Antibiotics and Fecundability Among Female Pregnancy Planners

A Prospective Cohort Study

Holly Michelle Crowe; Amelia Kent Wesselink; Lauren Anne Wise; Tanran R. Wang; Charles Robert Horsburgh Jr.; Ellen Margrethe Mikkelsen; Elizabeth Elliott Hatch


Hum Reprod. 2021;36(10):2761-2768. 

In This Article


Overall, we found little evidence of an association between antibiotic use and fecundability. The use of sulfonamides and lincosamides may be associated with slightly improved fecundability, while the use of macrolides or use of antibiotics for pelvic or vaginal indications may be associated with slightly reduced fecundability, although estimates were imprecise.

Macrolides and lincosamides, while chemically distinct, share a common mechanism of action (interruption of bacterial protein synthesis) (Leclercq and Courvalin, 1991). However, we found opposite associations between macrolides and fecundability (FR: 0.70, 95% CI: 0.47–1.05) and lincosamides and fecundability (FR: 1.58, 95% CI: 0.96–2.60). These opposite observed associations may be more attributable to chance or confounding by indication rather than actual effects of these specific types of antibiotics. Although there were not sufficient numbers to assess combinations of indication and antibiotic class, macrolides were more frequently used orally to treat illness, while lincosamides were more frequently used topically for skin conditions such as acne, which may partially explain the divergent findings.

Most lincosamides and macrolides are categorized as Food and Drug Administration (FDA) Category B for use in pregnancy, meaning that animal studies show no harm or animal studies show harm that has been unconfirmed in human studies (Sá Del Fiol et al., 2005). Clarithromycin, a specific macrolide, has been associated with spontaneous abortion, which may share a common etiological pathway with prolonged time to pregnancy, and is classified as FDA category C, meaning that use during pregnancy should be avoided whenever possible (Sá Del Fiol et al., 2005; Andersen et al., 2013b). While there were too few clarithromycin users to examine clarithromycin independently in this sample, a strong association between clarithromycin and prolonged time to pregnancy or early spontaneous abortion could contribute to the lower fecundability observed in macrolide users.

Lower fecundability was observed among participants using antibiotics for vaginal or pelvic infections, but not among participants using antibiotics for respiratory infections, despite similar patterns of macrolide use (17% of participants with vaginal or pelvic infections and 24% of participants with respiratory infections used macrolides), suggesting potential confounding by indication. The reduced fecundability observed among women with a history of reproductive health conditions and women taking an antibiotic for pelvic or vaginal infections also provides evidence for confounding by indication. Residual confounding by indication is a concern, as we were not able to jointly stratify by antibiotic class and indication.

Bacterial vaginosis is commonly treated with nitroimidazole or lincomycin antibiotics. It is believed to be the most common vaginal infection among women of reproductive age (Casari et al., 2010) and has been associated with subfecundity and infertility in several studies (Salah et al., 2013; Haahr et al., 2016; Lokken et al., 2021). A recent prospective cohort study of 458 Kenyan pregnancy planners found that patients with bacterial vaginosis at the health care visit prior to pregnancy testing had 17% lower fecundability than those without bacterial vaginosis (FR: 0.83, 95% CI: 0.6–1.1). The effect was stronger in women who had persistent bacterial vaginosis, which was associated with a 43% reduction in fecundability (FR: 0.57, 95% CI: 0.4–0.8) (Lokken et al., 2021). A 2013 cohort study found that women with infertility had over five times the odds of bacterial vaginosis than fertile women (odds ratio: 5.23, 95% CI: 3.06–8.12) (Salah et al., 2013). When stratifying by cause of infertility, investigators found that the 6-month cumulative pregnancy rate was higher among those with treated bacterial vaginosis than those with untreated bacterial vaginosis, regardless of the underlying cause of fertility (hormonal or unexplained). A 2019 meta-analysis of 12 studies found that the prevalence of tubal factor infertility was greater in patients with bacterial vaginosis than in those with normal vaginal microbiota (45% vs 28%) (Haahr et al., 2019). In addition to the vaginal microbiome, the microbiome of the upper reproductive tract may be important for fertility, as semen protects sperm from the acidic vaginal environment, but vaginal bacteria can ascend up the reproductive tract and adversely impact fertility (Suarez and Pacey, 2006; Moreno and Simon, 2019). These findings suggest a complex relation between the reproductive tract microbiome, hormonal disturbances, and fertility issues, which may be, at least in some cases, improved with antibiotic treatment (Salah et al., 2013; Haahr et al., 2019; Lokken et al., 2021). While we were unable to specifically analyze use of an antibiotic for bacterial vaginosis, when analyzing use of nitroimidazole, lincomycins, and use of an antibiotic for a vaginal infection overall, our study found mixed results. Given the differing mechanisms by which antibiotics may be helpful or detrimental to fertility, the overall null results may mask heterogeneous effects. While treating an active infection with antibiotics may improve the chances of conception, repeated use of antibiotics may interfere with conception. Some studies have provided evidence of an association between levels of reproductive hormones, vaginal pH, and relative bacterial abundance in the vagina (Casari et al., 2010; Farage et al., 2010; Wira et al., 2015; Emanuele Levi-Setti, 2016). Alterations in the acidity of the genital tract may interfere with the mobility of sperm and, therefore, conception. While antibiotics can aid fertility by resolving an acute infection, excessive antibiotic therapy may alter the normal reproductive tract flora and disrupt the optimal bacterial environment for conception and implantation (Emanuele Levi-Setti, 2016). Repeated exposure to small doses of antibiotics, with subsequent superinfections and changes in vaginal acidity, was proposed as a mechanism for the observed effect of antibiotics on time to pregnancy in the Danish study of pharmacy assistants (Schaumburg and Olsen, 1989).

