Evaluation of Antibiotic Allergy in the Ambulatory Setting Using a Standardized Questionnaire

Abstract

Patients are sometimes mislabeled as having an immune-mediated antibiotic allergy in their medical records. Therefore, the aim of this study was to investigate the prevalence of subjects with non-immune mediated reactions to antibiotics using a standardized questionnaire. Subjects aged 18 years and older with a documented antibiotic allergy were identified and recruited from 2 outpatient clinics in the greater Chicago area. Subjects completed a standardized questionnaire during a single visit regarding their previous adverse reaction to an antibiotic. For subjects with multiple documented antibiotic allergies, 1 questionnaire was filled out for each antibiotic allergy. Investigators subsequently evaluated the questionnaire responses to determine whether the adverse reaction was a true immune-mediated allergic reaction or an adverse drug reaction. A total of 98 subjects were recruited with completion of 159 questionnaires. Eighteen subjects (18.37%, 95% CI: 10.7%, 26.3%) had antibiotic allergy labels with no corresponding immune-mediated reaction history. There were 35 allergy labels (22.0%, 95% CI: 14.7%, 29.4%) that were unlikely to be immune-mediated. Antibiotics with the highest percentage of clinical histories that were unlikely to be immune-mediated were macrolides (8 of 11 subjects), nitrofurantoin (1 of 2 subjects), and amoxicillin/clavulanate (2 of 8 subjects). The most common antibiotic allergy labels were penicillin (43 of 159 subjects), sulfonamides (25 of 159 subjects), and fluoroquinolones (21 of 159 subjects). Identification of adverse reactions to antibiotics that are unlikely to be immune-mediated can be accomplished using a standardized questionnaire in the outpatient setting. Improved identification of low-risk antibiotic allergy labels can guide de-labeling initiatives to improve antibiotic prescribing.

Introduction

Adverse reactions to antibiotics are commonly reported. Previous studies reported that 13%−25% of their study populations were allergic to at least 1 antibiotic.^1–3 However, the prevalence of antibiotic allergy in these study populations may be inflated by inaccurate antibiotic allergy labeling in the medical record. A recent study demonstrated that up to 80% of penicillin allergies in the medical record are mislabeled.⁴

This incongruence can be attributed to many factors such as the prevalence of non-immune mediated reactions including expected drug side effects and the unreliability of the patient’s memory of the reaction.^4,5 Other inaccuracies can be explained by a lack of follow-up with the patient after initial documentation of antibiotic allergy to confirm an immune-mediated adverse reaction and incomplete documentation of the reaction.⁶ It has been demonstrated that antibiotic allergy labels accumulate over time, further accentuating a need to validate antibiotic allergy labels.⁷

The listing of non-immune mediated adverse medication reactions, such as nausea, in the allergy section of the electronic medical record can make it difficult to distinguish between immune-mediated adverse reactions from non-immune mediated adverse reactions. To reduce erroneous documentation of drug allergies, Macy et al.⁸ recommends that electronic health record systems have a designated place to document drug intolerances instead of drug allergies. This better distinguishes a true immune-mediated allergic reaction from other causes of adverse reactions.

Investigating the hypersensitivity reactions to antibiotics is further complicated by a lack of standardized clinical procedure.⁹ For subjects displaying allergic reactions to a class of antibiotics (eg, beta-lactam antibiotics), they are typically prescribed other classes of antibiotics (eg, vancomycin) to avoid potential risk of cross-reactivity.² The use of alternate antibiotics often leads to a deviance from the standard of care in order to account for the allergy.¹⁰ There are a number of associated burdens that result from restricting the clinician’s arsenal of antibiotics and the deference of first-line antibiotics. Poorer health outcomes for hospitalized patients is likely the most significant consequence.¹¹ Studies looking at beta-lactam allergies in particular have also discovered that subjects with a penicillin allergy are at an increased risk for Methicillin Resistant S. aureus (MRSA) and Clostridioides difficile infections from the use of non-beta-lactam antibiotics and longer average hospital stay.^12,13 Furthermore, use of suboptimal and alternate antibiotic therapies increases the overall risk for antibiotic resistance in the community.^14,15

Antibiotics are typically de-labeled after negative skin prick test, intradermal test, and a monitored oral challenge. De-labeling consists of removing the documented allergy label from the patient chart. However, proper diagnostic testing is costly and requires a significant time investment, which ultimately limits their widespread use.¹⁶ To overcome these challenges, recent initiatives aim to de-label antibiotic allergies through patient interviews.^4,5,17,18

This study investigated whether a symptom-based questionnaire will be able to discern immune-mediated adverse reactions to antibiotics from reactions that are likely non-immune mediated.

