Randomized Multicenter Trial for the Validation of an Easy-to-Administer Algorithm to Define Penicillin Allergy Status in Sexually Transmitted Infection Clinic Outpatients

Abstract

Background

Approximately 15% of patients in sexually transmitted infection (STI) clinics report penicillin allergies, complicating treatment for syphilis and gonorrhea. Nonetheless, >90% do not have a penicillin allergy when evaluated. We developed and validated an algorithm to define which patients reporting penicillin allergy can be safely treated at STI clinics with these drugs.

Methods

Randomized controlled trial to assess feasibility and safety of penicillin allergy evaluations in STI clinics. Participants with reported penicillin allergy answered an expert-developed questionnaire to stratify risk. Low-risk participants underwent penicillin skin testing (PST) followed by amoxicillin 250 mg challenge or a graded oral challenge (GOC)—amoxicillin 25 mg followed by 250 mg. Reactions were recorded, and participant/provider surveys were conducted.

Results

Of 284 participants, 72 (25.3%) were deemed high risk and were excluded. Of 206 low-risk participants, 102 (49.5%) underwent PST without reactions and 3 (3%) had mild reactions during the oral challenge. Of 104 (50.5%) participants in the GOC, 95 (91.3%) completed challenges without reaction, 4 (4.2%) had mild symptoms after 25 mg, and 4 (4.2%) after 250-mg doses. Overall, 195 participants (94.7%) successfully completed the study and 11 (5.3%) experienced mild symptoms. Of 14 providers, 12 (85.7%) completed surveys and 11 (93%) agreed on the safety/effectiveness of penicillin allergy assessment in STI clinics.

Conclusions

An easy-to-administer risk-assessment questionnaire can safely identify patients for penicillin allergy evaluation in STI clinics by PST or GOC, with GOC showing operational feasibility. Using this approach, 67% of participants with reported penicillin allergy could safely receive first-line treatments for gonorrhea or syphilis.

Clinical Trials Registration. Clinicaltrials.gov (NCT04620746).

In patients attending sexually transmitted infection (STI) clinics, a risk-assessment questionnaire can safely identify patients for penicillin allergy evaluation by skin testing or graded oral challenge, enabling most patients to received first-line STI treatments.

Up to 20% of patients in medical settings [1, 2] and 15 in sexually transmitted infection (STI) clinics [3] report penicillin or beta-lactam allergy, of which less than 5% are confirmed when evaluated [4–7]. Overreported beta-lactam allergy is linked to broad-spectrum antimicrobial overuse, adverse drug events, higher healthcare costs, longer hospitalizations [4, 8, 9], surgical-site infections [10], antimicrobial resistance [11], and increased mortality [12]. This issue is particularly important for syphilis and gonorrhea since penicillin and ceftriaxone are, respectively, the primary therapies [13].

Sexually transmitted infections are largely treated in ambulatory care settings, such as STI clinics where access to allergy/immunology evaluations is limited. Penicillin allergy evaluation has traditionally involved skin testing followed by an oral challenge with amoxicillin [2] and has been increasingly performed by non-allergists, including hospitalists [14], infectious diseases physicians [15], and pharmacists [16, 17], as well as part of preoperative assessments [5].

While penicillin skin testing (PST) is safe, implementation requires well-trained personnel, supplies, and time. Furthermore, although the penicillin major determinants are commercially available (PRE-PEN; AllerQuest, LLC, Plainville, CT), the minor determinants are not, which then requires compounding on-site using Penicillin-G pharmaceutical powder. After a negative PST, a 1-dose oral amoxicillin challenge is administered for confirmation. These complexities present implementation difficulties in high-volume, resource-limited settings such as STI clinics, where rapid treatment is crucial.

Graded oral challenge (GOC) is an alternative that uses a 2-dose sequential, escalating amoxicillin oral challenge. A small amoxicillin test dose is administered, and if there is no reactivity, the full dose is given [18, 19]. Standard GOC protocols only include individuals with low-risk penicillin allergy histories, such as prior gastrointestinal symptoms, unknown history, or cutaneous-only reactions [20, 21]. In contrast, high-risk categories include prior severe cutaneous adverse reactions or symptoms suggesting an immunoglobulin E (IgE)–mediated response. In 2022, a joint task force of major American allergy professional organizations recommended that allergists use GOC for adults with low-risk penicillin allergy histories [22].

