Safety of Antimicrobials During Pregnancy: A Systematic Review of Antimicrobials Considered for Treatment and Postexposure Prophylaxis of Plague

Abstract

Background

The safety profile of antimicrobials used during pregnancy is one important consideration in the decision on how to treat and provide postexposure prophylaxis (PEP) for plague during pregnancy.

Methods

We searched 5 scientific literature databases for primary sources on the safety of 9 antimicrobials considered for plague during pregnancy (amikacin, gentamicin, plazomicin, streptomycin, tobramycin, chloramphenicol, doxycycline, sulfadiazine, and trimethoprim-sulfamethoxazole [TMP-SMX]) and abstracted data on maternal, pregnancy, and fetal/neonatal outcomes.

Results

Of 13 052 articles identified, 66 studies (case-control, case series, cohort, and randomized studies) and 96 case reports were included, totaling 27 751 prenatal exposures to amikacin (n = 9), gentamicin (n = 345), plazomicin (n = 0), streptomycin (n = 285), tobramycin (n = 43), chloramphenicol (n = 246), doxycycline (n = 2351), sulfadiazine (n = 870), and TMP-SMX (n = 23 602). Hearing or vestibular deficits were reported in 18/121 (15%) children and 17/109 (16%) pregnant women following prenatal streptomycin exposure. First trimester chloramphenicol exposure was associated with an elevated risk of an undescended testis (odds ratio [OR] 5.9, 95% confidence interval [CI] 1.2–28.7). Doxycycline was associated with cardiovascular malformations (OR 2.4, 95% CI 1.2–4.7) in 1 study and spontaneous abortion (OR 2.8, 95% CI 1.9–4.1) in a separate study. First trimester exposure to TMP-SMX was associated with increased risk of neural tube defects (pooled OR 2.5, 95% CI 1.4–4.3), spontaneous abortion (OR 3.5, 95% CI 2.3–5.6), preterm birth (OR 1.5, 95% CI 1.1–2.1), and small for gestational age (OR 1.6, 95% CI 1.2–2.2). No other statistically significant associations were reported.

Conclusions

For most antimicrobials reviewed, adverse maternal/fetal/neonatal outcomes were not observed consistently. Prenatal exposure to streptomycin and TMP-SMX was associated with select birth defects in some studies. Based on limited data, chloramphenicol and doxycycline may be associated with adverse pregnancy or neonatal outcomes; however, more data are needed to confirm these associations. Antimicrobials should be used for treatment and PEP of plague during pregnancy; the choice of antimicrobials may be influenced by these data as well as information about the risks of plague during pregnancy.

Plague, pregnancy, antimicrobials, safety

Plague has occurred since ancient times, from major pandemics in 541–750 AD and in 1346–1353 AD, to modern outbreaks in sub-Saharan Africa, including a 2017 outbreak in Madagascar [1–6]. In the United States, plague typically occurs in western states with an average of 7 cases per year between 2000 and 2017 [7]. Yersinia pestis, the causative bacterium of plague, is classified as a bioterrorism agent and in the case of an intentional wide-area release, can cause a high degree of morbidity and mortality [8]. In 2000, a working group of 25 experts published consensus recommendations for antimicrobial treatment and postexposure prophylaxis (PEP) of plague [9]. Since then, additional data have become available, including animal efficacy studies that supported the Food and Drug Administration’s (FDA) approval of ciprofloxacin and newer antimicrobials (eg, levofloxacin, moxifloxacin) for plague.

The choice of antimicrobials for treatment and PEP of plague during pregnancy may be the same as those recommended for nonpregnant individuals or may differ depending on the maternal/fetal risk-benefit profile of each antimicrobial. Therefore, we conducted a systematic review of the literature to evaluate the risks of adverse maternal, pregnancy, and fetal/neonatal outcomes from prenatal exposure to antimicrobials considered for treatment or PEP of plague. These data, as well as information regarding the risks of plague during pregnancy, will be the basis for informing the Centers for Disease Control and Prevention’s (CDC) clinical guidelines for plague.

METHODS

Protocol and Registration

The systematic review on the safety of 9 antimicrobials during pregnancy was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement and registered with the International Prospective Register of Systematic Reviews (PROSPERO; number CRD42018095490) [10].

Selection of Antimicrobials

Antimicrobials considered for treatment and PEP of plague include amikacin, gentamicin, plazomicin, streptomycin, tobramycin, chloramphenicol, doxycycline, sulfadiazine, and trimethoprim/sulfamethoxazole (TMP-SMX). Although some of these antimicrobials may not currently be widely distributed in the United States, they were included because of their global availability (ie, streptomycin and chloramphenicol). Fluoroquinolones, also considered for plague treatment and PEP, were excluded from this review due to 2 previously published systematic reviews on this class of antimicrobials [11, 12].

Information Sources and Searches

A CDC librarian searched CINAHL, Cochrane Library, Embase, Medline, and Scopus for English language articles from inception of each database through February 2018. Search terms focused on the generic antimicrobial name combined with pregnancy, fetal, or neonatal terms. The full search strategy may be accessed here: https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=95490. We also searched https://clinicaltrials.gov for relevant publications, hand searched electronic resources that summarize medication safety during pregnancy, searched reference lists from review articles for additional primary sources, and reviewed the US FDA package inserts for each antimicrobial [13,–17].

