Penicillin allergies and selection of intrapartum antibiotic prophylaxis against group B Streptococcus at a safety-net institution

Introduction

Intrapartum antibiotic prophylaxis (IAP) is an important factor in the prevention of neonatal group B streptococcal (GBS) disease. Early-onset disease can be prevented by screening for GBS colonization and administering appropriate IAP [1], 2]. Guidelines for antibiotic administration have contributed to a decreasing incidence of early-onset GBS disease, which still carries a mortality rate of 2.1–20 % depending on an infant’s gestational age [3].

Guidelines issued by Centers for Disease Control and Prevention (CDC), American College of Obstetricians and Gynecologists (ACOG), and American Academy of Pediatrics (AAP) outline recommendations for IAP [3], [4], [5]. Penicillin, and alternatively ampicillin, remain the prophylaxis antibiotics of choice, because of their narrow spectrum of activity and excellent penetration into fetal tissues. Cefazolin achieves high intra-amniotic concentrations and is therefore an acceptable prophylaxis for patients with low-risk penicillin allergies. For patients with high-risk penicillin allergies, clindamycin is the preferred agent, as long as GBS isolates demonstrate susceptibility to clindamycin and erythromycin. If clindamycin cannot be used, vancomycin is the last-line agent [4]. As there are limited efficacy data for clindamycin and vancomycin in neonates, both agents are considered second-line (or “alternative”) options for GBS prophylaxis [3], [4], [5], [6].

Reported penicillin allergies are estimated in 10 % of both the general and obstetric populations. However, fewer than 1 % have a true IgE-mediated, high-risk penicillin allergy, as manifested by anaphylaxis, angioedema, respiratory distress, or urticaria [7], [8], [9], [10]. Recognizing the importance of first-line GBS prophylaxis, the objectives of this study were to determine the types and severities of allergic reactions in penicillin-allergic, GBS-colonized patients and to assess the appropriateness of IAP selection.

Subjects and methods

A retrospective cohort study was conducted at a single, safety-net, large-volume, academic medical center comprised of a level III neonatal intensive care unit (NICU), between April 27, 2017 and October 31, 2022. The start date of the study was based on the institutional adoption of an electronic health record (EHR). Data were extracted electronically from the EHR and supplemented by manual chart review for missing data.

Patients with a recorded penicillin allergy and GBS colonization based on routine prenatal care testing were identified. Testing is performed in a centralized laboratory in our institution, and all rectovaginal samples for GBS testing undergo susceptibility testing for clindamycin and erythromycin. Additional exclusions included precipitous vaginal births (resulting in insufficient time to receive IAP), scheduled cesarean sections, cesarean sections unrelated to labor, fetal demise, or patient declining IAP. A penicillin allergy was defined as any allergy listed in the EHR to penicillin, amoxicillin, ampicillin, or piperacillin. Allergies were classified as high- or low-risk, where a high-risk penicillin allergy was defined as documentation of anaphylaxis, angioedema, respiratory distress, or urticaria, and low-risk allergy was defined as the absence of those symptoms (including the presence of unspecified non-urticarial rash). If multiple reactions were selected that fell into both the high- and low-risk allergy definition, the patient was classified as having a high-risk allergy. If no allergy reaction was documented, it was assumed the patient had low-risk allergy and was therefore categorized as such. These approaches are consistent with standard practice in allergy literature [7], [8], [9], [10].

The following variables were abstracted from the EHR: maternal demographic and obstetric characteristics, reported antibiotic allergies, GBS test results and isolate susceptibilities, antibiotic prophylaxis administered during labor, maternal outcomes, and neonatal outcomes. As per standard of care, patients were screened for GBS in their third trimester and received antibiotics at delivery according to the institutional policy, which is in line with the current ACOG recommendations for intrapartum antibiotic prophylaxis selection [5].