We developed an easy-to-administer assessment tool designed to stratify patients with reported penicillin allergy into high- and low-risk groups. We then randomized low-risk patients into either PST or GOC procedure groups and compared the outcomes. Our primary objectives were to validate an algorithm for penicillin allergy evaluation in STI clinics and determine what proportion of patients reporting allergy can be safely treated with first-line therapies for syphilis and gonorrhea. The secondary objectives were to assess the safety and feasibility of penicillin allergy evaluation in STI clinic settings.

METHODS

This was a randomized clinical trial (Clinicaltrials.gov NCT04620746) conducted at 4 geographically diverse STI clinics and was approved by the Johns Hopkins University Institutional Review Board (IRB). Eligible subjects had a reported history of penicillin or other beta-lactam allergy and were aged 18 years or older. Exclusion criteria included human immunodeficiency virus (HIV)–positive persons with a CD4 count less than 200 cells/mm³, antihistamine use within the prior week, and rashes or extensive tattoos in the area needed for skin testing. Pregnant people were not excluded from skin testing but were excluded from subsequent oral challenge. After informed consent, participants were administered a questionnaire for their demographics and to stratify their beta-lactam allergy risk (see Supplementary Material Appendix 1).

We developed our risk questionnaire after literature review and consultation from expert consultants (see Acknowledgments). Since our settings were low-resource outpatient STI clinics, our objective was to identify patients without high-risk penicillin allergies. High-risk histories included immediate, potentially IgE-mediated reactions, such as angioedema, shortness of breath, throat tightness, hypotension, arrhythmia, syncope, anaphylaxis, and hives. Hives were considered high risk only if occurring within the past 5 years. The questionnaire also queried for delayed (non–IgE-mediated) reactions, such as severe cutaneous allergic reactions (eg, Stevens-Johnson syndrome), drug-induced nephritis or hepatitis, serum sickness, blood dyscrasias, and drug fever. Those classified as high risk were excluded from testing and referred to an allergist for further evaluation. We defined low-risk history as the following: unknown remote reaction (>10 years ago), family history of drug allergy, gastrointestinal symptoms, rigors, fatigue, headache, non-urticarial rash (ie, morbilliform rash), hives more than 5 years ago, pruritus without rash, and flushing. These participants defined as low risk underwent allergy evaluation.

Randomization to Penicillin Allergy Testing Evaluation Method

Low-risk participants were block-randomized at each site to either PST followed by oral amoxicillin challenge or the 2-step GOC.

Penicillin Skin Testing Arm

Study providers were trained on-site by a board-certified allergist (S. A.-A.) who was accessible for consultation throughout the study. Instructional videos were also provided. To address potential severe reactions and anaphylaxis, training and rescue medications were provided. The PST process involved both pinprick and intradermal techniques, following allergy protocols [2]. First, participants underwent pinprick testing on the volar forearm. If nonreactive responses were observed, intradermal skin testing was conducted using a 27-gauge needle and syringe. Wheals of 3 mm or larger compared to the negative control indicated positive reactions and were documented in forms and photos to ensure quality.

For testing, benzylpenicillin polylysine, commercially available PRE-PEN (AllerQuest, LLC) was used and on-site preparation of Penicillin-G (10 000 U/mL) mixture. Histamine (AllerQuest, LLC) served as the positive control, while saline served as the negative control. Comprehensive details outlining the assessment protocol and procedures are provided in the Supplementary Material Appendix 2 and preparation of the Penicillin G 10 000-U/mL preparation is provided in the Supplementary Material Appendix 3.

Participants with negative PST testing received an oral challenge with 250 mg amoxicillin. Pre-challenge physical examination included vital signs, directed examination, and peak expiratory flow evaluation. Peak expiratory flow meter assessments were performed with the oral challenges as a precaution, given the outpatient clinic setting; the single-use devices are affordable, disposable, and easily procured.