Eligibility Criteria

Aggregate studies (ie, case-control, case series, cohort, and randomized studies) and case reports were included if they: 1) were peer-reviewed and contained data related to the use of amikacin, gentamicin, plazomicin, streptomycin, tobramycin, chloramphenicol, doxycycline, sulfadiazine, or TMP-SMX in singleton pregnancies; and 2) contained maternal, pregnancy, or fetal/neonatal outcomes. Aggregate studies were included if prenatal exposure to relevant antimicrobial(s) were described with or without information on dosage or duration of antimicrobial exposure, whereas case reports required information on dosage and/or duration. Intrapartum antimicrobial exposures (defined as within 72 hours of delivery) to chloramphenicol and sulfadiazine or TMP-SMX were included to ascertain risk of gray baby syndrome and jaundice/kernicterus, respectively. Intrapartum administration of all other antimicrobials and postpartum exposures were excluded. Articles were excluded if they were animal or in vitro studies, policy guidelines, reviews, or only reported drug class (ie, not delineated further by specific antimicrobial).

Outcomes collected included fetal/neonatal outcomes (eg, birth defects, low birthweight, small for gestational age, preterm birth, neonatal death), pregnancy outcomes (eg, spontaneous abortion, induced abortion, intrauterine fetal death [IUFD]), and maternal outcomes (ie, antimicrobial-related adverse events [AEs], maternal death). Fetal/neonatal AEs and birth defects associated with known etiology (ie, prenatal toxoplasmosis and genetic polymorphism-associated hearing loss) were excluded to avoid confounding contributed by these etiologies [18, 19].

Article Selection and Data Abstraction

Seven reviewers (2 independent reviewers for each article) screened search results by title and abstract (C.P., D.M.D., E.T., E.Y., H.K., P.S., P.Y.). Any discordances were discussed and adjudicated. Selected articles underwent full text review by a single reviewer to assess eligibility. Data from full text reviewed articles were abstracted using a standard form developed in a Microsoft Access database. A second, independent reviewer examined the abstracted primary data from included articles to ensure accuracy and completeness. Discrepancies were discussed and adjudicated.

Data Synthesis and Analysis

Data cleaning and descriptive analysis for aggregate studies and case reports were conducted in Microsoft Access and R [20]. We conducted meta-analyses for antimicrobials with sufficient information by calculating the pooled odds ratios (OR) from case-control studies to estimate the effect of antimicrobial exposures on outcomes using Open Meta-Analyst software [21]. Estimates of ORs were pooled across studies using a DerSimonian and Laird random-effects model and heterogeneity was quantified using the I² statistic with high heterogeneity signified by I² value >50%.

For case reports, the frequency of outcomes was calculated by combining patients with prenatal exposures to the same antimicrobial. Duration of antimicrobial exposure was dichotomized using 14 days as the cutoff because duration of plague treatment and PEP is expected to be no greater than 14 days. Subgroup analysis was performed by comparing the effect of short (<14 days) and long duration (≥14 days) antimicrobial exposure on fetal/neonatal, pregnancy, and maternal outcomes; comparisons were performed using χ ² test, or if frequencies were <5, Fisher exact test.

Quality Assessment

Quality of the included articles was evaluated using a 10-point modified Newcastle-Ottawa Scale assessing selection, comparability, and outcome characteristics for aggregate studies and individual case reports [22] (Supplementary Table 1). One reviewer (Z.R.) conducted primary assessment of the risk of bias, which was reviewed by a second reviewer (E.Y., H.K., or P.Y.). Scores of 1–4, 5–6, and 7–10 represented very serious, serious, and low risk of bias, respectively.

RESULTS

A total of 13 052 articles were identified from the literature search. After initial exclusion of duplicates, non-English articles, and articles excluded from a related, previously conducted systematic review, titles and abstracts of 9097 articles were screened (Figure 1) [12]. Following title/abstract screening, 767 full text articles were reviewed. Overall, 162 articles were included following full-text review: 66 aggregate studies (14 case-control, 9 case series, 36 cohort, and 7 randomized studies) and 96 case reports (Figure 1, Tables 1–2, Supplementary Table 2).

Figure 1.

Flow diagram of article selection.

*Articles excluded from previous systematic review by Meaney-Delman, et al., 2013 were also excluded in this systematic review.

**Articles describing intrapartum administration of chloramphenicol and sulfadiazine or trimethoprim-sulfamethoxazole were included to assess potential risks of gray baby syndrome or jaundice/kernicterus, respectively.

*** Aggregate studies included case-control, case series, cohort, and randomized studies. Case reports (including patients with detailed clinical information per patient in case series) may include more than 1 relevant antimicrobial prenatal exposure.

Table 1.