The primary study outcome was the proportion of patients receiving inappropriate IAP for GBS. Inappropriate IAP was defined as administration of alternative antibiotics (clindamycin, vancomycin, or cefotetan) in place of standard penicillin or cefazolin in the setting of a low-risk penicillin allergy. Secondary maternal outcomes included chorioamnionitis, endometritis, and/or surgical site infection within 6 weeks after delivery. Neonatal outcomes included sepsis, bacteremia, meningitis, antibiotic treatment, and/or readmission within 30 days of life.

Descriptive statistics were used to characterize the study cohort and to evaluate outcomes. Continuous variables are presented as means and standard deviations (SD) or median and interquartile range (IQR) while categorical variables are presented as percentages. Statistical analysis was conducted using SAS Version 9.4 (Cary, NC). The study was approved and determined to be exempt from review by our Institutional Review Board (COMIRB 22-1593). The authors have complied with the World Medical Association Declaration of Helsinki regarding ethical conduct of research involving human subjects and/or animals.

Results

During the study period there were 18,931 births. Demographics and obstetric characteristics of all birthing people, those with penicillin allergies, and those with penicillin allergies and GBS colonization are summarized in Table 1.

Among the total patient cohort, 1,262 (6.7 %) patients had a documented penicillin allergy. Of those, 244 (19.3 %) were GBS-positive, 786 (62.3 %) were GBS-negative, and 232 (18.4 %) were not tested for GBS (Figure 1). Among those with a reported allergy, 1,009 (80.0 %) had a documented allergy to penicillin, 361 (28.6 %) to amoxicillin; 32 (2.5 %) to ampicillin; 1 (<1 %) to piperacillin; and 138 (10.9 %) of patients had documented allergies to multiple penicillin category antibiotics.

Figure 1:

Flowchart of analysis of patient cohort.

After additional a priori specified exclusions, among the 196 penicillin-allergic, GBS-colonized patients, 110 (56.1 %) had a high-risk reaction and 86 (43.9 %) had a low-risk reaction. Of those categorized as having a low-risk reaction, 26 did not have a reaction listed. The most common allergic reactions included urticaria (high-risk reaction; n=75, 38.3 %) and unspecified rash (low-risk reaction; n=69, 35.2 %). The two were listed simultaneously in 10.6 % of patients, therefore classifying those patients as having high-risk allergies. The prevalence and type of allergic reactions are further summarized in Table 2.

Out of the total cohort of 196 penicillin-allergic, GBS-colonized patients, 55 (28.1 %) received inappropriate antibiotics. Of the 86 GBS-colonized patients with low-risk penicillin reactions, 54 (62.8 %) received inappropriate prophylaxis, 32 (37.2 %) received appropriate prophylaxis [penicillin (n=7), ampicillin, (n=2), or cefazolin (n=24)]. Of the 110 patients with high-risk reactions, 1 (0.9 %) received inappropriate antibiotic prophylaxis [cefotetan], and 109 (99.1 %) received appropriate prophylaxis.

Fewer than 7 % of pregnant patients developed a complication, including chorioamnionitis, endometritis, or surgical site infection within 30 days. There were 198 neonates born to the 196 penicillin-allergic, GBS-colonized patients. Only one neonate was diagnosed with GBS bacteremia. Maternal and neonatal complications were rare; outcomes are summarized in Table 3 and Table 4, respectively.

Discussion

The overall prevalence of penicillin allergy among pregnant patients at our institution during the study period was 6.7 %. Among penicillin-allergic patients, the rate of GBS colonization was 19.3 %. Among the penicillin-allergic, GBS-colonized patients, about half had high-risk penicillin allergies (56.1 %). Of those with a low-risk penicillin allergy, the rate of inappropriate antibiotic selection was 62.8 %, a rate similar to those of other studies examining antibiotic selection for IAP based on allergy stratification and antimicrobial sensitivity testing [11], [12], [13], [14], [15], [16]. Significant maternal and neonatal complications were low.