When we stratified the analysis of overall time-varying antibiotic use and fecundability, we observed slightly lower fecundability among antibiotic users in women <30 years of age, and current smokers. Distribution of medication class and indication were similar among younger and older women, although younger women were slightly less likely to use antibiotics for a pelvic infection than older women (8% vs 11% of ever antibiotic users). As antibiotic use for a pelvic infection was associated with slightly lower fecundability, fewer pelvic infections in this group would not explain the slightly lower fecundability observed among antibiotic users.

Current smokers and nonsmokers also had similar distributions of antibiotic class and indications, although current smokers were slightly more likely to use sulfonamides (8% vs 4%, respectively) and an antibiotic for a vaginal infection (9% vs 5%, respectively). As sulfonamides and use of an antibiotic for a vaginal infection were associated with fecundability in opposite directions, it is unlikely that differing antibiotic classes or indications explain the stronger findings among smokers.

Stratified analysis showed that antibiotic use may have a slightly different association with fecundability among different subgroups of pregnancy planners, although these associations were modest, further illustrating the potentially complex relation between antibiotic use, underlying health, and fecundability.

In addition to confounding by indication, potential sources of bias in this study include misclassification, selection bias, and unmeasured or residual confounding. Participants may misremember the timing or type of antibiotic taken for a particular infection. Such misclassification is likely to be non-differential and reduced by the 4-week recall period. As participants were surveyed every 8 weeks and asked about their medication use in the past 4 weeks, medication use in the 4 weeks immediately following a follow-up questionnaire was not assessed. This missing antibiotic use would be unlikely to materially impact our results, as singular instances of antibiotic use are likely to be transient exposures, and we did not have a sufficient number of women taking multiple antibiotics over follow-up to analyze the association between repeated antibiotic use and fecundability.

Misclassification of time to pregnancy is possible, particularly if participants incorrectly report their cycles of attempt time at study entry, although such misclassification is less likely in a study of pregnancy planners and likely non-differential. Selection bias due to differential loss to follow-up is unlikely, as antibiotic users were not substantially more or less likely to be lost to follow-up than non-users. Residual confounding by factors such as diet quality, health care quality, or health care accessibility is possible, although this study collected extensive data on participant demographic and lifestyle factors, as well as reproductive and medical history.

FR for several antibiotic classes and indications for use were imprecise, owing to the small numbers of participants reporting these use patterns. Small numbers also prevented more detailed analysis of the association between specific antibiotic and indication combinations and fecundability, which may aid in identifying how antibiotics may be associated with increased or decreased fecundability.

Our results do not support the hypothesis that antibiotic use overall is associated with fecundability, however, specific types of antibiotics and specific indications for their use may be associated with either increased or decreased fecundability. Several factors, including medical history, underlying health status, inflammatory responses, and the reproductive tract microbiome, likely contribute to these differences in observed effects, although stratified analyses show that antibiotic use may be slightly more likely to be associated with lower fecundability among younger women, smokers, and those with a history of reproductive health conditions.