Number of Included Aggregate Studies and Case Reports, Number of Prenatal Exposures, and Trimester of Exposure, by Antimicrobial

	Aggregate Studiesa			Case Reportsb
	No. of Studies	No. of Prenatal Exposures	Trimester of Exposurec	No. of Prenatal Exposures	Trimester of Exposurec
Aminoglycosides
Amikacin	2	5	First: 1 (20%) Second: 0 (0%) Third: 0 (0%) Other: 4 (80%) Not reported: 0 (0%)	4	First: 1 (25%) Second: 0 (0%) Third: 0 (0%) Other: 3 (75%) Not reported: 0 (0%)
Gentamicin	11	310	First: 79 (25%) Second: 37 (12%) Third: 2 (<1%) Other: 49 (16%) Not reported: 143 (46%)	35	First: 4 (11%) Second: 14 (40%) Third: 15 (43%) Other: 2 (6%) Not reported: 0 (0%)
Plazomicin	0	0	…	0	…
Streptomycin	10	226	First: 27 (12%) Second: 3 (1%) Third: 28 (12%) Other: 156 (69%) Not reported: 12 (5%)	59	First: 5 (8%) Second: 15 (25%) Third: 20 (34%) Other: 17 (29%) Not reported: 2 (3%)
Tobramycin	4	38	First: 6 (16%) Second: 11 (29%) Third: 21 (55%) Other: 0 (0%) Not reported: 0 (0%)	5	First: 0 (0%) Second: 3 (60%) Third: 2 (40%) Other: 0 (0%) Not reported: 0 (0%)
Chloramphenicol	8	240	First: 2 (1%) Second: 1 (<1%) Third: 0 (0%) Other: 15 (6%) Not reported: 222 (93%)	6	First: 0 (0%) Second: 1 (17%) Third: 4 (67%) Other: 1 (17%) Not reported: 0 (0%)
Doxycycline	8	2348	First: 314 (13%) Second: 45 (2%) Third: 28 (1%) Other: 1896 (81%) Not reported: 65 (3%)	3	First: 1 (33%) Second: 2 (67%) Third: 0 (0%) Other: 0 (0%) Not reported: 0 (0%)
Sulfonamides
Sulfadiazine	8	860	First: 5 (<1%) Second: 2 (<1%) Third: 9 (1%) Other: 158 (18%) Not reported: 686 (80%)	10	First: 1 (10%) Second: 3 (30%) Third: 3 (30%) Other: 3 (30%) Not reported: 0 (0%)
Trimethoprim-sulfamethoxazole	27	23 579	First: 7430 (32%) Second: 77 (<1%) Third: 41 (<1%) Other: 3295 (14%) Not reported: 12 736 (54%)	23	First: 5 (22%) Second: 10 (43%) Third: 4 (17%) Other: 4 (17%) Not reported: 0 (0%)

^aArticles describing prospective and observational studies including case-control, case series, cohort, and randomized studies. An article may be listed more than once if it contained primary data on more than 1 relevant antimicrobial.

^bAn article that was abstracted as a case report may have included description of more than 1 pregnant woman and more than 1 relevant antimicrobial. A total of 96 unique case reports described 145 prenatal exposures.

^cTrimester listed as antimicrobial exposure during first, second, or third trimester only. Other denotes prenatal exposure during >1 trimester (eg, first and second, second and third) or during intrapartum (for chloramphenicol, sulfadiazine, or trimethoprim-sulfamethoxazole).

Aminoglycosides

Amikacin

We identified 2 studies containing aggregate data on 5 pregnant women and 4 case reports describing prenatal amikacin use across all trimesters of pregnancy [25–30]. No consistent pattern of adverse maternal, pregnancy, and fetal/neonatal outcomes were reported.

Gentamicin

We identified 11 studies containing aggregate data on 310 gentamicin-treated pregnant women [26, 31–40] and 35 gentamicin-treated pregnant women from case reports [28, 41–72].

Birth Defects

One population-based case-control study from the Hungarian Case-Control Surveillance for Congenital Abnormalities (HCCSCA) compared aminoglycoside use between pregnant women who had newborns without any birth defects (control group) and pregnant women who had fetuses or newborns with birth defects [31]. Gentamicin administered primarily during the second or third gestational months was not significantly associated with birth defects (OR 1.7, 95% CI .9–3.2).

A prospective cohort study of 25 gentamicin exposures did not find a significant association with neonatal hearing deficit (OR 1.8, 95% CI .7–4.5) [33]. One retrospective cohort study found no audiologic deficits among 40 infants during routine hearing screening [32].

A combined total of 52 gentamicin exposures during the first or second trimesters in 2 separate prospective cohort studies reported 1 fetus or infant each with congenital luxation of the knee, club foot, or choroid plexus cyst [34, 35]. A retrospective study of 26 first trimester gentamicin exposures reported 1 neonate (4%) with a dilated renal pelvis without caliectasis, compatible with bilateral hydronephrosis, resulting in a neonatal death 4 hours after birth [26]. The infant’s mother received a combination of antimicrobials, including ciprofloxacin, gentamicin, and methenamine for 7–10 days for treatment of a urinary tract infection.

In a case series of 6 gentamicin exposures during the second and third trimesters, no reports of birth defects or other adverse fetal/neonatal outcomes were reported [36]. One case report described a neonate born with renal dysplasia characterized by small kidneys with increased echotexture on ultrasound [51].

Pregnancy and Fetal/Neonatal Outcomes

Two randomized studies with a combined total of 90 gentamicin-treated pregnant women compared to untreated women found no increased risk of IUFD [37, 38]. A retrospective cohort study reported no neonatal deaths among 15 gentamicin exposures during the first 2 trimesters of pregnancy [39].

Among 35 patients from case reports [28, 41–72], the median duration of gentamicin exposure in utero was 8 days (range 1–56 days). The percent of any adverse fetal/neonatal outcomes associated with ≥14 days (1/5, 20%) of use was compared to <14 days (1/18, 6%) of use, but was not statistically significant (P value = .4).