Participants were excluded from oral challenge if they had abnormal vital signs, findings that would impede allergy assessment on subsequent exams, or were pregnant. After oral administration of amoxicillin 250 mg, participants were observed for 60 minutes, after which the physical examination parameters were repeated. Participants who demonstrated no reaction with normal vital signs, peak flow, and focused exam after the observation period were counseled that they were not allergic to penicillin. If participants had any subjective and/or objective symptom (eg, rash, flushing, pruritus) during the observation period, treatment was provided as deemed appropriate by the clinician, and was defined as an adverse event (AE). Study staff documented all reaction specifics, including type, severity, and the time elapsed from amoxicillin administration to the AE. Participants with signs or symptoms consistent with allergy were counseled regarding their reaction to penicillin.

Graded Oral Challenge Arm

Prior to GOC, vital signs, peak flow, and a focused physical exam were assessed. Exclusions for GOC were the same as outlined for the PST subgroup. After confirming the participant's eligibility, the initial step involved administering amoxicillin 25 mg oral suspension, followed by a 30-minute observation period during which vital signs and peak flow were monitored. If the participant exhibited normal vital signs and remained asymptomatic, a dose of 250 mg amoxicillin was then administered, followed by 30 minutes’ observation, during which assessments were repeated as per allergy assessment protocols [2].

Participants in both subgroups who exhibited no reactions during the evaluation were counseled that they were not allergic and provided a confirmatory letter. Participants who demonstrated signs or symptoms consistent with allergy were counseled that they have penicillin allergy. Participants in both assessment subgroups were provided contact information for study personnel to notify in case of a delayed reaction.

Participant and Provider Acceptability Survey

Participants who completed testing were administered a second questionnaire regarding their opinions about the testing procedures and how confident they were that they could take penicillin in the future. Similarly, participating providers were queried at the study start and completion on the feasibility of penicillin allergy evaluation in this setting.

RESULTS

From June 2021 through December 2022, we screened 325 patients; 41(12.6%) declined and 284 enrolled (Figure 1). Demographics (Table 1) were similar between the enrolled and treated groups, except that a higher percentage of Black participants were randomized to the GOC group (24.8% vs 12.6%; P = .024). Of those who declined, the most common reasons were insufficient time in 17 (41.5%), no interest in knowing their status in 13 (31.7%), and not wanting an oral challenge in 6 (14.6%).

Figure 1.

Screening and enrollment pre-randomization. ¹Ideally the Treated population only include subjects within the low-risk group who received a test. However, any subjects who received a test will be included as the Treated population, even if the study accidentally gave test(s) to a subject not in the low-risk group. ²Multiple answers can be checked for higher-risk allergy history and late effect symptoms. A subject could have higher-risk allergy history and late effect symptoms at the same time. The number is based on the whole high-risk population regardless of treated of not. ³Questionnaire version 3 did not specify the time that hives occurred (both high-risk and late effect symptoms). During the analysis, answers for hives from version 3 are re-categorized into “Hives (More than 5 years ago)”/”Hives 5 years ago or less” using the subject's answer to question 1, How long ago was the reaction to the antibiotic?

Table 1.

Summary of Participants’ Baseline Demographic Characteristics—Enrolled and Treated Populations

