Penicillin allergy testing in pregnancy improves maternal and neonatal clinical outcomes Authors: Jamie L Waldron, Nerlyne Desravines, Carolina Alvarez, Rebecca J Cleveland, Kim A Boggess, Mildred Kwan Published: 2026 ABSTRACT AbstractBackgroundPenicillin allergy is commonly reported in pregnancy, a time of high antibiotic utilization wherein β-lactam antibiotics have multiple indications. Unverified penicillin allergy leads to alternative antibiotic receipt and is associated with increased rates of cesarean delivery, prolonged maternal hospitalization, increased infectious complications, and increased number of antibiotics courses, while neonates born to those with penicillin allergy have higher infection rates, hospitalization time, and readmissions.ObjectiveOur primary aim was to evaluate the impact of penicillin allergy evaluation on antibiotic utilization through 30 days postpartum. Our secondary aim was to compare maternal and neonatal clinical outcomes to those who did not undergo penicillin allergy evaluation (usual care group).MethodsParticipants in this prospective cohort study underwent single outpatient visit for penicillin allergy evaluation with supervised drug challenge to amoxicillin after negative skin prick and intradermal skin testing during second or third trimester. Antibiotic utilization, length of stay, and other outcomes for mother and infant were compared to a usual care group by linear regression models.ResultsForty-six participants completed outpatient penicillin allergy evaluation. Compared to 105 pregnant patients in the usual care group, tested individuals were more likely to be married and to have private insurance, and they had significantly increased perinatal β-lactam receipt (96% vs 38%, P < .001). Tested patients had 69% decreased odds of cesarean delivery (odds ratio = 0.31; 95% confidence interval, 0.13. 0.72), while infants of tested patients had 69% lower mean length of stay (odds ratio = 0.31; 95% confidence interval, 0.18, 0.55) and 59% lower odds of high neonatal sepsis scores (odds ratio = 0.41; 95% confidence interval, 0.19, 0.86).ConclusionsProactive penicillin allergy evaluation during pregnancy benefits both maternal and neonatal clinical outcomes and is safe to perform.Key words: Penicillin, β-lactam, pregnancy, group B Streptococcus, neonatal sepsis score, drug challenge, skin test, cesarean section, antibiotic, hypersensitivity In the United States, 10% of pregnant patients report penicillin allergy, a label associated with poor clinical outcomes during pregnancy.1,2 Pregnancy is a time of high antibiotic utilization,3, 4, 5, 6, 7 and β-lactam antibiotics are the preferred antimicrobial for prevention of vertical transmission of group B Streptococcus (GBS) in patients with colonization.8 With penicillin allergy labels, alternative and less effective antibiotic regimens are often used. Studies show that 90% of pregnant patients with self-reported penicillin allergy are delabeled after guideline-supported allergy evaluation; however, many obstetricians have limited awareness of penicillin allergy evaluation methods.4,9, 10, 11 Current guidelines for the management of pregnant patients recommend universal screening during the third trimester for genital GBS colonization and treatment with penicillin in labor to reduce early-onset neonatal GBS sepsis.12 Patients with penicillin allergy require second-line antibiotics including clindamycin, which has growing GBS resistance, and vancomycin, which has variable concentrations in fetal circulation and higher maternal toxicity than other antibiotics.12,13 Use of non–β-lactam antibiotics in GBS-colonized pregnant patients is associated with increased risk for surgical site infections after cesarean section.14 Additionally, parturients with a penicillin allergy are more likely to experience adverse drug reactions (defined as new drug-class allergies within 30 days’ exposure to peripartum antibiotics) because of receipt of alternative (non–β-lactam) antibiotics.13 The most recent American College of Obstetricians and Gynecologists guidelines highlight the complexity of the self-reported penicillin allergy label and call for consideration of penicillin allergy testing in the antenatal period.8 There is growing literature for safety, feasibility, and maternal clinical outcomes of penicillin allergy evaluation in pregnancy, with most studies focused on testing in third trimester.9,10 However, many obstetricians remain reluctant, and current standards do not incorporate penicillin allergy testing into routine prenatal care for those reporting allergies.4 Additionally, prospective data for evaluation during second trimester