Maternal Outcomes

Transient nephrotoxicity characterized by increased serum creatinine by 1.5–1.6 mg/dl occurred in 2/18 pregnant women treated with 5 days of gentamicin in a nonrandomized study; none resulted in permanent nephrotoxicity [40]. No maternal cases of nephrotoxicity or ototoxicity were reported among 35 patients from case reports [28, 41–72].

Plazomicin

We did not identify any articles on the safety of plazomicin during pregnancy.

Streptomycin

We identified 10 studies containing aggregate data on 226 streptomycin-treated pregnant women [31, 34, 73–80] and 59 streptomycin-treated pregnant women from case reports [29, 81–112].

Birth Defects

Combined data from 121 prenatal streptomycin exposures in 2 cohort studies and 2 case series evaluated ototoxicity in children; 18/121 (15%) of children had either hearing or vestibular deficits based on audiogram and caloric testing [74–76, 79]. The previously mentioned population-based case-control study from HCCSCA found no significant association between streptomycin administered primarily in the second to third months of pregnancy and birth defects (OR 1.3, 95% CI .4–4.6) among 9 streptomycin exposures (1–2g daily for 5–60 days) [31]. No birth defects or adverse fetal/neonatal outcomes were reported among a total of 17 additional pregnancies with streptomycin use in 2 other cohort studies and 1 other case series [34, 73, 78].

Among 59 prenatal streptomycin-exposed pregnancies described in case reports [29, 81–112], the proportion of any birth defect or adverse fetal/neonatal outcome among pregnancies with ≥14 days of use was 13/43 (30%), compared to <14 days (1/4, 25%) of use. This was not statistically significant (P value = 1). Neonatal hearing deficits were reported in 4 infants, including severe bilateral sensory neural hearing loss in 2 children [90, 97, 99].

Maternal Outcomes

From 2 cohort studies and 2 case series that evaluated ototoxicity following prenatal streptomycin exposure, 17/109 (16%) women had either hearing or vestibular deficits based on audiogram and caloric testing [74–76, 79]. Two other cohort studies reported that no maternal AEs occurred among 79 streptomycin-treated pregnant women; however, 68 of these cases involved use of a single-dose of streptomycin [77, 80]. Among 59 prenatal streptomycin-exposed pregnancies from case reports, hearing and/or vestibular dysfunction were reported following streptomycin use in 8 pregnant women (median duration 77 days, range 3 to approximately 100 days) [90, 96, 97, 99, 105, 109, 111].

Tobramycin

We identified 4 studies containing aggregate data on 38 tobramycin-treated pregnant women [31, 113–115] and 5 tobramycin-treated pregnant women from case reports [116–119]. In the previously mentioned HCCSCA case-control study, tobramycin administered primarily during the second or third months of gestation was not associated with birth defects (OR .8, 95% CI .2–3.9); however, this included only 6 tobramycin-treated pregnancies [31]. In 3 separate, prospective cohort studies conducted by the same investigators totaling 32 pregnant women treated with tobramycin and other antimicrobials during the second and third trimesters, no birth defects or adverse fetal/neonatal outcomes were reported [113–115].

Only 1 of these cohort studies reported information on maternal AEs, which found normal renal and subjectively normal auditory functions in 18 tobramycin-treated pregnant women [114]. Among 5 prenatal tobramycin-exposed pregnancies from case reports, no cases of maternal or neonatal nephrotoxicity or ototoxicity were reported [116–119].

Chloramphenicol

We identified 8 studies containing aggregate data on 240 chloramphenicol-treated pregnant women [120–127] and 6 chloramphenicol-treated pregnant women from case reports [28, 112, 128–130]. One case-control study that included 103 prenatal chloramphenicol exposures with 2 and 7 exposures among cases and controls in the first trimester, respectively, reported an elevated risk of an undescended testis (OR 5.9, 95% CI 1.2–28.7) [120]. Two cohort studies with 22 prenatal chloramphenicol exposures reported genu valgum, hiatal hernia, renal agenesis, strabismus, and an undescended testis in 5 infants [121, 122]. Three additional cohort studies, 1 case-control study, and 1 case series reported no evidence of birth defects or gray baby syndrome among a total of 115 prenatal chloramphenicol exposures [123–127].

Among case reports describing 6 prenatal chloramphenicol exposures, 1 neonate developed gray baby syndrome characterized by abdominal distension, vomiting, hypotension, and cyanosis that resulted in death following intrapartum and postnatal exposure to chloramphenicol [130]. No other reports of gray baby syndrome following in utero exposure to chloramphenicol were found [28, 112, 128, 129].

Doxycycline

We identified 8 studies containing aggregate data on 2348 doxycycline-treated pregnant women [34, 125, 131–136] and 3 doxycycline-treated pregnant women from a case series [137].

Birth Defects

In 1 population-based case-control study comparing 164 first trimester doxycycline-exposed pregnancies to 124 469 unexposed pregnancies, the odds of cardiovascular birth defects (cardiac and circulatory system birth defects) in infants was found to be 2- to 3-fold higher among infants born to exposed versus unexposed women (OR 2.4, 95% CI 1.2–4.7) [135]. Subset analyses revealed statistically significant ORs for cardiac birth defects exclusively (OR 2.5, 95% CI 1.2–5.0) and ventricular/atrial septal defect (OR 3.2, 95% CI 1.6–6.5) [135]. Two population-based case-control studies of the HCCSCA did not identify a statistically significant risk of birth defects from doxycycline exposure during the second and third gestational months (OR 1.8, 95% CI .7–5.0) or neural tube defects (NTD) from doxycycline exposure during the second gestational month (OR 2.0, 95% CI .3–15.5) [125, 132].