Methods

This was a prospective study aimed at estimating the prevalence of individuals who are labeled incorrectly as having an antibiotic allergy despite not having a history of an immune-mediated adverse reaction. The study was approved by the Loyola University IRB Board, LU number 214762. Adult subjects with a documented antibiotic allergy were identified and recruited from 2 outpatient clinics within the Loyola University Health System, which are located in Burr Ridge, IL and Park Ridge, IL. Potential participants were identified through the electronic medical record and approached if they met the following inclusion criteria: 18 years or older and had at least 1 documented antibiotic allergy in their chart. Participants were consented and completed a standardized questionnaire regarding their previous adverse reaction to the antibiotic. Patients with multiple documented antibiotic allergies completed 1 questionnaire for each of their antibiotic allergies.

The questionnaire requested information on demographics, name of the antibiotic allergy, date of adverse reaction, indication for antibiotic use, route of administration, adverse reaction characteristics, timing of the adverse reaction, whether they were hospitalized from the adverse reaction, what occurred after the cessation of the antibiotic, previous allergy testing to the antibiotic, and subsequent antibiotic use following the adverse reaction. A copy of the questionnaire can be found in Appendix A. A team of infectious disease physicians, infectious disease pharmacists, primary care physicians, and study personnel developed the questionnaire. The questionnaire was developed based on questions likely to be asked in a clinical investigation of a prior allergic reaction. We referenced the questionnaire developed by Devchand et al.⁵ in their study of a pharmacist-led penicillin de-labeling initiative.

Investigators reviewed the responses to the questionnaires to determine if the reaction was classified as immune-mediated by categorizing the adverse reaction. Symptoms that could be explained by Type I, II, III, IV hypersensitivity reactions were designated as immune-mediated.¹⁹

Statistical Analysis

Generalized estimating equations are used to estimate the percentage of antibiotic allergies that were incorrectly labeled. Frequency and percentage of incorrectly labeled antibiotic allergies were assessed with qualitative measures. Symptoms of non-immune mediated antibiotic use are summarized with frequency and percentage per antibiotic. A sample size of 100 subjects were required to obtain a confidence interval half-width of 10% or less, using a 95% Wald confidence interval, for estimating the percentage of individuals who had no immune-mediated adverse reactions. This was made assuming a conservative hypothesis in which 50% of individuals had no immune-mediated adverse reaction. All analyses were conducted using SAS 9.4 (Cary, NC) and P-values <.05 were deemed statistically significant. The study was powered at the patient level. 100 patients were estimated as a sufficient sample size to estimate the CI with a half-width of 10%.

Results

The study recruited and consented 104 individuals. Of these, 3 later withdrew from the study, and 3 were excluded from analysis due to incorrectly completed patient questionnaires. The average age (range) was 57.2 (18.1 – 87.8) years, with 71.1% of subjects identifying as female, and 64.3% of subjects identifying as Caucasian. One hundred fifty-nine questionnaires were completed with most subjects having only 1 antibiotic allergy listed in their medical record (69.4%) (Table 1).

Table 1.

Patient Characteristics.

Patient Characteristic	n = 98
Mean age (range)	57.2 (18.1 – 87.8)
Female, n (%)	69 (70.4)
Race, n (%)
White	63 (64.3)
Other	8 (8.2)
Not reported	27 (27.6)
No. of antibiotic allergies, n (%)
1	68 (69.4)
2	17 (17.4)
3+	13 (13.3)

Eighteen of 98 subjects (95% CI: 10.7%, 26.3%) had adverse reactions to antibiotics with symptoms that could not be explained by Type I-IV hypersensitivity reactions, resulting in a half-width of 7.7%. Thirty-five of 159 antibiotic labels were determined to be non-immune mediated (95% CI: 14.7%, 29.4%). Macrolides had the highest prevalence of non-immune mediated adverse reactions (n = 8 of 11), followed by nitrofurantoin (n = 1 of 2), and amoxicillin/clavulanate (n = 2 of 8). In contrast, adverse reactions to clindamycin (n = 10 of 10), metronidazole (n = 5 of 5), and topical antimicrobials (n = 3 of 3) were consistent with immune-mediated reactions.