and neonatal outcomes are limited. The primary aim of this study was to observe antibiotic receipt in parturients with self-reported penicillin allergy who underwent allergy testing during pregnancy, compared to those who did not. Our secondary aim was to compare the clinical outcomes of participants who proactively underwent penicillin allergy testing in pregnancy and their neonates to a matched group of penicillin-allergic pregnant patients who did not undergo testing and their respective neonates. We hypothesized that tested participants would have greater β-lactam and penicillin receipt and improved clinical outcomes, including length of hospitalization and neonatal sepsis scores, compared to those who were not tested. Methods Study design Penicillin allergy evaluation Patients receiving prenatal care through the University of North Carolina (UNC) health care system were recruited for study participation during routine prenatal care clinic visit. After informed written consent, enrolled participants were scheduled for an outpatient allergy visit for penicillin allergy evaluation. Penicillin allergy evaluation visits were scheduled from 14 weeks 0 days gestation through estimated delivery date. Penicillin allergy evaluation was conducted during a single in-person visit lasting approximately 3 hours at a stand-alone outpatient adult allergy clinic also within the UNC health care system under the supervision of the study allergist. Our testing protocol has previously been published.15 It included skin testing with controls, benzylpenicilloyl polylysine, and penicillin G, followed by a 2-step graded oral amoxicillin drug challenge. Participants maintained their penicillin allergy label if they had positive skin testing or developed symptoms during the amoxicillin challenge that were consistent with an immediate IgE-mediated reaction (anaphylaxis, urticaria, vomiting, wheezing, coughing, chest tightness). Participants were deemed nonallergic after completing the graded amoxicillin challenge without development of symptoms consistent with an IgE-mediated reaction. Results were discussed with the participant, the electronic medical record was updated, and a copy of the allergy clinic visit was sent to primary obstetrician. Outside of the single visit to allergy clinic for penicillin allergy evaluation, prenatal care, including evaluation of clinical need for antimicrobial treatment and specific antibiotic utilized, was continued at the discretion of the primary obstetrics provider. Patients who met inclusion criteria for study participation during screening, but who did not undergo penicillin allergy testing during pregnancy, served as the control (usual care) group. Participant criteria Inclusion and exclusion criteria for this prospective cohort study have been previously published.15 We recruited of 18- to 55-year-olds with singleton or multifetal pregnancies at 14 weeks 0 days’ gestation to 36 weeks 6 days’ gestation who intended to deliver within the UNC health care system and who had a self-reported penicillin allergy. The UNC institutional review board approved this study. Predefined clinical characteristics Maternal characteristics collected included age, race, ethnicity, marital status, insurance status, prenatal care site, gravidity and parity, medical comorbidities, and penicillin allergy history (specific causative drug and reaction details). Clinical outcomes included gestational age at delivery, length of labor (defined as time of hospital admission to time of delivery), mode of delivery, length of hospitalization, intrapartum and postpartum complications up to 30 days after delivery, and antibiotic requirement (indication, type, duration) as recorded in the electronic medical record. Antibiotic receipt was grouped into the following stages: antepartum (defined as date of established care with obstetrics through admission to labor and delivery), peripartum (defined as admission to labor and delivery through delivery), and postpartum (defined as delivery through 30 days after delivery). Antibiotic indication (including active infection vs prophylaxis), antibiotic type, and duration of antibiotics were verified by the associated coding in the health record and pharmacy records. We defined “independent antibiotic indications” as those that required an antibiotic prescription for a unique medical indication; multiple antibiotics courses prescribed for the same indication were counted as a single indication. Risk stratification We utilized current published guidelines1 to stratify reported allergic symptoms into low-, moderate-, and high-risk groups. Low risk was defined as isolated cutaneous