A retrospective cohort study of neonates born to doxycycline-exposed (n = 1691) and unexposed pregnant women (n = 3400) did not find an elevated risk of cardiac birth defects when treated during the first 4 lunar months (defined as the period from last menstrual period through 112 days) (relative risk [RR] 0.9, 95% CI .5–1.6) [131]. The overall risk of birth defects was similar between infants born to exposed versus unexposed women (relative risk [RR] 0.9, 95% CI .6–1.2). In another cohort of 41 first trimester doxycycline exposures during pregnancy, 1 infant was reported to have diastasis recti [34].

Pregnancy and Fetal/Neonatal Outcomes

An analysis of HCCSCA data examining the effects of prenatal doxycycline exposure during any trimester on preterm birth and low birth weight did not identify a significant association with preterm birth (OR 0.4, 95% CI .1–1.4) or low birth weight (OR 1.2, 95% CI .5–1.8) [134]. In a different population-based case-control study, doxycycline use was significantly associated with spontaneous abortion (OR 2.8, 95% CI 1.9–4.1) [136]. In a cohort study of 53 women treated with doxycycline during the first 2 trimesters of pregnancy compared to untreated women, lower rates of spontaneous abortion (11% vs. 21%), preterm birth (2% vs. 4%), and neonatal death (2% vs. 5%) were reported; these comparisons were not statistically significant (P values >.05) [133].

No specific patterns of adverse maternal, pregnancy, and fetal/neonatal outcomes were identified from case reports because data were limited to 1 case series describing 3 prenatal doxycycline exposures [137].

Sulfonamides

Sulfadiazine

We identified 8 studies containing aggregate data on 860 sulfadiazine-treated pregnant women [138–145] and 10 sulfadiazine-treated pregnant women from case reports [87, 146–154].

Pregnancy and Fetal/Neonatal Outcomes

A retrospective cohort study of 94 sulfadiazine women treated any time during pregnancy reported 9/94 (10%) infants with jaundice; none progressed to kernicterus [141]. No cases of jaundice or kernicterus were reported from 2 additional cohort studies of 47 exposures [143, 144], 1 case series of 5 exposures [142], and 10 exposures from case reports [87, 146–154]. No other patterns of fetal/neonatal outcomes were found among included articles.

Maternal Outcomes

The largest aggregate study for sulfadiazine was a retrospective cohort of 685 pregnant women treated with sulfadiazine, pyrimethamine, and folinic acid, which reported no serious maternal AEs related to sulfadiazine, other than a single hypersensitivity reaction among 140 women treated who had available information on AEs [138]. In a prospective cohort study of 9 pregnant woman treated with sulfadiazine in the third trimester, 1 woman was reported to have occasional macrocytic anemia, successfully treated with folinic acid [140]. In a case series of 18 sulfadiazine-treated pregnant women, 1 patient (6%) had a cutaneous allergic reaction following 3 weeks of sulfadiazine [139]. No specific patterns in maternal adverse events were identified from case reports describing 10 sulfadiazine exposures [87, 146–154].

Trimethoprim-Sulfamethoxazole

We identified 27 studies containing aggregate data on 23 579 TMP-SMX treated pregnant women [26, 34, 121, 125, 135, 145, 155–175] and 23 TMP-SMX treated pregnant women from case reports [47, 152, 176–186].

Birth Defects

Seven case-control studies reported on the risk of birth defects. The risk of NTDs was higher in women exposed to TMP-SMX during the first trimester compared with unexposed women across 3 case-control studies (pooled OR 2.5, 95% CI 1.4–4.3, I² 0%) (Figure 2) [145, 155, 159]. In 1 of these 3 studies that included 11 sulfamethoxazole and 2 sulfadiazine exposures during the first trimester, there was a significantly greater association with NTDs among exposed women not taking folic acid supplementation (OR 13.3, 95% CI 2.9–61.4), whereas the OR was 1.2 (95% CI .1–12.7) for exposed women taking daily folic acid [145]. Although 1 case-control study reported an increased risk of cardiac birth defects in children born to TMP-SMX exposed versus unexposed women [159], 2 other case-control studies found no increased risk of cardiac birth defects from first trimester TMP-SMX exposure, resulting in a nonstatistically significant pooled OR of 1.2 (95% CI .6–2.5, I² 68%) across the 3 studies (Figure 2) [135, 158, 159]. A different case-control study reported an increased risk of cardiac birth defects (OR 5.39, 95% CI 1.42–20.49), but the trimester of exposure was not reported [162].

Figure 2.

Pooled estimates of effect for birth defects following exposure to trimethoprim sulfamethoxazole during pregnancy. Forrest plots were only able to be completed for trimethoprim-sulfamethoxazole due to the sufficient number of available aggregate studies with a comparison group of unexposed pregnant women. Abbreviation: CI, confidence interval.