	Enrolled Population			Treated Population
	PST Group (n = 139)	GOC Group (n = 145)	Total (n = 284)	PST Group (n = 103)	GOC Group (n = 109)	Total (n = 212)
Age, y
Mean (SD)	35.9 (11.82)	37.4 (11.62)	36.6 (11.72)	36.9 (12.18)	35.9 (10.94)	36.4 (11.54)
Median	33	33	33	34	33	33
Minimum–maximum	19–68	20–67	19–68	19–68	20–67	19–68
Gender,a n (%)
Male	91 (65.5)	98 (67.6)	189 (66.5)	78 (75.7)	78 (71.6)	156 (73.6)
Female	43 (30.9)	35 (24.1)	78 (27.5)	21 (20.4)	20 (18.3)	41 (19.3)
Gender diverse	4 (2.9)	11 (7.6)	15 (2.9)	3 (2.9)	10 (9.2)	13 (6.1)
Not listed/reported	1 (0.7)	1 (0.7)	2 (0.7)	1 (1)	1 (0.9)	2 (0.9)
Race,b n (%)
Black or African Americanc	29 (20.9)	42 (29)	71 (25)	13 (12.6)	27 (24.8)	40 (18.9)
White or Caucasian	104 (74.8)	92 (63.4)	196 (69)	84 (81.6)	72 (66.1)	156 (73.6)
Hawaiian or Pacific Islander	1 (0.7)	1 (0.7)	2 (0.7)	1 (1)	1 (0.9)	2 (0.9)
Asian	1 (0.7)	6 (4.1)	7 (2.5)	1 (1)	6 (5.5)	7 (3.3)
Native American or American Indian	3 (2.2)	6 (4.1)	9 (3.2)	3 (2.9)	1 (0.9)	4 (1.9)
Other	4 (2.9)	12 (8.3)	16 (5.6)	3 (2.9)	9 (8.3)	12 (5.7)
Ethnicity, n (%)
Not Hispanic/Latino/Spanish	117 (84.2)	130 (89.7)	247 (87)	86 (83.5)	98 (89.9)	184 (86.8)
Hispanic/Latino/Spanish	19 (13.7)	15 (10.3)	34 (12)	14 (13.6)	11 (10.1)	25 (11.8)
Unknown/missing	3 (2.2)	0 (0)	3 (1.1)	3 (2.9)	0 (0)	3 (1.4)
HIV status, n (%)
Negative	97 (69.8)	111 (76.6)	208 (73.2)	87 (84.5)	95 (87.2)	182 (85.8)
Positive	16 (11.5)	31 (21.4)	47 (16.5)	16 (15.5)	11 (10.1)	27 (12.7)

Abbreviations: GOC, graded oral challenge; HIV, human immunodeficiency virus; PST, penicillin skin testing; SD, standard deviation.

^aGender diverse includes trans-women, trans-men, gender-queer, and nonbinary individuals.

^bMultiple races might apply.

^cFor the Enrolled Population, the difference in the percentage of Black race is not statistically significant, with P = .12. However, it is statistically significant for the Treated population, with P = .024. The chi-square test is used for both populations.

Of the 284 enrollees, 72 (25.3%) were excluded from further allergy evaluation because of high-risk immediate reactions in 56 (77.8%), high-risk delayed reactions in 11 (15.3%), declined testing by 3 (4.2%), and other reasons in 9 (12.5%). Questionnaire results are presented in Table 2. Most participants (97%) indicated allergies to penicillin-class antibiotics and 3% to cephalosporin-class antibiotics. The most frequently reported drugs were penicillin or amoxicillin (84%). Notably, 10% of participants were unable to recall the specific name of the penicillin-class antibiotic to which they had an allergic reaction.

Table 2.