symptoms (rash, hives, pruritus), symptoms inconsistent with IgE-mediated reaction (reflux, yeast infection, abdominal discomfort, diarrhea), and remote unknown reaction (≥10 years). Moderate risk included recent unknown reaction (<10 years), isolated respiratory symptoms, and cutaneous symptoms with addition of angioedema, dizziness, or altered consciousness. High risk included self-reported anaphylaxis and symptoms that would otherwise meet anaphylaxis criteria (cutaneous symptoms with additional organ system involvement). A penicillin allergy risk stratification tool, PEN-FAST, was published after our protocol was developed; this tool utilizes reaction characteristics (timing, symptoms, treatment) to calculate a score, with higher scores correlating to increased risk of maintaining penicillin allergy.16 These criteria have been applied retrospectively to further risk-stratify our tested cohort; however, because of limited information regarding primary allergic reaction, we were unable to apply PEN-FAST criteria to our usual care group. GBS risk classification and predefined neonatal clinical outcomes GBS colonization status (presence, absence, unknown) was collected and categorized into low versus high risk of vertical transmission. Participants who tested negative for GBS colonization or who were not tested but underwent planned repeat cesarean section were categorized as low risk on the basis of American College of Obstetricians and Gynecologists guideline recommendation against GBS prophylaxis.12 Participants who had positive GBS colonization, or who were undergoing labor and had unknown GBS colonization, were deemed high risk on the basis of guideline recommendation to administer prophylactic GBS treatment.12 Antibiotic administration was left to the discretion of obstetric and pediatric providers. Data collected on infant outcomes included length of hospitalization, gestational age, Apgar score, neonatal sepsis score (recorded in neonate’s electronic medical record or calculated by study staff with the Kaiser Permanente Neonatal Sepsis Score Calculator17,18), neonatal intensive care unit (NICU) requirement, antibiotic requirement, and associated complications during hospitalization and up to 30 days of life. Neonatal sepsis score was assessed continuously and categorically using a 3-level definition as <0.1, 0.1 to 0.3, and >0.3.19 Statistical analysis We used multiple logistic regression to calculate adjusted odds ratios for antibiotic type utilized according to allergy evaluation; we adjusted for age, race, parity, marital status, and private insurance. Chi-square or Fisher exact tests for categorical data and 2-sample t tests for continuous data were used to compare clinical outcomes between the two groups. Descriptive statistics were used to summarize demographic, delivery, antibiotic receipt, and allergy characteristics of study subjects by tested or usual care group status. Counts and percentages were produced for categorical variables, while mean and standard deviation were computed for continuous variables. The Fisher exact test was used to assess associations of penicillin allergy tester or usual care status with type of antibiotic receipt (such as β-lactam or penicillin) throughout pregnancy strata (overall, antepartum, peripartum, postpartum). P values for the Fisher exact test were found to account for small (<5) and zero cell counts. To assess associations of maternal clinical outcomes (gestational age, length of stay, delivery method, high-risk GBS status, antibiotic requirements) with penicillin allergy tested or usual care status, each maternal outcome was modeled separately as a function of group status and adjusted for age, race, and insurance. Gestational age at delivery was modeled by linear regression to produce adjusted beta values and 95% confidence intervals (CIs). Maternal length of stay, a skewed count outcome, was modeled by negative binomial regression to produce adjusted mean ratio (MR) and 95% CI. MRs can be interpreted as how many times higher (if >1) or lower (if <1) the mean outcome of the tested group is compared to the control. Dichotomous outcomes like cesarean delivery, high-risk delivery, and antibiotic requirement were modeled by logistic regression to produce adjusted odds ratios (OR) and 95% CI. To assess associations of neonatal clinical outcomes (infant length of stay, NICU requirements, clinical complications, sepsis score, antibiotic requirements) with penicillin allergy tested or usual care status, each neonatal outcome was modeled separately as a function of group status and adjusted for maternal age, race, and insurance. Infant length of stay and neonatal