Pregnancy and Fetal/Neonatal Outcomes

Several studies found significant associations between prenatal TMP-SMX exposure and pregnancy and fetal/neonatal outcomes. One population-based case-control study reported an increased risk of spontaneous abortion among 7039 first trimester TMP-SMX exposed women (OR 3.5, 95% CI 2.3–5.6) [167]. A cohort study comparing 447 women exposed to TMP-SMX any time during pregnancy to 14 537 women not exposed to any drugs in the former FDA pregnancy categories C, D, or X reported an elevated risk of preterm birth (OR 1.5, 95% CI 1.1–2.1) [174]. A case-control study, which included 214 second and third trimester exposures to TMP-SMX, found an increased risk of infants born small for gestational age (OR 1.6, 95% CI 1.2–2.2) [170]. In a cohort study of TMP-SMX exposure in utero during the second and third trimesters, 1/67 (1%) was hospitalized with neonatal jaundice; no cases of kernicterus were reported [171]. Rates of spontaneous abortion, IUFD, preterm birth, neonatal death, and low birth weight generally varied across cohort studies [26, 121, 156, 157, 160, 161, 163,–166, 168, 169, 171, 172, 174].

No cases of NTDs, jaundice, or kernicterus were reported following 23 in utero exposures from case reports and no patterns of fetal/neonatal outcomes were identified [47, 152, 176,–186].

Maternal Outcomes

Nine articles containing aggregate data reported on maternal AEs from prenatal TMP-SMX exposure; no serious maternal AEs were reported [121, 156, 160, 163, 165, 166, 168, 173, 175]. Among 23 exposures during pregnancy from case reports, the only reported maternal AE was 1 case of esophageal candidiasis in a pregnant woman with human immunodeficiency virus (HIV) infection [47, 152, 176,–186].

Risk of Bias

Risk of bias assessment of aggregate studies revealed that 29 (44%) articles had low risk of bias, 20 (30%) articles had serious risk of bias, and 17 (26%) articles had very serious risk of bias (Supplementary Table 2). Case reports are generally considered highly biased reports due to publication bias. Detailed assessment of the quality of case reports, accounting for those that described >1 patient, indicated a higher proportion with serious or very serious risk of bias compared to aggregate studies. When comparing case reports across the review, 41 (28%) had low risk of bias, 76 (52%) had serious risk of bias, and 28 (19%) had very serious risk of bias.

DISCUSSION

This systematic review summarizes data on the safety of 9 antimicrobials considered for treatment and PEP of plague during pregnancy and will inform CDC clinical guidelines for plague. For most antimicrobials in this review, adverse maternal/fetal/neonatal outcomes were not observed in a consistent pattern, although data were generally limited for all antimicrobials reviewed. Antimicrobials should be used for treatment and PEP of plague during pregnancy; the choice of antimicrobials may be influenced by these data along with considerations of the risks of plague as described in the systematic review of plague during pregnancy [Fleck-Derderian et al in this CID supplement]. Our review suggests prenatal exposure to streptomycin may be associated with a potential risk of maternal/neonatal hearing and/or vestibular deficits, particularly with longer duration of exposure. In addition, TMP-SMX appears to carry a potential risk of NTDs with first trimester exposure. Although chloramphenicol exposure was associated with undescended testis in 1 study and doxycycline exposure was associated with cardiovascular birth defects in 1 study and spontaneous abortion in another study, these associations were based on limited data in single studies and not substantiated by other studies included in our review. Amikacin, gentamicin, tobramycin, and sulfadiazine during pregnancy were not associated with adverse maternal/fetal/neonatal outcomes. However, the limited number of prenatal exposures may affect the ability to detect rare associations. Fluoroquinolones, which would be considered for treatment and PEP of plague in pregnant women, were not included in our systematic review due to the existence of systematic reviews on the safety of fluoroquinolones [11, 12]. The clinical guidelines for plague will take into consideration all 3 systematic reviews.

A pattern of maternal and neonatal hearing and/or vestibular deficits following streptomycin prenatal exposure was reported in several studies and case reports, most of which were for treatment of tuberculosis (TB) and often involved prolonged streptomycin therapy in combination with other medications throughout pregnancy. Streptomycin-related articles included in our review were primarily published before 1970 when streptomycin was more widely used in the United States. Most pregnant women exposed to streptomycin (>60%) were treated for ≥14 days (range, 1–280 days) in aggregate studies. The deficits identified may also be related to more pregnancies exposed to streptomycin beginning in the first trimester because of the longitudinal nature of the treatment. Because the expected duration of therapy for treatment and PEP of plague is expected to be limited to <14 days, the potential risks of streptomycin use for treatment and PEP of plague during pregnancy may be lower.

No cases of maternal or fetal nephrotoxicity or ototoxicity were identified from prenatal exposure to amikacin, gentamicin, and tobramycin; however few articles were identified and the durations of treatment were generally shorter compared to streptomycin. Although no articles describing plazomicin exposure during pregnancy were identified, clinical trials conducted in nonpregnant adults have reported nephrotoxicity and the risk of ototoxicity cannot be excluded based on limited data [187].

First-trimester prenatal exposure of TMP-SMX was associated with a >10-fold higher risk of NTDs among women not taking folic acid supplementation, but this risk was mitigated by daily folic acid supplementation [145]. This is consistent with the mechanism of action of TMP-SMX as a folic acid inhibitor interrupting DNA biosynthesis or methylation required for proper closure of the neural tube during embryogenesis [188]. However, included studies of TMP-SMX were generally for therapy ≥14 days in pregnant women with comorbidities (eg, HIV-positive in malaria-endemic areas). Sulfonamides carry a potential risk of bilirubin displacement and kernicterus [13]; however, in our review we did not find cases of kernicterus among TMP-SMX and sulfadiazine prenatal exposures. Systematic review and meta-analysis of TMP-SMX that included other sulfonamides found no reports of neonatal kernicterus attributable to maternal ingestion of sulfonamides [189].

