during the study period, and 58% more likely to have a cardiovascular death. 
Assuming antibiotic pills for over 20 days, at any time in early-to-middle adulthood caused a 69% increase in colorectal adenomas, which could progress to invasive cancers over time. 
Competing interests: No competing interests
Re: Associations between macrolide antibiotics prescribing during pregnancy and adverse child outcomes in the UK: population based cohort study
I read the article with great concern and anguish. Macrolides are class C drugs in most of the western world, specifically during the first trimester. Oxford University studies have repeatedly pointed that out but still this has not yet been taken up by guidelines. These drugs are routinely being prescribed in third world countries where modern antibiotics are scarce, even in pregnancy.
I really appreciate the authors for bringing that fact to a wider audience: they deserve applause and credit. Still this disclaimer is too little and too late.
Källén, B.A., Olausson, P.O. and Danielsson, B.R., 2005. Is erythromycin therapy teratogenic in humans? Reproductive Toxicology, 20(2), pp.209-214.
Lin KJ, Mitchell AA, Yau WP, Louik C, Hernández-Díaz S. Safety of macrolides during pregnancy. American journal of obstetrics and gynecology. 2013 Mar 1;208(3):221-e1.
Competing interests: No competing interests