sepsis score, skewed count outcomes, were modeled by negative binomial regression to produce adjusted MRs and 95% CIs. Neonatal sepsis score was also modeled with the 3-level definition using a proportional odds model for an ordinal outcome using cumulative logistic regression to produce adjusted ORs and 95% CIs. Dichotomous outcomes such as NICU requirement, neonatal clinical complications, and neonatal antibiotic requirement were modeled by logistic regression to produce adjusted ORs and 95% CIs. All analyses were performed by SAS v9.4 (SAS Institute, Cary, NC). Statistical significance was determined at the .05 alpha level. Results Participant recruitment, demographics, and allergy history During the study period, 176 pregnant patients with self-reported penicillin allergy were screened from March 2019 through August 2020. Among these, 74 (42%) were enrolled; 46 completed penicillin allergy testing in an outpatient allergy clinic, with one participant excluded from analysis because of twin gestation (Fig 1). Of those who enrolled but failed to undergo testing (n = 24), 5 patients withdrew from the study for various reasons (including spousal concern for safety, transportation, time commitment, and no reason specified). Ten scheduled an appointment but did not show up or respond to rescheduling requests. Nine patients delivered their babies before scheduling an appointment. Our usual care group (n = 126) consisted of patients who declined penicillin allergy evaluation (n = 53), failed to undergo testing (n = 24), or were unable to be approached for participation (n = 49). Ultimately, 21 people were excluded from the usual care group for the following reasons: loss to follow-up (n = 14), incorrect or duplicate medical record number (n = 5), fetal loss (n = 1), and loss of penicillin allergy status (n = 1). Our final usual care group totaled 105 patients. Exclusions were due to various reasons, including loss to follow-up, fetal loss, clinically delabeled allergy, incorrect medical record number, and duplicate medical record. Compared to the usual care group, participants who underwent testing were slightly older, and they were more likely to be married and be privately insured (Table I). There were no differences in racial and ethnic backgrounds between groups. Table IDemographics and descriptive characteristics of allergy history of pregnant participants with self-reported penicillin allergy in tested and usual care groupsCharacteristicTested group (n = 45)Usual care group (n = 105)Demographics Age (years), mean ± SD31.1 ± 5.028.9 ± 5.6 Non-White race31% (14)44% (46) Married∗73% (33)51% (52) Private insurance87% (39)51% (54)Reported allergic symptoms† Hives38% (17)33% (35) Unspecified rash42% (19)30% (32) Pruritis2% (1)4% (4) Vomiting4% (2)1% (1) Angioedema4% (2)8% (8) Anaphylaxis7% (3)6% (6) Other4% (2)10% (10) Unknown9% (4)1% (1) Unspecified or missing015% (16) Total reported symptoms50110Risk-stratified allergy symptoms‡ Low risk91% (42)82% (73) Moderate/high risk9% (4)18% (16) PEN-FAST scoreNA Very low risk (score = 0)64% (29) Low risk (score = 1-2)31% (14) Moderate risk (score = 3)4% (2) High risk (score = 4-5)0Data are presented as % (nos.) unless otherwise indicated. NA, Not applicable; SD, standard deviation.∗Marital status missing for 4 participants in usual care group.†Not mutually exclusive.‡P = .237; unspecified or missing symptoms (n = 16) were excluded from data analysis for risk stratification of allergy symptoms. Demographics and descriptive characteristics of allergy history of pregnant participants with self-reported penicillin allergy in tested and usual care groups Data are presented as % (nos.) unless otherwise indicated. NA, Not applicable; SD, standard deviation. Marital status missing for 4 participants in usual care group. Not mutually exclusive. P = .237; unspecified or missing symptoms (n = 16) were excluded from data analysis for risk stratification of allergy symptoms. In both groups, penicillin and amoxicillin accounted for most reported antibiotic allergies. Cutaneous symptoms were most reported for initial reaction in both groups (Table I). Reactions inconsistent with IgE-mediated reaction were rarely reported; 1 participant (2.2%) in the tested group reported reflux, while 5 individuals (4.7%) in the usual care group reported non–immune-mediated reactions (2 reports of yeast infections, 1 diarrhea, 1 stomach pain, 1 flulike symptoms). Similar rates of low (73 [82%] vs 42 [91%]) and moderate-to-high risk (16 [18%] vs 4 [9%]) histories were reported among tested and usual care groups, respectively (P = .237, Table I). When PEN-FAST scoring was applied, all but 2 patients