An association between chloramphenicol prenatal exposure and undescended testes was reported by one population-based case-control study; however, only a small number of chloramphenicol exposures were in the case and control groups. Although one case report described gray baby syndrome primarily due to postnatal, iatrogenic exposure to supratherapeutic chloramphenicol doses, our review did not find gray baby syndrome resulting from in utero chloramphenicol exposure [130, 190]. However, given the limited data available, the potential for gray baby syndrome to occur with prenatal chloramphenicol exposure cannot be excluded, and the risk of undescended testes has not been verified in other studies.

The elevated risk for cardiac birth defects and spontaneous abortion reported for doxycycline were limited to single population-based case-control studies and not confirmed by other studies. Thus, the data are too limited to conclude that there is a risk of birth defects and adverse pregnancy and fetal/neonatal outcomes. Of note, we only identified 8 studies on use of doxycycline during pregnancy despite its long history of use and initial US approval in 1967 [191]. This class of antimicrobials is rarely used during pregnancy likely due to the tetracycline-class warning of permanent discoloration of the teeth from exposure during the last half of pregnancy and through childhood (to 8 years of age), although studies have shown no doxycycline-associated dental staining in young children [192–194].

Our systematic review has several limitations. The body of evidence reporting safety of antimicrobials in pregnancy, particularly prospective controlled clinical studies, is limited. Although our inclusion criteria were intentionally broad to include antimicrobial exposures for various treatment indications to provide comprehensive assessment of available data, other factors including preexisting conditions, severity of illness (eg, diabetes, HIV, listeria, malaria, TB), and concomitant treatments may have contributed to reported outcomes. These results, however, may be comparable to the estimated background risk of major birth defects (approximately 3%) and spontaneous abortions (approximately 10% among clinically recognized pregnancies in the U.S. general population) [195, 196].

Because we limited our search to specific antimicrobials considered for treatment and PEP of plague and excluded studies of the antimicrobial drug class, we cannot generalize results to other antimicrobials within the same drug class. In addition, many of the studies were observational, did not have an unexposed comparison group, and had serious or very serious risk of bias. Our search was limited to English articles and therefore may have missed additional articles from outside the United States, particularly for chloramphenicol and streptomycin, which are no longer commonly available in the United States. Also, we only included articles published in peer-reviewed literature and publication bias may have contributed to the identification of associations between antimicrobial exposure and adverse maternal/fetal/neonatal outcomes. The small number of prenatal exposures for many of the antimicrobials may have precluded the detection of associations with adverse maternal/fetal/neonatal outcomes, particularly those that are rare. Some of the findings of elevated odds of birth defects were based on single population-based case-control studies with low numbers of antimicrobial exposures among case and control groups. Analytic methods may have adjusted for potential confounding factors (eg, maternal age, race/ethnicity, parity); however, other confounders (eg, disease severity, dosage, duration of antimicrobial exposure, preexisting conditions, other potential teratogenic exposures such as smoking) may not have been assessed and controlled for. Furthermore, missing information on trimester of exposure, small numbers of pregnancy exposures, and limited number of exposures in the first trimester limit the ability to further assess birth defects and fetal/neonatal outcomes. First trimester exposures are especially important in assessing risks of birth defects because this is the critical period of organogenesis.

Despite these limitations, our review is consistent with other reviews and resources of drug safety during pregnancy (Briggs’ Drugs in Pregnancy and Lactation, Lexicomp, REPROTOX, and TERIS) (Table 2) [12, 189, 197]. For most antimicrobials, maternal or fetal/neonatal outcomes observed during pregnancy did not consistently demonstrate adverse outcomes. Several studies suggest a potential risk associated with prenatal streptomycin and TMP-SMX exposure during pregnancy. The risk of adverse maternal, pregnancy, and fetal/neonatal outcomes associated with chloramphenicol and doxycycline is less clear. All information about the safety of antimicrobials during pregnancy should be considered in the context of the risks of the infection and associated morbidity and mortality if patients are not treated. During a public health emergency involving plague or other bacterial biothreats, collection of more data regarding antimicrobial use among pregnant women would be important since antimicrobials are the mainstay for treatment and/or PEP.

Table 2.

Summary of Overall Findings Related to Fetal/Neonatal, Pregnancy, and Maternal Outcomes Compared to Other Antimicrobial Safety References

