Antibacterial Medication Use During Pregnancy and Risk of Birth Defects: National Birth Defects Prevention Study

Objective  To estimate the association between antibacterial medications and selected birth defects.

Design, Setting, and Participants  Population-based, multisite, case-control study of women who had pregnancies affected by 1 of more than 30 eligible major birth defects identified via birth defect surveillance programs in 10 states (n = 13 155) and control women randomly selected from the same geographical regions (n = 4941).

Main Exposure  Reported maternal use of antibacterials (1 month before pregnancy through the end of the first trimester).

Main Outcome Measure  Odds ratios (ORs) measuring the association between antibacterial use and selected birth defects adjusted for potential confounders.

Results  The reported use of antibacterials increased during pregnancy, peaking during the third month. Sulfonamides were associated with anencephaly (adjusted OR [AOR] = 3.4; 95% confidence interval [CI], 1.3-8.8), hypoplastic left heart syndrome (AOR = 3.2; 95% CI, 1.3-7.6), coarctation of the aorta (AOR = 2.7; 95% CI, 1.3-5.6), choanal atresia (AOR = 8.0; 95% CI, 2.7-23.5), transverse limb deficiency (AOR = 2.5; 95% CI, 1.0-5.9), and diaphragmatic hernia (AOR = 2.4; 95% CI, 1.1-5.4). Nitrofurantoins were associated with anophthalmia or microphthalmos (AOR = 3.7; 95% CI, 1.1-12.2), hypoplastic left heart syndrome (AOR = 4.2; 95% CI, 1.9-9.1), atrial septal defects (AOR = 1.9; 95% CI, 1.1-3.4), and cleft lip with cleft palate (AOR = 2.1; 95% CI, 1.2-3.9). Other antibacterial agents that showed associations included erythromycins (2 defects), penicillins (1 defect), cephalosporins (1 defect), and quinolones (1 defect).

Conclusions  Reassuringly, penicillins, erythromycins, and cephalosporins, although used commonly by pregnant women, were not associated with many birth defects. Sulfonamides and nitrofurantoins were associated with several birth defects, indicating a need for additional scrutiny.

Antimicrobials, and antibacterial drugs in particular, are among the most commonly used medications during pregnancy because treatment of infections is critical to the health of a mother and her fetus.^1,2 Although it appears that some classes of antibiotics have been used relatively safely during pregnancy, to our knowledge there have been no large-scale studies addressing safety or risk for many classes of antibacterial drugs.^1,3 Obvious ethical and logistical difficulties in studying drug safety among pregnant women have limited the amount of information available to women and their health care providers in choosing a drug for needed treatment. Some antibiotics, such as penicillins and erythromycins, have had a long history of use without reports of deleterious effects in fetuses. However, due to the continually expanding problem of antibacterial resistance worldwide, health care providers have been forced to use a wider array of newer antibiotics with no existing information on safety during pregnancy. In addition, emerging infections and prophylactic use during potential bioterrorism threats may expose even more women to unevaluated antibacterial and other antimicrobial drugs during pregnancy.

The objective of this study is to evaluate the association between selected birth defects and antibacterial drug use during the critical developmental stages in early pregnancy (1 month before pregnancy through the third month of pregnancy).

The National Birth Defects Prevention Study is an ongoing population-based, case-control study of birth defects in the United States.⁴ Case infants with at least 1 of more than 30 categories of major birth defects were identified via birth defects surveillance systems in Arkansas, California, Georgia, Iowa, Massachusetts, New Jersey, New York, North Carolina, Texas, and Utah. Case information was reviewed by clinical geneticists at each site to determine study eligibility. Infants with recognized or strongly suspected chromosomal abnormalities or single-gene conditions were excluded from the study. After inclusion in the study, all cases with 1 specific defect were then classified by clinical geneticists to establish consistency for the defect and to determine whether the defect pattern was isolated or multiple (>1 major malformation). Case infants could be live born (all sites), still born (all sites except New Jersey), or induced abortions (all sites except Massachusetts and New Jersey). Case infants included infants with isolated and multiple defects; infants with multiple eligible defects were included in each of those categories. Our analyses include roll-up categories, eg, we assessed associations for all included heart defects, conotruncal heart defects, and tetralogy of Fallot.