Summary of Participants’ Allergy History—Enrolled Population

Screening Questions: Allergy History	High-Risk Group (n = 59)	Low-Risk Group (n = 225)	Total (n = 284)
How long ago was the beta-lactam reaction?
Within the past year	3 (5.1)	5 (2.2)	8 (2.8)
2–5 y	13 (22)	10 (4.4)	23 (8.1)
6–10 y	4 (6.8)	11 (4.9)	15 (5.3)
>10 y	38 (64.4)	187 (83.1)	225 (79.2)
Don't remember/unknown	1 (1.7)	12 (5.3)	13 (4.6)
Drug administration
Pill or liquid by mouth	30 (50.8)	116 (51.6)	146 (51.4)
Injection (a shot) into a muscle or through a vein	19 (32.2)	31 (13.8)	50 (17.6)
Don't remember/unknown	10 (16.9)	78 (34.7)	88 (31)
Time to from administration to development of allergy reaction
<1 h	21 (35.6)	30 (13.3)	51 (18)
1–6 h	12 (20.3)	30 (13.3)	42 (14.8)
6–24 h	5 (8.5)	16 (7.1)	21 (7.4)
>24 h	10 (16.9)	41 (18.2)	51 (18)
Unknown	11 (18.6)	108 (48)	119 (41.9)
Low-risk allergy history
Isolated gastrointestinal upset	11	17	28
Chills	10	6	16
Headache	12	10	22
Fatigue	11	10	21
Itching (pruritus), self-limited	24	68	94
Non-urticarial rash (morbilliform)	22	119	141
Hives (>5 y ago)a	18	63	81
Diffuse flushing/redness	25	36	61
Family history	13 (22)	40 (17.8)	53 (18.7)
No history but present in medical record	1 (1.7)	4 (1.8)	5 (1.8)
High-risk allergy history
Angioedema	40	0	40
Wheezing	37	0	37
Throat tightness	31	0	31
Hypotension	5	0	5
Arrhythmia	6	0	6
Syncope	11	0	11
Anaphylaxis (self-report)	18	0	18
Hives ≤5 y agoa	11	0	11
High-risk conditions (≥24 h after drug administration)
Stevens-Johnson syndrome/toxic epidermal necrolysis	1	0	1
Organ injury	0	0	0
Low platelets	1	0	1
Drug reaction eosinophilia and systemic symptoms	0	0	0
Acute generalized exanthem	4	0	4
Dystonia or muscles became either very stiff or very weak	1	0	1
Serum sickness	2	0	2
Anemia	1	0	1
Documented drug fever	2	0	2
Erythema multiforme	2	0	2
Hives ≤5 y agoa	5	0	5
The antibiotic use in the years following the reaction
Penicillin class antibiotic	16 (27.1)	60 (21.1)	76 (26.8)
Cephalosporin class antibiotic	9 (15.3)	95 (42.2)	104 (36.6)
Other non–beta-lactam antibiotic	7 (11.9)	34 (15.1)	41 (14.4)
Ever treated for gonorrhea, syphilis, or strep throat after the antibiotic reaction	32 (54.2)	174 (77.3)	206 (72.5)
Ever skin-tested for allergy to penicillin?	6 (10.2)	7 (3.1)	13 (4.6)
Positive results on skin test	6 (100)	5 (71.4)	11 (84.6)

Data are presented as n (%). Participants could report responses from multiple groups, including both high-risk and low-risk histories.

^aHives was the only response for which a specific time frame was asked due to hives within the past 5 years indicating a higher-risk group. All other responses could have occurred at any point in the past.

After excluding those defined as high risk using our predefined study criteria, 212 (74.6%) participants were eligible for penicillin allergy evaluation. Six participants with high-risk histories were skin-tested in error and were excluded from the analysis, although all had a negative challenge. Of the 212 low-risk participants, 206 (97.1%) completed the allergy assessment and are our final per-protocol population, including 26 (12.7%) with HIV.

In the per-protocol population of 206 participants, 102 (49.5%) were randomized to PST and 104 (50.5%) to GOC (Figure 2). In the PST group, all had negative pinprick tests. After intradermal testing, 3 were excluded from oral challenge—1 for uninterpretable intradermal results and 2 with abnormal vital signs (elevation of blood pressure). Of the 99 participants who received oral challenge with amoxicillin 250 mg, 96 (97%) had a negative challenge and were identified as not penicillin allergic (Table 3) and 3 (3.0%) experienced possible allergic reactions. One participant experienced nasal drainage, itching throat and eyes, and conjunctival injection 50 minutes after the challenge; 1 participant developed a pruritic erythematous rash later in the evening, which subsided after diphenhydramine, and 1 participant experienced a rash 24 hours later that was treated with prednisolone. During the oral challenge, 2 (2%) participants reported nonspecific symptoms: minor itching and a sensation of something in their throat. The investigators deemed the cases as not allergic. We defined 96 out of 102 (94%) PST participants as not allergic.

Figure 2.

Per-protocol population randomization to study arms and results. ¹Low-risk population includes all the subjects who have none of the high-risk or late effect symptoms listed in the questionnaire that happened during their allergic history. Abbreviations: OC, oral challenge; PE, physical exam.

Table 3.