Antimicrobial [Previous FDA pregnancy categorya] (No. of articles/prenatal exposures)	Systematic Reviewb	TERISc	REPROTOXd	Risk Assessment Summary from Briggs et al,2017e
Aminoglycosides
Amikacin [D] (6 articles/ 9 prenatal exposures)	Data too limited to draw a conclusion [current review]	Teratogenic risk is undetermined based on very limited data	Amikacin and other aminoglycosides often are avoided due to theoretical concerns about ototoxicity and nephrotoxicity. These concerns do not preclude use if indicated for serious infection	Human data suggest low risk
Gentamicin [D] (44 articles/ 345 prenatal exposures)	Limited data on nephrotoxicity and ototoxicity; no clear association with birth defects [current review]	Teratogenic risk is undetermined based on limited data. A small risk cannot be excluded, but there is no indication that the risk of malformations in children of women treated with gentamicin during pregnancy is likely to be great. Because it is an aminoglycoside, maternal treatment during pregnancy may be associated with an increased risk for fetal auditory nerve or renal damage	Ototoxicity and nephrotoxicity in the fetus are theoretical possibilities, but these adverse effects have not been documented clinically	Human data suggest low risk
Plazomicin (0 articles)f	No data	No summary	No summary	No summary
Streptomycin [D] (43 articles/ 285 prenatal exposures)	Maternal and neonatal hearing and/or vestibular deficit possible, particularly for prolonged exposure (≥14 days) [current review]	Risk of deafness minimal based on limited to fair data. Therapeutic doses are unlikely to pose a substantial teratogenic risk; data are insufficient to say there is no risk	Might be associated with ototoxicity in offspring, but appears to be infrequent if it occurs at all	Human data suggest risk
Tobramycin [D] (8 articles/ 43 prenatal exposures)	Very limited data on nephrotoxicity and ototoxicity; no clear association with birth defects [current review]	Because it is an aminoglycoside, maternal treatment during pregnancy may pose a risk for fetal auditory nerve damage	Based on animal studies, not expected to increase risk of congenital anomalies. Nephrotoxicity and ototoxicity with aminoglycosides have been reported in adults. Ototoxicity has rarely been reported in children born after intrauterine exposure to some aminoglycosides	Human data suggest low risk
Chloramphenicol
Chloramphenicol [C] (13 articles/ 246 prenatal exposures)	Limited data to assess risk of birth defects. Inadequate data in the third trimester to calculate risk of gray baby syndrome [current review] [12]	Teratogenic risk is unlikely based on limited to fair data; data insufficient to say there is no risk; maternal treatment in late pregnancy may be associated w/ vascular collapse in the newborn infant	Avoided during pregnancy due to its bone marrow toxicity and the possibility of a neonatal syndrome that includes circulatory collapse and death	Chloramphenicol should be used with caution at term
Doxycycline
Doxycycline [D] (9 articles/2351 prenatal exposures)	No consistent pattern of specific birth defects. Inadequate data to assess risk of cardiac birth defects and spontaneous abortion [current review] [12]	Therapeutic doses are unlikely to pose a substantial risk of fetal malformations but the data are insufficient to state there is no risk. Other tetracyclines cause staining of the primary dentition in fetuses exposed during second or third trimesters	Based on animal studies and human reports, not anticipated to increase risk of congenital anomalies. Avoided because other tetracyclines have been associated with transient suppression of bone growth and with staining of developing teeth	Contraindicated in second and third trimesters
Sulfonamides
Sulfadiazine [C] (18 articles/870 prenatal exposures)	No consistent pattern of adverse fetal/neonatal outcomes based on limited data. Inadequate data in the third trimester to calculate risk of kernicterus [current review]	Therapeutic doses are unlikely to pose a substantial teratogenic risk; data are insufficient to say there is no risk	No summary	No summary
TMP-SMX [C] (40 articles/23 602 prenatal exposures)	Potential risk for birth defects with first trimester exposure (neural tube and cardiac birth defects). Inadequate data in the third trimester to calculate risk of kernicterus [current review]	Small to moderate teratogenic risk based on good data; risk appears to be lower if folic acid supplement is taken early in pregnancy; maternal treatment of women with some infectious diseases during pregnancy appears to improve the outcome	TMP-SMX caused malformations in rats. Human studies have not shown consistent increases in specific types of malformations	• TMP: Human and animal data suggest risk • SMX: No summary

Abbreviations: FDA, Food and Drug Administration; REPROTOX, reproductive toxicology; TERIS, Teratogen Information System; TMP, trimethoprim; TMP-SMX, trimethoprim-sulfamethoxazole.

^aPrior FDA pregnancy letter categories denoting the teratogenic effects of a drug. These categories were retired with implementation of FDA’s 2014 Pregnancy and Lactation Labeling Rule [23].

^bSummary of systematic review based on current review and a previously published systematic review by Meaney-Delman et al, 2013 [12].

^cTeratogen Information System (TERIS) is a computerized database of agent summaries to assist in assessing risk of possible teratogens in pregnant women [14].

^dREPROTOX contains summaries on the effects of medications on pregnancy, reproduction, and development [13].

^eRisk assessment summary from Briggs et al, 2017, a reference guide to fetal and neonatal risks of drugs [17].

^fPlazomicin was FDA-approved in 2018 after implementation of FDA’s Pregnancy and Lactation Labeling Rule; therefore, a pregnancy category is not available for plazomicin. Per Zemdri (plazomicin) Package Insert, “There are no available data on the use of Zemdri in pregnant women to inform a drug associated risk of adverse developmental outcomes.” In animal studies, “no drug-related visceral or skeletal malformations were observed in pregnant rats and rabbits administered subcutaneous plazomicin during organogenesis” [24].

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. The authors thank Joanna Taliano, Shana Godfred-Cato, Katharine Cooley, Shannon Fleck-Derderian, Niki Davis, Megan Brown, and Elizabeth Ailes.

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.

Financial support. No specific funding was provided for this work.

Supplement sponsorship. This article appears as part of the supplement “Plague and Bioterrorism Preparedness,” sponsored by the Centers for Disease Control and Prevention.

Potential conflicts of interest. C. P. was employed by Janssen Research & Development, LLC, from July 2017 to June 2018 and was a contractor providing services to Janssen Pharmaceuticals, Inc., from June 2018 to August 2018, prior to C. P.’s contribution to the work described in this article. None of the drugs described in this article are manufactured by Janssen. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.