Control infants were a random sample of live births without major birth defects, selected from the same geographical areas from either birth certificates or birth hospitals. Case and control infants whose estimated dates of delivery were from October 1, 1997, through December 31, 2003, were included in these analyses. Mothers with incomplete interviews as well as mothers who reported preexisting type 1 or type 2 diabetes mellitus were excluded from the analysis. The study has been approved by the institutional review boards of all participating institutions.

As described previously,⁴ mothers of case and control infants completed a detailed telephone interview in English or Spanish language within 24 months after the estimated date of delivery. Oral informed consent was obtained. The interview addressed preconceptional, periconceptional, and pregnancy exposures. It included questions about maternal diseases, pregnancy history, demographics, education, and medication use. Mothers could report medication use either in response to questions related to specific diseases (such as respiratory illness and kidney, bladder, or urinary tract infections) or in response to general questions about medication use. Specific names of antibacterials were read to help the mothers recall (eg, amoxicillin, Augmentin [amoxicillin and clavulanate potassium combination], Bactrim [trimethoprim and sulfamethoxazole combination], and Keflex [cephalexin]). When a mother reported use of a medication, she was asked for details about timing and duration but not for dosage information. Mothers' responses were combined into exposures for each of the pregnancy months such that a mother could have an exposure to a specific antibacterial in the month before pregnancy and an antibacterial not otherwise specified in the second month of pregnancy.

To identify medications and active ingredients in multicompound products, all drugs were linked to the Slone Drug Dictionary.⁵ We focused on the most common and narrowest reference encompassing the following categories of antibacterial medications: (1) penicillins, (2) erythromycins, (3) nitrofurantoins, (4) sulfonamides, (5) cephalosporins, (6) quinolones, (7) tetracyclines, (8) other miscellaneous β-lactams, (9) aminoglycosides, (10) antimycobacterial agents, and (11) antibiotics that were specified but not part of the previously mentioned categories or not otherwise specified. Antimicrobial drugs that were primarily antiviral, antifungal, or antiparasitic were not included. Women who reported these drugs were not excluded from analyses.

To assess the association between antimicrobial drugs and birth defects, we defined a mother as exposed if she used an antibacterial during the periconceptional period from the month prior to the estimated date of conception through the end of the first trimester (defined as the end of the third month of pregnancy). Mothers who reported that they did not know whether they were exposed or said they were exposed but were uncertain of the timing were excluded from regression analysis. This was a limited number of participants, eg, for the analysis of any antibacterial exposure, 245 case mothers (1.9%) and 75 control mothers (1.5%) were excluded from regression for these reasons. Mothers were considered unexposed if they reported no antibacterial use (any class) in the periconceptional period.

Odds ratios (ORs) adjusted for potential confounders were estimated using logistic regression models if there were at least 4 exposed cases. After review of the literature, we selected an a priori list of confounders that included maternal age (aged <35 vs ≥35 years), race (white vs other), education (≤12 vs >12 years), prepregnancy body mass index (calculated as weight in kilograms divided by height in meters squared) (<30 vs ≥30), time from the estimated date of delivery to the interview (≤12 vs >12 months), use of folic acid or multivitamins containing folic acid from 1 month before pregnancy through the first month of pregnancy, and any periconceptional smoking or alcohol use. We created separate logistic regression models with the same set of confounders for each combination of birth defect and antibacterial medication. We performed subgroup analyses including only those cases and controls without a family history of the birth defect studied.

The participation rate in the maternal interview was 70.5% among case mothers and 67.2% among control mothers, resulting in 13 586 case mothers and 5008 control mothers participating. Mothers with incomplete interviews (122 case mothers and 41 control mothers) as well as mothers who reported preexisting type 1 or type 2 diabetes mellitus (310 case mothers and 26 control mothers) were then excluded from the analysis, resulting in a total of 13 155 case mothers and 4941 control mothers included in the analysis.

Use of antibacterials was common among our control mothers, ranging from a prevalence of 2.0% in the month that was 3 months prior to conception to 5.8% during the fourth month after conception (Figure). The most common single category of antibacterial exposure was penicillins (Table 1). No control mothers reported use of other β-lactams, chloramphenicol, or antimycobacterial drugs. A limited number of case and control mothers reported exposure to multiple classes of antibacterials during the periconceptional period: 113 (0.6%) reported 2 classes, 14 (0.1%) reported 3 classes, and 3 (0.02%) reported 4 classes.