in the tested group had scores of <3. The average gestational age at time of penicillin allergy testing was 26 weeks. The results of penicillin allergy evaluation for this cohort have been previously published.15 Antibiotic receipt across groups Antibiotic receipt was more common among the usual care group compared to the tested group (80% vs 62%, P = .045) (Table II). Of the 43 participants who had penicillin allergy removed, 26 received antibiotics, with a significantly higher percentage receiving penicillin (65% vs 7%, P < .001) or a β-lactam (96% vs 38%, P < .001) compared to the usual care group (Table III). When antibiotic receipt was evaluated by gestational period, there was no difference in antibiotics between the groups during peripartum and postpartum periods, but there was a significantly higher antepartum receipt of antibiotics in the usual care group compared to the tested group (38% vs 20%, P = .040) (Table I). Specific antibiotic indications are reported for each gestational period in Table II. Most antibiotic indications during peripartum were prophylactic, with 24 and 64 independent indications among tested patients who had their penicillin allergy removed (n = 28) and usual care group (n = 84), respectively. There were 9 and 48 independent antibiotic indications among patients who received antepartum antibiotics in the tested (n = 9) and usual care (n = 40) groups, respectively. There was a significant difference in the antepartum antibiotic requirement for indication of genitourinary infections in the usual care group compared to the tested group (85% vs 44%, P = .007). Antibiotic indications during the postpartum period included postpartum infection, postdelivery surgical or procedural prophylaxis, urinary tract infection, and mastitis. Because indications for postpartum antibiotics were limited and varied widely, statistical analysis for differences in indications between the two groups was not performed. Table IIDescriptive characteristics of maternal and neonatal clinical outcomes in patients with self-reported penicillin allergy in tested and usual care groupsCharacteristicTested group (n = 45)Usual care group (n = 105)Delivery characteristics∗Maternal GBS status Positive38% (17)23% (24) Negative60% (27)67% (70) Unknown2% (1)10% (11)High risk38% (17)24% (25)Cesarean section24% (11)45% (47)Maternal length of stay (days), median [IQR] range2 [2-3] 1-143 [2-4] 1-18Infant length of stay (days), median [IQR] range2 [1-2] 1-122 [2-3] 1-90Gestational age at delivery (days), mean ± SD273.2 ± 10.4267.3 ± 15.9Neonatal ICU requirement18% (8)23% (24)Neonatal clinical complications33% (15)38% (40)Premature07% (7)Neonatal sepsis score among not premature, median [IQR] range0.10 [0.07-0.14] 0.02-0.840.12 [0.07-0.19] 0.02-8.55Neonatal sepsis score among not premature <0.147% (21)38% (37) 0.1-0.347% (21)43% (42) >0.37% (3)19% (19)Infant antibiotic requirement4% (2)12% (13)Antibiotic receipt in pregnancyOverall (throughout pregnancy)62% (28)80% (84) Antepartum20% (9)38% (40) Peripartum53% (24)61% (64) Postpartum13% (6)10% (11)Antepartum indication (% out of antepartum antibiotic receipt)† GU infection‡44% (4)85% (34) Non-GU infection§33% (3)10% (6) Prophylaxis¶11% (1)6% (2)Peripartum indication (% out of peripartum antibiotic receipt)† Surgical/procedural prophylaxis50% (12)77% (49) GBS prophylaxis58% (14)36% (23) PPROM02% (1) Chorioamnionitis02% (1)Postpartum indication (% out of postpartum antibiotic receipt)† Postpartum infection33% (2)64% (7) Prophylaxis¶33% (2)18% (2) GU infection‡17% (1)18% (2) Mastitis17% (1)0Data are presented as % (nos.) unless otherwise indicated. GU, Genitourinary; IQR, interquartile range; PPROM, premature rupture of membranes; SD, standard deviation.∗From all delivery characteristics, one woman was excluded because of twin delivery.†Some individuals had >1 indication of diagnosis, so n may sum to more than total reported.‡GU infection was defined as urinary tract infection, bacteriuria, pyelonephritis, bacterial vaginosis, and sexually transmitted infections (trichomonas, chlamydia, gonorrhea, syphilis, pelvic inflammatory disease).§Non-GU infections included sinusitis, dental abscess, appendicitis, bronchitis, and upper respiratory infections.