Penicillin was the most prevalent antibiotic allergy listed in the medical record (n = 43 of 159), followed by sulfonamides (n = 25 of 159), and fluoroquinolones (n = 21 of 159) (Table 2).

Table 2.

List of Antibiotic Allergy Labels Stratified by Immune-Mediated Classification

	Immune-Mediated?
Antibiotic	Yes n (%)	No n (%)	Total n
Amoxicillin	13 (81.3)	3 (18.8)	16
Amoxicillin/Clavulanate	6 (75.0)	2 (25.0)	8
Cephalosporins	7 (87.5)	1 (12.5)	8
Clindamycin	10 (100.0)	0 (0)	10
Fluoroquinolones	16 (76.2)	5 (23.8)	21
Macrolides	3 (27.3)	8 (72.7)	11
Metronidazole	5 (100.0)	0 (0)	5
Nitrofurantoin	1 (50.0)	1 (50.0)	2
Penicillin	34 (50.0)	9 (20.9)	43
Sulfonamides	20 (80.0)	5 (20.0)	25
Tetracyclines	6 (85.7)	1 (14.3)	7
Topical antimicrobials	3 (100.0)	0 (0)	3
Total	124 (78.0)	35 (22.0)	159

The date of allergic reaction was collected in a categorical/ordinal manner. Ninety-eight of 159 surveys (61.6%) indicated the reaction date was 10+ years ago [Table 3]. However, immune-mediated reaction categorization was independent of the reaction date.

Table 3.

Date of Reaction.

	Reaction Date
Immune-Mediated	Missing Response n (%)	Within Past Year n (%)	1–5 Years Ago n (%)	5–10 Years Ago n (%)	10+ Years Ago n (%)	Unknown n (%)	Total n
No	2 (100.00)	3 (60.00)	16 (80.00)	18 (85.71)	77 (78.57)	8 (61.54)	124
Yes	0 (.00)	2 (40.00)	4 (20.00)	3 (14.29)	21 (21.43)	5 (38.46)	35
Total	2	5	20	21	98	13	159

Immune-mediated reactions included: difficulty swallowing, rash, hives, itchy skin, skin ulcers, shortness of breath, wheezing, swelling, and facial swelling. Non-immune mediated reactions included: gastrointestinal symptoms (ie, nausea, vomiting, constipation, diarrhea), cough, joint pain or stiffness, fatigue, and headache.

Rash (30.8%), hives (25.5%), and itchy skin (21.9%) were the most commonly reported immune-mediated adverse reactions (Table 4). Aside from cutaneous reactions, shortness of breath/wheezing was the most common immune-mediated adverse reaction (8.1%). Gastrointestinal symptoms (48.2%), headache (27.2%), and fatigue (17.3%) were the most prevalent non-immune mediated reactions reported (Table 5).

Table 4.

List of Immune-Mediated Adverse Events.

Adverse Events	Frequency n = 247	Percent (%)
Rash	76	30.8
Hives	63	25.5
Itchy skin	54	21.9
Shortness of breath/Wheezing	20	8.1
Swelling	14	5.7
Difficulty swallowing	10	4.0
Facial swelling (angioedema)	10	4.0

^aSubjects were able to select multiple adverse events to most accurately characterize their previous adverse reaction to the antibiotic. The total number of adverse events documented was 247.

Table 5.

List of Non-Immune-Mediated Adverse Events.

Adverse Events	Frequency n = 81	Percent (%)
Gastrointestinal	39	48.2
Headache	22	27.2
Fatigue	14	17.3
Joint pain or stiffness	4	4.9
Cough	1	1.2
Muscle pain	1	1.2

Discussion

This study demonstrated the feasibility of using a clinical history questionnaire to identify non-immune mediated adverse reactions to various antibiotics in an outpatient setting. The implication of this study is the possibility of implementing a protocol that can involve members of the patient care team with minimal to no specialized training. The next step of this initiative would be to perform confirmatory testing to corroborate the nature of the adverse reaction by performing skin prick testing, intradermal testing, and oral challenge. The ultimate goal is to directly de-label low-risk subjects that are incorrectly labeled to have an antibiotic allergy listed in the medical record without any further testing or inquisition. Low-risk subjects are individuals with antibiotic allergy histories that are inconsistent to be immune-mediated by their clinical presentation. This study differs from previous antibiotic stewardship initiatives as our study created a standardized questionnaire to investigate previous adverse reactions to investigate all antibiotics as opposed to a specific class of antibiotics.^4,5,15,18