The distributions for maternal race/ethnicity, age, and prepregnancy body mass index were very similar for cases and controls (Table 1). Overall, case mothers were interviewed later than control mothers. Use of antibacterials from 3 months before pregnancy through the end of pregnancy was reported by 3863 case mothers (29.4%) and 1467 control mothers (29.7%). The type of antibacterial could not be specified by 1342 case mothers (34.7%) and 463 control mothers (31.6%).

Penicillins were the most commonly reported class of antibacterials used during the critical fetal development window, and our results showed a single elevated OR in a defect with limited sample size (intercalary limb deficiency, 4 exposed cases of 24 total cases; adjusted OR [AOR] = 3.1; 95% confidence interval [CI], 1.0-9.4) (Table 2). To clarify, this means that women who have children with an intercalary limb deficiency were 3 times more likely to report use of penicillins vs no antibacterials compared with mothers of children without a birth defect.

Erythromycins were the next most commonly reported class of antibacterials. Only anencephaly (AOR = 2.4; 95% CI, 1.1-5.3) and transverse limb deficiency (AOR = 2.1; 95% CI, 1.0-4.2) were associated with erythromycin exposure (Table 2).

Nitrofurantoins had 4 associations: anophthalmia or microphthalmos (AOR = 3.7; 95% CI, 1.1-12.2), hypoplastic left heart syndrome (AOR = 4.2; 95% CI, 1.9-9.1), atrial septal defects (AOR = 1.9; 95% CI, 1.1-3.4), and cleft lip with cleft palate (AOR = 2.1; 95% CI, 1.2-3.9) (Table 3).

The use of sulfonamides was associated with more defects than any other antibacterial class. The defects included anencephaly (AOR = 3.4; 95% CI, 1.3-8.8), 2 left-sided heart defects (hypoplastic left heart syndrome: AOR = 3.2; 95% CI, 1.3-7.6; and coarctation of the aorta: AOR = 2.7; 95% CI, 1.3-5.6), choanal atresia (AOR = 8.0; 95% CI, 2.7-23.5), transverse limb deficiency (AOR = 2.5; 95% CI, 1.0-5.9), and diaphragmatic hernia (AOR = 2.4; 95% CI, 1.1-5.4). In examining all estimated AORs, one can appreciate the general shift toward positive associations as there are a number of additional elevated associations whose 95% CIs barely included 1: tetralogy of Fallot (AOR = 1.9; 95% CI, 0.8-4.3), small intestinal atresia or stenosis (AOR = 2.6; 95% CI, 0.9-7.5), and craniosynostosis (AOR = 2.0; 95% CI, 0.9-4.5) (Table 3).

Cephalosporins had 1 statistically significant increased OR for atrial septal defects (AOR = 1.9; 95% CI, 1.1-3.2). Quinolones had a limited number of exposures (42 case mothers and 14 control mothers) and had 1 defect associated with reported use: the conotruncal defect tetralogy of Fallot (AOR = 3.7; 95% CI, 1.3-10.5) (Table 3).

Tetracyclines were reported by 36 case mothers and 6 control mothers, and 4 ORs were calculated. For any heart defects, left ventricular outflow tract obstruction defects, septal heart defects, and all oral clefts, the AORs were 2.2 (95% CI, 0.8-5.9), 3.5 (95% CI, 1.0-12.6), 1.8 (95% CI, 0.5-6.2), and 2.0 (95% CI, 0.6-6.7), respectively.

Limiting these analyses to those who did not have a family history reported for the same birth defects increased most ORs, but not by more than 15%. The largest effect of excluding subjects with a family history was seen for hypoplastic left heart syndrome. When subjects with a first-degree family history of heart defects were excluded, the OR increased from 3.2 to 3.6 for exposure to sulfonamides and from 4.2 to 4.8 for exposure to nitrofurantoins.

Data from the National Birth Defects Prevention Study showed that antibacterial drug use during pregnancy is common and increases during early pregnancy. Our study lends support to the established safety profiles for certain classes of antibacterials such as penicillins, erythromycins, and cephalosporins. However, several increased ORs were observed among women taking sulfonamides and nitrofurantoins, indicating a need for additional scrutiny.

Determining the causes of birth defects is problematic. A single defect can have multiple causes, or multiple seemingly unrelated defects may have a common cause. This study could not determine the safety of drugs during pregnancy, but the lack of widespread increased risk associated with many classes of antibacterials used during pregnancy should be reassuring.