¶GBS colonization, procedural or surgical prophylaxis. Descriptive characteristics of maternal and neonatal clinical outcomes in patients with self-reported penicillin allergy in tested and usual care groups Data are presented as % (nos.) unless otherwise indicated. GU, Genitourinary; IQR, interquartile range; PPROM, premature rupture of membranes; SD, standard deviation. From all delivery characteristics, one woman was excluded because of twin delivery. Some individuals had >1 indication of diagnosis, so n may sum to more than total reported. GU infection was defined as urinary tract infection, bacteriuria, pyelonephritis, bacterial vaginosis, and sexually transmitted infections (trichomonas, chlamydia, gonorrhea, syphilis, pelvic inflammatory disease). Non-GU infections included sinusitis, dental abscess, appendicitis, bronchitis, and upper respiratory infections. GBS colonization, procedural or surgical prophylaxis. Table IIIAntibiotic type utilized in tested and usual care groups during pregnancy among 110 participants who received antibioticsCharacteristicTested groupUsual care groupP valueOverall antibiotics (N = 110)(n = 26)(n = 84) β-Lactam antibiotics96% (25)38% (32)<.001∗ Penicillins65% (17)7% (6)<.001∗Antepartum antibiotics (n = 49)(n = 9)(n = 40) β-Lactam antibiotics100% (9)15% (6)<.001∗ Penicillins78% (7)5% (2)<.001∗Peripartum antibiotics (n = 86)(n = 22)(n = 64) β-Lactam antibiotics100% (22)42% (27)<.001∗ Penicillins59% (13)6% (4)<.001∗ β-Lactam alternatives†063% (40)<.001∗Postpartum antibiotics (n = 17)(n = 6)(n = 11) β-Lactam antibiotics67% (4)9% (1).0276∗ Penicillins17% (1)9% (1)1∗First-line antibiotic for GBS prophylaxis (n = 37)(n = 14)(n = 23) Penicillin G93% (13)17% (4)<.001∗∗Data are shown as % (nos.). Shown are descriptive characteristics and test of differences, with P values via Fisher exact test and exact OR and 95% CIs, according to tested or usual care group for antibiotic type received during pregnancy among 110 participants who received antibiotics and did not have penicillin allergy. Data are excluded from those who did not require antibiotics (n = 16) or had penicillin allergy (n = 3) in tested group; and those who did not require antibiotics (n = 21) in usual care group.∗Two-sided P values and ∗∗P value.†Defined as vancomycin, clindamycin, or gentamicin. Antibiotic type utilized in tested and usual care groups during pregnancy among 110 participants who received antibiotics Data are shown as % (nos.). Shown are descriptive characteristics and test of differences, with P values via Fisher exact test and exact OR and 95% CIs, according to tested or usual care group for antibiotic type received during pregnancy among 110 participants who received antibiotics and did not have penicillin allergy. Data are excluded from those who did not require antibiotics (n = 16) or had penicillin allergy (n = 3) in tested group; and those who did not require antibiotics (n = 21) in usual care group. Two-sided P values and ∗∗P value. Defined as vancomycin, clindamycin, or gentamicin. GBS status and delivery outcomes There were no significant differences in GBS colonization between tested and usual care groups (Table II). When stratified into high versus low risk for GBS transmission, there was no statistical difference in GBS treatment indications between the two groups (P = .224). Tested participants had significantly fewer cesarean sections compared to usual care group (24% vs 45%, P = .047). There was no clinical significance between gestational age at delivery between infants born to those in usual care (267.3 days or 38 weeks 1 day) versus tested group (273.2 days or 39 weeks 0 days). There were no differences in maternal length of stay or in delivery or postpartum complications between the two groups. Multivariable analysis of clinical outcomes adjusting for maternal demographic differences The adjusted multivariable analysis estimating the association of group status and clinical outcomes is shown in Table IV. There was no clinical significance in gestational age at delivery. Patients from the tested group had a significantly lower odds of cesarean section (OR = 0.31; 95% CI, 0.13, 0.72) but a significantly higher risk of requiring GBS prophylaxis (high-risk delivery, OR = 2.75; 95% CI, 1.17, 6.48) compared to those from the usual care group. There was no association between maternal length of stay or antibiotic requirement with group status. Infants born to women in the tested group had significantly lower mean length of stay (OR = 0.31; 95% CI, 0.18, 0.55) compared to infants born in the usual care group. Additionally, infants born to tested mothers had a significant (59%, OR = 0.41; 95% CI, 0.19, 0.86) decreased odds of having a high neonatal sepsis score compared to those born to mothers in the usual care group. We found no association between NICU requirements, associated complications, or antibiotic requirements with group status. Table IVIndependent association of being patient or neonate born to patient in tested (vs usual care) group with effect on maternal and neonatal clinical outcomes in 150 