Penicillin Allergy Testing Results—Per-Protocol Population

	Test Results
	n	% (95% CI)
Any penicillin skin test
Positive	0	0 (0, 3.55)
Negative	101	99.02 (94.66, 99.98)
Uninterpretable	1	0.98 (0.02, 5.34)
Discontinued early	0	0 (0, 3.55)
Total	102	…
Skin testing + oral challengea
Positive	3	3.03 (0.63, 8.6)
Negative	96	96.97 (91.4, 99.37)
Total	99	…
Graded oral challenge
Positive	8	7.69 (3.38, 14.6)
Negative	95	91.35 (84.21, 95.97)
Discontinued early	1	0.96 (0.02, 5.24)
Total	104	…
Overall allergy status
Positive	11	5.34 (2.7, 9.35)
Negative	191	92.72 (88.27, 95.87)
Uninterpretable	1	0.49 (0.01, 2.67)
Discontinued earlya	3	1.46 (0.3, 4.2)
Total	206	…

Abbreviation: CI, confidence interval.

^aParticipants with negative skin testing proceeded to an oral challenge.

All 104 participants in the per-protocol GOC group received an amoxicillin 25-mg challenge. Of those, 4 (3.8%) experienced a possible allergic reaction: 2 had upper body rash, 1 had itching accompanied by an upper body rash, 1 had isolated itching of the torso. One additional participant withdrew due to anxiety. A total of 99 (95.1%) participants received an amoxicillin 250-mg challenge and 4 (4.0%) experienced possible allergic reactions: 1 participant had upper body rash, 1 had arm itching and ear flushing, 1 had pruritic upper body rash, and 1 had abdominal rash accompanied by a warm sensation and pruritus. A total of 95 (91.4%) participants successfully completed the GOC and were thus not penicillin allergic.

Of the total 206 participants evaluated, we defined 191 (92.7%) as not allergic. All immediate reactions were mild and responded to antihistamines. One participant had a delayed reaction approximately 24 hours later that required a short course of corticosteroids. In sum, we determined that 92.7% of tested low-risk participants were not at increased risk for penicillin allergy and had their allergy label removed from the medical record. Of all enrolled participants (including the high-risk group), we were able to de-label 191 of 284 (67%).

In the per-protocol population, 17 AEs were reported in 16 participants—4 in the PST and 13 in the GOC subgroups. These included subjective complaints and abnormal examination findings, all of which manifested after oral challenge. Notably, within the GOC, more AEs emerged following the first dose compared with the second dose (7 vs 5). Utilizing our predefined grading scale, 88.2% of AEs were classified as mild, while 11.8% were categorized as moderate. There were no severe AEs.

After completing the allergy assessment, we queried participants on their acceptability of penicillin allergy testing in the clinic, and 99% found the testing procedures helpful (Table 4). Most participants (95.3%) characterized the procedures as very helpful or extremely helpful, were very comfortable or extremely comfortable (95.6%) with the testing in an STI clinic, and would refer a friend or family member with a penicillin allergy for testing (97%). When participants were asked how confident they were that they could take penicillin or a similar antibiotic after negative allergy testing results, 70.2% were very confident, 20.4% were confident, and 7.9% were moderately confident. Participants in the PST group had a higher level of confidence in their negative penicillin testing than those in the GOC group (very confident: 77.1% for PST vs 63.2% for GOC; P = .02).

Table 4.