The prevalence of reported penicillin allergy at our institution in this population is slightly lower than the rates reported in similar studies, while the rate of high-risk allergies was higher [9], 10]. This likely represents an overestimate skewed by the ability in our EHR during the study period for providers to document both a high-risk allergy and low-risk allergy for a given patient, which occurred in 19 % of patients with any documentation. For example, 10.6 % of patients had documented reactions that included both urticaria (considered high-risk in our institution) and unspecified rash (considered low-risk), which likely reflects inaccuracy in allergy documentation rather than the true occurrence of both rash types.

Moreover, 13.3 % of patients had no reaction documentation associated with their penicillin allergy. These shortcomings in documentation can make it difficult for providers to determine the appropriate antibiotic selection at the time of delivery. Even though these patients should be considered low-risk by accepted standards – and were considered low-risk in this study – in actual practice, providers may nonetheless treat such patients with alternative antibiotics out of an abundance of caution [17]. This lack of clarity indicates areas for improvement in allergy documentation and antibiotic selection. To address these two points of confusion, our institution recently instituted a hard-stop in allergy documentation that requires documenting the reaction when entering a new drug allergy, a quality improvement change that was undertaken independent of the focus of this study.

It is important to consider the potential negative impacts of inappropriate antibiotic usage in this context. While some studies indicate minimal adverse effects with inappropriate antibiotic usage, other studies have found associations with longer neonatal length of stay, higher likelihood of a postnatal lab draw, and higher endometritis incidence in obstetric patients [9], 10], [13], [14], [15], [16, 18]. Additionally, concerns about the effects of broad-spectrum antibiotic usage on both the fetal microbiome and development of multi-drug-resistant organisms argue for striving for best practices in antibiotic selection.

A variety of tools have been explored for improving allergy documentation, including provider education, in-person allergy evaluations, allergy clarification tools, order sets, identifying physician champions, and risk stratification for skin testing or direct oral challenges [19], [20], [21], [22]. As a quality improvement intervention based on these findings, we plan to implement provider education regarding the difference between urticarial and non-urticarial rash, as well as a detailed allergy screen during prenatal care to better identify low-vs. high-risk allergies. The goal of these will be to more accurately document reactions and thereby be able to de-escalate or de-label a documented allergy based on history alone. This presumption is based on prior literature showing low rates of positive skin testing or drug challenge results in obstetric patients (0–7% positivity), as well as low rates of adverse events after allergy de-labeling (no/extremely limited (<1 %) reported events in patients receiving intrapartum penicillin after being de-labeled) [20], [21], [22], [23], [24], [25], [26], [27], [28], [29]. Moreover, such interventions have been shown effective in reducing the usage of inappropriate alternative antibiotics for intrapartum antibiotic prophylaxis [30], 31].

Our study has several notable strengths. It addresses the common and important issue of the interaction of penicillin allergies with intrapartum antibiotic prophylaxis for GBS. It is one of few studies to focus on this question in a safety-net population. There is a wide range of obstetric providers within our system, including attending and resident obstetrician-gynecologists, certified nurse-midwives, family medicine physicians, and advanced practice providers, broadening the internal and external validity of our findings to many practice settings and provider types. We also had minimal exclusion criteria, which helps to strengthen the generalizability of our study. Finally, we combined electronic chart review with manual chart review to ensure minimal missing data.

The study also has several limitations. Retrospective data collection means that several data components – non-documented reactions for drug allergies, potential misclassification of reactions as high-risk or low-risk (e.g. urticaria vs. unspecified rash), patients with unclear antibiotics administered – cannot be reconciled or obtained after the fact. Patients with penicillin allergies but without GBS testing were excluded from further analyses, given that they would not have received antibiotic prophylaxis. Lastly, we solely reviewed antibiotic selection, not other factors such as timing or dosing. While these may have confounded our secondary outcomes – allergy reaction type and symptom, maternal and neonatal outcomes – it is unlikely that any of the above would have altered our primary outcome.

In conclusion, more than half of penicillin-allergic, GBS-colonized patients received inappropriate intrapartum antibiotic prophylaxis at our institution. These findings highlight opportunities to implement intervention(s) towards the goal of clarifying allergies and reactions in order to better resolve perceived contraindications to standard IAP antibiotics.