subjectsModel no.OutcomeEffectAdjusted effect (95% CI)∗Maternal clinical outcomes1 Gestational age at delivery (days)β5.7 (0.4, 10.9)2 Maternal length of stay (days)MR0.90 (0.65, 1.26)3 Cesarean section (ref = SVD)OR0.31 (0.13, 0.72)4 High-risk delivery (ref = not high risk)OR2.75 (1.17, 6.48)5 Any antibiotic requirement (ref = none)OR0.45 (0.20, 1.03)Neonatal clinical outcomes6† Infant length of stay (days)MR0.31 (0.18, 0.55)7 NICU requirement (ref = no NICU)OR0.59 (0.22, 1.52)8 Neonatal clinical complications (ref = none)OR0.75 (0.34, 1.65)9A‡ Neonatal sepsis scoreMR0.39 (0.15, 1.01)9B‡ Neonatal sepsis score categoriesOR0.41 (0.19, 0.86)10 Infant antibiotic requirement (ref = none)OR0.31 (0.06, 1.53)∗Adjusted for age, race, and insurance status.†This model excludes 1 infant missing length of stay.‡These models exclude 7 premature infants for whom sepsis score is not computed. Adjusted beta values are results from normal regression models for continuous outcomes (gestational age at delivery, model 1); adjusted MRs are results from negative binomial regression models for skewed count outcomes (maternal length of stay, infant length of stay, and neonatal sepsis score; models 2, 6, and 9A); adjusted ORs are shown for categorical outcomes (all others; model 9B is proportional odds model for 3-level ordinal outcome using cumulative logistic regression). Independent association of being patient or neonate born to patient in tested (vs usual care) group with effect on maternal and neonatal clinical outcomes in 150 subjects Adjusted for age, race, and insurance status. This model excludes 1 infant missing length of stay. These models exclude 7 premature infants for whom sepsis score is not computed. Adjusted beta values are results from normal regression models for continuous outcomes (gestational age at delivery, model 1); adjusted MRs are results from negative binomial regression models for skewed count outcomes (maternal length of stay, infant length of stay, and neonatal sepsis score; models 2, 6, and 9A); adjusted ORs are shown for categorical outcomes (all others; model 9B is proportional odds model for 3-level ordinal outcome using cumulative logistic regression). Discussion Our findings demonstrate that proactive penicillin allergy evaluation enhances antibiotic selection during pregnancy with increased guideline-driven β-lactam utilization and is associated with lower maternal rates of cesarean section, shorter neonatal lengths of stay, and lower neonatal sepsis scores. We previously reported a rate of 94% (43 of 46) negative penicillin allergy test results, while 3 patients retained their penicillin allergy (1 [5%] with positive intradermal testing result, 2 [10%] with failed oral challenge).15 Like recent studies, we found penicillin allergy evaluation in pregnancy increased appropriate antibiotic receipt with increased utilization of β-lactam antibiotics for peripartum, but also in antepartum and postpartum time periods.9, 10, 11 In our study, penicillin allergy evaluation enhanced antibiotic selection throughout pregnancy, with the most notable increase in guideline-recommended penicillin utilization during peripartum: All pregnant patients who had the penicillin allergy label removed were able to receive a β-lactam antibiotic for peripartum prophylaxis, with 100% of GBS-colonized patients receiving first-line antibiotics with penicillin compared to 6% in the usual care group. Despite the majority of women in the usual care group reporting a low-risk penicillin allergy (82%), only 42% requiring peripartum antibiotics received a β-lactam, suggesting continued reluctance among obstetricians to administer a guideline-supported alternative cephalosporin in patients with penicillin allergy.20 As cefazolin remains first-line treatment for GBS for those with low-risk penicillin allergy, this further supports the utility of penicillin allergy testing in pregnancy. We previously reported safety of penicillin allergy evaluation as early as the second trimester,15 while our current study supports the utility of a second-trimester evaluation. Negative allergy evaluation immediately allowed the option of administering β-lactams. We found that 100% of tested participants received a β-lactam (78% penicillin) antepartum compared to 15% (5% penicillin) in the usual care group, suggesting that penicillin allergy evaluation enhances β-lactam receipt for antepartum indications, such as asymptomatic bacteriuria and urinary tract infections.21,22 This underlines that penicillin allergy testing may have benefits beyond patients colonized with GBS and should not be deferred until late in the third trimester. Penicillin allergy