Under the current system, the US Food and Drug Administration defines risk categories (A, B, C, D, X) for all drugs based on the level of risk the drug poses to the fetus and the nature of evidence supporting that risk assessment. The categories are as follows: (A) controlled studies do not demonstrate a risk to the human fetus; (B) animal studies do not show risk (no controlled studies in humans), or animal studies have shown adverse effect but controlled studies in humans show no risk; (C) animal studies have shown an adverse effect on the fetus (no controlled studies in humans); (D) investigational studies have shown evidence of human fetal risk; and (X) the drug is contraindicated.⁶ No antibacterials in this study are currently classified in category A or category X.

As this was a retrospective case-control study, it is impossible to determine causal relationships between exposure and birth defect outcome; only associations can be determined. The intent of this specific study was to systematically determine the associations of many different categories of birth defects with common classes of antibacterial exposures that increased the risk of observing spurious associations due to the large number of tests conducted in the analysis. Additional subanalyses are necessary to explore possible mechanisms of increased risk and the association of isolated vs multiple defect types as well as other defect types not included in the National Birth Defects Prevention Study. Defects with at least 4 exposed cases were assessed using logistic regression to get estimates that are relatively stable (as opposed to estimates based on, for instance, 2 exposed cases) and to be able to identify any possible dramatically elevated risks among the rarer defect groups. The number of exposed cases is presented and can inform the reader (along with the width of the 95% CI) as to the confidence in each estimate. An additional limitation is that because our subjects were interviewed 6 weeks to 2 years after the pregnancy, they may not fully or accurately recall all exposures and/or the exact timing of each exposure. This was evidenced by the large proportion of women who reported antibacterial use but could not provide a product name. Selection bias can also be of concern. However, the participation rates are similar for cases and controls, and in a recent investigation it was shown that participating controls are similar to those not participating.⁷

Another limitation is that it is not possible to determine whether the birth defect is associated with the antibacterials used or the underlying infection. This is a difficult analytical challenge because particular types of infections tend to be treated with the same antibacterials, reducing the number of cases exposed to the infection and treated with different antibacterials. Additionally, those receiving different classes of antibacterials for the same type of indication (eg, sinus infections) may have more aggressive or different types of infections. For instance, both sulfonamides and nitrofurantoins are used for the treatment of urinary tract infections. However, in our analysis they did not seem to be associated with the same birth defects; the only birth defect in common between the 2 groups was hypoplastic left heart syndrome. This suggests that although these classes of antibacterials have similar indications, they do not appear to be associated with similar types of birth defects. Although we found that the urinary germicide nitrofurantoin (category B) was associated with cleft lip with cleft palate, other studies did not indicate a teratogenic potential of nitrofurantoin use during early gestation.^8,9

Penicillins (category B), cephalosporins (category B), and other β-lactam antibacterials target bacterial cell wall–synthesizing enzymes, which are not found in mammalian cells and may account for the relatively low toxicity of these antibacterials. Penicillins were the most commonly reported single class of antibacterials; our results showed only 1 elevated OR in a defect with limited sample size, which is consistent with the long-standing safety profile of these drugs.¹⁰ Similarly, in the literature most cephalosporins have not been associated with many birth defects,⁸ and our study showed a single elevated risk of a specific category of heart defects (atrial sepal defects: AOR = 1.9; 95% CI, 1.1-3.2). Erythromycins are the first and most widely known of the macrolide class of antibacterials, which includes clarithromycin and azithromycin. Erythromycins (category B) and other macrolides are widely used with rarely reported serious adverse events.⁸ Our results show elevated AORs for 2 distinct birth defects.

Potential fetal adverse effects are well known for several antibacterial classes. It is encouraging to note that the reported use of category C and D antibacterials such as aminoglycosides (category C or D), chloramphenicol (category C), and tetracyclines (category D) that have been associated with possible negative effects was extremely low to none at all. For example, aminoglycosides have been known to carry a small risk of fetal ototoxic effects,⁸ and problems attributable to the use of tetracycline (category D) during pregnancy include adverse effects on fetal bones and possible birth defects.^3,8 It is encouraging that these antibacterials were not commonly reported by pregnant women in our study.