Participant Acceptability—Per-Protocol Population

	PST Group (n = 101)	GOC Group (n = 104)	Total (n = 205)	P
Did you find the penicillin testing procedures to be helpful?				<.001
Extremely helpful	84 (83.2)	67 (64.4)	151 (73.7)
Very helpful	15 (14.9)	22 (21.2)	37 (18)
Somewhat helpful	1 (1)	12 (11.5)	13 (6.3)
Slightly helpful	0 (0)	2 (1.9)	2 (1)
Not at all helpful	1 (1)	1 (1)	2 (1)
Took too long	0 (0)	1 (33.3)	1 (25)
Testing was uncomfortable	0 (0)	0 (0)	0 (0)
Other	1 (100)	1 (33.3)	2 (50)
Missing	0 (0)	1 (33.3)	1 (25)
Were you comfortable with allergy testing conducted in settings where STD services are provided rather than by an allergist?				.002
Extremely comfortable	92 (91.1)	77 (74)	169 (82.4)
Very comfortable	5 (5)	22 (21.2)	27 (13.2)
Somewhat comfortable	2 (2)	3 (2.9)	5 (2.4)
Slightly comfortable	2 (2)	1 (1)	3 (1.5)
Not comfortable at all	0 (0)	1 (1)	1 (0.5)
If a friend or family member had a history of penicillin allergy, would you refer them for testing?				.112
Definitely yes	95 (94.1)	91 (87.5)	186 (90.7)
Probably yes	5 (5)	8 (7.7)	13 (6.3)
Possibly	0 (0)	2 (1.9)	2 (1)
Probably no	1 (1)	1 (1)	2 (1)
Definitely no	0 (0)	2 (1.9)	2 (1)
How confident are you that you can now take penicillin or similar antibiotics? (Responses from de-labeled participants only)				.020
Very confident	74 (77.1)	60 (63.2)	134 (70.2)
Confident	18 (18.8)	21 (22.1)	39 (20.4)
Moderately confident	3 (3.1)	12 (12.6)	15 (7.9)
Slightly confident	0 (0)	1 (1.1)	1 (0.5)
Not at all confident	1 (1)	1 (1.1)	1 (0.5)

Data are presented as n (%).

Abbreviations: GOC, graded oral challenge; PST, penicillin skin testing; STD, sexually transmitted disease.

A total of 12 of 14 study providers completed the post-study feasibility assessments, and 11 (93%) agreed with the statement that penicillin allergy assessment is safe and effective in STI clinics (data not shown). Implementation barriers identified by the providers included cost of reagents, reimbursement, and time constraints.

DISCUSSION

A brief, easy-to-administer questionnaire can safely identify persons at low risk for penicillin allergy who can be evaluated and de-labeled by either PST/oral challenge or GOC. Both approaches were safe in appropriately selected patients. Our major operational finding was that allergy assessment for lower-risk patients is safe and practical in STI clinic settings. Since penicillin is the first-line treatment for syphilis and sexual partners of people with syphilis [13], confident use of penicillin among patients with previously reported penicillin allergy history expands clinical and public health efforts toward STI control. Similarly, although ceftriaxone is the first-line therapy for gonorrhea, many clinicians refrain from prescribing it in patients with reported penicillin allergy, even though cross-reactivity is minimal [23–25], resulting in treatments with suboptimal second-line drugs. In fact, most of these patients can be safely evaluated, de-labeled, and treated appropriately.

During our trial, 45 participants were subsequently safely treated with either penicillin or ceftriaxone at the STI clinics after de-labeling. The small number of AEs experienced were minor and there were no serious reactions in either study arm. Our randomized clinical trial demonstrated that, in the STI-risk population, after screening for high-risk penicillin allergy, GOC can be performed without the need for skin testing, and we demonstrated safety by careful vital signs and expiratory peak flow assessment. This has important implications for the implementation of penicillin allergy evaluation, especially in low-resource outpatient settings. Although not studied here, this type of protocol may be amenable to a nursing standing order.

Our results extend the rapidly evolving clinical science of penicillin allergy management. This includes an expanded understanding of the natural history of penicillin allergy [26], developing oral challenge procedures easily adaptable to nonspecialty clinic workflows [27], and implementation of screening procedures that facilitate identifying low-risk patients able to receive penicillin without skin testing, similar to those that have been adopted in major health systems [6, 20, 21, 28, 29]. We also showed that these procedures had high acceptability among both patients and providers in the program. Providers preferred GOC due to the shorter amount of time required to complete the procedure and ease of performance. However, some providers did note participants’ increased confidence in PST results over GOC, most likely because allergy testing is widely assumed to include skin testing. Further research should concentrate on identifying clinical settings where on-site penicillin evaluation could prove valuable and on clinicians’ willingness to engage in such efforts.

We demonstrated that, with appropriate support, STI healthcare providers can effectively acquire proficiency in both the PST and GOC procedures. Compared with GOC, implementing PST in a busy STI clinic required more time, and required supply preparation including the various syringe and pinprick devices, reagents including histamine and saline controls, and the major and minor determinants of penicillin for skin testing. Graded oral challenge stands out as a simpler alternative, devoid of the need for specialized in-person training for skin tests, aligns with the latest allergy society guidance [22], and can be easily integrated into clinic operations.