evaluation was associated with improved maternal and neonatal clinical outcomes. Tested participants had a decreased odds of cesarean section compared to our usual care group, consistent with other publications.13 Interestingly, our study found that those with an increased odds of high-risk delivery (those with indication for GBS prophylaxis) were more likely to be in the tested group; all tested patients with high-risk GBS status received a β-lactam antibiotic as prophylaxis. Despite those with high-risk delivery having an increased likelihood of being in the tested group, our study found no difference in antibiotic requirements between the tested and usual care groups for peripartum prophylaxis. Retrospective studies have shown that infants born to mothers with penicillin allergy have increased hospital stay, readmission rates, and increased blood draws.20,23 To our knowledge, this is the first prospective study reporting improved clinical outcomes for infants of those who underwent penicillin allergy evaluation. Though maternal lengths of stay were not statistically different, infants born to our tested cohort had 70% shorter duration of stay compared to those born to our usual care group. Additionally, our study found that neonates born to tested participants had significantly lower neonatal sepsis scores. Neonatal sepsis scores influence several aspects of neonatal clinical outcomes, including laboratory evaluation, antibiotic administration, level of care, and duration of hospitalization. Although we found no difference in neonatal antibiotic receipt, increased neonatal sepsis scores may warrant more invasive monitoring, lab work, and additional studies that may otherwise not be required with preferred maternal prophylaxis. The specific impact of lower sepsis scores on neonatal clinical outcomes was beyond the scope of our study; larger studies are needed to further investigate the relationship between sepsis scores and neonatal outcomes. Our study was strengthened by the multidisciplinary collaboration between the allergy and obstetrics departments. Prenatal care was allowed to proceed uninterrupted, and decisions regarding antibiotic choice were at the discretion of the primary obstetrician. Because most studies on penicillin allergy evaluation in pregnancy are retrospective, the prospective nature of this study is a strength. Limitations include a small cohort at a single health care center, with halted ability to perform allergy evaluation because of allergy clinic closures during the early phase of the coronavirus disease 2019 pandemic. This small sample size may prevent generalizable results, and further studies with larger cohorts are needed to support these improved clinical outcomes. Additional limitations include a tested cohort that had higher association with private insurance compared to the usual care group. The location of our allergy clinic, which is distant from the main hospital campus and obstetrics practices, may have restricted study participation for those with limited resources. However, this demographic difference in insurance, a proxy for socioeconomic status, has also been reported in other studies,23,24 and further studies are needed to investigate this association and improve access to allergy evaluation. Similar to other studies,4 the majority of participants reported low-risk symptoms, limiting application to higher-risk cohorts. Our study supports the enhancement of antibiotic options, not only peripartum but also antepartum, with proactive penicillin allergy evaluation. The increased utilization of first-line β-lactams for prophylaxis resulted in improved maternal and neonatal outcomes. However, our findings suggest that socioeconomic status may influence penicillin allergy evaluation. Future studies should investigate ways to improve participation regardless of socioeconomic status, as well as the impact on neonatal sepsis scores, and long-term maternal and neonatal outcomes. Disclosure statement Funded by a Cefalo-Bowes grant and internal support from the Department of Obstetrics and Gynecology, Division of General Obstetrics and Gynecology at the University of North Carolina at Chapel Hill, and the Department of Internal Medicine, Division of Allergy and Immunology at the University of North Carolina at Chapel Hill. Disclosure of potential conflict of interest: The authors declare that they have no relevant conflicts of interest. Contributor Information Jamie L. Waldron, Email: JLWaldron@mgh.harvard.edu. Mildred Kwan, Email: mildred_kwan@med.unc.edu. Supplementary data Supplementary Table 1 References Associated Data Supplementary Materials Supplementary Table 1