Sulfonamides (category C or D) are among the oldest classes of antibacterials, and resistance has developed. Sulfonamides have been shown to be teratogenic in animal studies, although it is unclear whether sulfonamides without trimethoprim pose a significant risk.^2,3,11-13 Czeizel et al¹¹ indicated higher odds of cardiovascular malformations (adjusted prevalence OR = 3.5; 95% CI, 1.9-6.4) and clubfoot (adjusted prevalence OR = 2.6; 95% CI, 1.1-6.2) in infants born to mothers receiving sulfonamide treatment. Trimethoprim (category C) is a diaminopyrimidine that is commonly given in combination with the sulfonamide sulfamethoxazole for treatment of urinary infections. There have been concerns about the use of trimethoprim-sulfamethoxazole (category C) and possible teratogenicity in early pregnancy as well as induction of hyperbilirubinemia near term. The 2 antibacterial drugs operate synergistically via a sequential blockade of 2 steps in the biosynthesis of reduced folates.¹³ In a study of 2296 Michigan Medicaid recipients, first-trimester trimethoprim exposure was associated with an increased risk for birth defects, particularly cardiovascular defects.³ In a retrospective study, treatment with co-trimoxazole during the second and third months of pregnancy was associated with an increased risk for cardiovascular malformations and particularly multiple congenital abnormalities including defects of the urinary tract and cardiovascular system. In 2 studies by Hernández-Díaz et al,^13,14 the risks for cardiovascular defects and oral clefts were increased 2- to 3-fold among women who took trimethoprim with sulfonamide during the periconceptional period; a later study found trimethoprim to be associated with an increased risk for neural tube defects.¹⁴ These studies indicate a possible role for the folate pathway in the association between these medications and birth defects. We did adjust for folic acid use in our analyses, but additional detailed analysis of the relationship between combined sulfonamide and trimethoprim exposure and birth defects is warranted.

The quinolones (category C) comprise a large and expanding group of synthetic antibacterial agents. Fluoroquinolones are a subset of later quinolones that can have adverse effects on growing cartilage and may cause fetal arthropathies.⁸ Although to our knowledge there have been no well-controlled studies of the safety of most quinolones in pregnant women, a recent review did not show evidence of teratogenicity in animal studies.¹⁰ Nevertheless, our results suggested the possibility of increased risk for certain cardiac defects. Quinolones are currently not recommended for use during pregnancy unless there are overriding reasons for their use.

Antibacterial exposure was common among pregnant women participating in the National Birth Defects Prevention Study, with use increasing during the first trimester—the critical developmental stage. Our study lends support to the established safety profiles for certain classes of antibacterials such as penicillins, erythromycins, and cephalosporins. Overall, there were limited numbers of associations between the included classes of antibiotics and categories of major birth defects. However, several increased ORs were observed among women taking sulfonamides and nitrofurantoins, indicating a need for additional scrutiny.

Correspondence: Krista S. Crider, PhD, Division of Birth Defects and Developmental Disabilities, National Center on Birth Defects and Developmental Disabilities, Centers for Disease Control and Prevention, 1600 Clifton Rd, MS E-86, Atlanta, GA 30333 (kcrider@cdc.gov).

Accepted for Publication: July 15, 2009.

Author Contributions: Drs Crider, Cleves, Reefhuis, and Hu had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Crider, Cleves, Berry, Hobbs, and Hu. Acquisition of data: Reefhuis and Hobbs. Analysis and interpretation of data: Crider, Cleves, Reefhuis, Hobbs, and Hu. Drafting of the manuscript: Crider, Reefhuis, and Hu. Critical revision of the manuscript for important intellectual content: Crider, Cleves, Reefhuis, Berry, Hobbs, and Hu. Statistical analysis: Crider, Cleves, and Reefhuis. Obtained funding: Reefhuis, Berry, and Hobbs. Administrative, technical, and material support: Reefhuis and Hu. Study supervision: Reefhuis, Berry, and Hu.

Financial Disclosure: None reported.

Funding/Support: The National Birth Defects Prevention Study is funded by cooperative agreement PA 01081 from the Centers for Disease Control and Prevention.

Disclaimer: The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

Additional Information: Coding of drug information in the National Birth Defects Prevention Study used the Slone Drug Dictionary, under license from the Slone Epidemiology Center at Boston University, Boston, Massachusetts.

Additional Contributions: We acknowledge the many scientists, participants, and collaborators who work with the National Birth Defects Prevention Study.