Previous studies of penicillin allergy assessment focused on diagnostic precision—that is, formal evaluation of how many patients with reported allergy actually have it [30, 31]. Similarly, Copaescu et al [31] recently evaluated the Penicillin Allergy Clinical Decision Rule (PEN-FAST), where 382 subjects were randomized to PST or direct oral challenge (DOC), and had results similar to ours. In contrast to Copaescu et al's study, we chose GOC over DOC for added safety, given our trial's setting in STI clinics, and we found that half of the participants deemed allergic by GOC experienced symptoms with the test dose. Therefore, the results together indicate that these assessments can be performed in a wide variety of outpatient venues by non-allergy providers. Furthermore, these results are consistent with the safety of DOCs reported in previous studies, particularly in low-risk populations. This approach offers a reasonable pathway for patients to receive first-line beta-lactam antibiotics.

Our objective was to identify patients with a low risk of true penicillin allergy, who often can be safely evaluated in low-resource settings. Our questionnaire was intentionally designed to have a high negative predictive value (NPV) and our approach was more conservative than other allergy assessment tools. For instance, while many allergists might disregard a reaction history that took place more than 5 years ago, we restricted this time frame solely to cases involving hives. In contrast, for all other indicators of high-risk history, we categorized individuals as high risk if they had ever reported immediate allergic symptoms. As a result, of the 284 participants enrolled, 56 (19.7%) were excluded from further testing because of high-risk symptoms and 11 (3.9%) because of late-effect reactions. These exclusions are higher than those reported by other studies, but we were intentionally conservative. Nevertheless, over 70% (212/284) of STI clinic participants reporting penicillin allergy were safely evaluated with no serious AEs, and of those who completed testing, only 11 of 206 (5.3%) retained their penicillin allergy.

The major limitation of our study was sample size. Due to coronavirus disease 2019 (COVID-19), the participating STI clinics were profoundly understaffed. Our sites also were impacted by Hurricane Ida (New Orleans) and a ransomware attack (Indiana). In 2022, the mpox epidemic negatively affected enrollment. As a result, our sample size was smaller than we originally anticipated and we were unable to assess positive or negative predictive values of specific questionnaire components. Nevertheless, since our approach prioritized NPV and not a specific diagnosis, we maintain the validity of our findings, especially when we consider the questionnaire instrument as a comprehensive instrument. Operationally, the administration time is 5–7 minutes, and it can be easily administered using a computer tablet. We also noted that the GOC group had a higher proportion of Black subjects, even after randomization, but do not believe that this biases or impacts our results.

In summary, we found that, in STI clinic settings with low-risk patients, using 2-step GOC is a feasible, acceptable, and effective way to de-label patients reporting penicillin allergy and provides a pathway to improving patient care and antibiotic stewardship.

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 their project officers: Dr. Carolyn Deal, PhD, and Krista Cato, MHA. They also thank the consultant allergists who provided critical guidance and suggestions: Eric Macy, MD; Alon Vaisman, MD; Kimberly Blumenthal, MD; and Guha Krishnaswamy, MD. They thank the project management and data teams at Family Health International: Pai Lien Chen, PhD, Senior Principal Statistician; Lalitha Venkatasubramanian, BS, Associate Director, Data Management; Mu-Tien Lee, MS, Biostatistician; and Linda McNeil, MA, Carolina Marchena, BA, Sandra Brindis, BA, and Anne Rinaldi, BA, for operational support. From our study sites, they thank Julia Dombrowski, MD; Angela LeClair, ND; Laura Rishel, APRN; Eduardo Munoz, PA; Lora Fortenberry, MSN, WHNP; Miriam Mancuso, PhD; John Williamson, PharmD; and the Wake Forest ID Clinical Pharmacy Practice Team. They finally thank the study participants at the clinic sites who made this project possible.

Financial support. This work was supported by the National Institutes of Health (NIH) STI Clinical Trials Group (contract no. HHSN272201300012I, Division of Microbiology and Infectious Diseases protocol number 18-0023).