Recent studies aiming to de-label antibiotics with clinical history alone showed promising results.^4,5,10,17,18 Reichel et al.¹⁷ created a 5-question algorithm to determine which of the past adverse reactions to antibiotics had a low risk of being an allergy. Subjects with histories determined to be low risk were directly de-labeled and were administered with the implicated antibiotic. Out of the 200 participants in the study, only 70 subjects provided a convincing antibiotic allergy history. One hundred twenty-four participants received the previously incriminated antibiotic (or a drug of the same class) and did not show an adverse reaction. Meanwhile, Lutomski et al.¹⁰ found 109 out of 139 (78%) reported antibiotic allergies to be immune-mediated after pharmacist-led patient interviews. The frequency of discrepancy found between patient interviewing and the patient’s medical record reinforces the importance of antibiotic allergy validation. Devchand et al.⁵ directly de-labeled 14 out of 106 subjects (13%) after standardized questioning of hospitalized subjects. Chua et al.¹⁸ de-labeled penicillin allergies in 29% (355/1225) of their study population directly and through oral challenge. Of the 29% of subjects ultimately de-labeled, 45% of them were directly de-labeled without an oral challenge. Our study results further reinforce the feasibility of using a standardized questionnaire to differentiate non-immune mediated reactions from immune-mediated allergies to a variety of antimicrobials.

There appears to be a wide range of prevalence for antibiotic allergy mislabeling in various studies. A possible explanation for the differences is the extent to which these allergies are investigated in these different studies. For example, Du Pleiss et al.⁴ performed oral challenges and referred patients to an immunologist for further evaluation when the questionnaire and interview results were inconclusive. Their study differs from this study which aimed to investigate the feasibility of de-labeling low-risk antibiotic allergies from subjects by only using subject medical history and questionnaire.

Our questionnaire was tailored by the investigators to help categorize reactions as immune-mediated or non-immune mediated. Anstey et al.²⁰ described the successful utilization of the PEN-FAST algorithm in multiple study cohorts to risk stratify penicillin allergies. While there is an accepted risk-stratification guideline for penicillin (PEN-FAST), there is no standardized tool for other classes of antibiotics. More research is needed to develop an algorithm to reliably risk-stratify adverse reactions to all antibiotics. This study contributes to the development of a tool to investigate other classes of antibiotics.

A strength of our study is that our questionnaire accommodated for all antibiotic allergies as opposed to similar studies that only focused on penicillin allergies.^4,5,16,18 The screening questionnaire can also be administered by all members of the care team with little to no additional training. The questionnaire can be completed by the patient in a short amount of time with little to no guidance after the initial instructions are given. This makes the questionnaire ideal in outpatient settings. Furthermore, though the study ultimately included less than 100 individuals, the percentage of individuals with adverse reactions that could not be explained by Type I-IV hypersensitivities was less than what was conservatively planned for, resulting in an even more precise and smaller 95% confidence interval half-width.

A limitation of our study is that the data collected is subjective to patient-reported events, including events that occurred several years ago or in childhood. However, relying on limited patient histories is common practice in clinical decision making. The poor memory of subjects affecting the quality of the data could be combatted by corroborating their self-reported history by investigating their medical record. An additional challenge is that the electronic medical record was implemented relatively recently so adverse events that happened in the distant past may not be documented.

An additional limitation of our study is the lack of objective confirmation of reported reactions (eg, skin testing, oral challenge). Regardless, this study showed that identification of de-labeling candidates from a screening questionnaire is feasible. Future studies are necessary to perform confirmatory clinical testing to corroborate our results.

Despite these limitations, our study achieved our objective of utilizing a standardized questionnaire to efficiently identify low-risk adverse reactions to antibiotics that are likely not immune-mediated. More questionable, high-risk adverse reactions should be investigated more thoroughly with an allergist.

Conclusion

We were able to demonstrate the success of using a standardized questionnaire in the outpatient setting to identify low-risk antibiotic allergy labels and can lead to the subsequent removal of inappropriate allergy labels. Greater efficiency in identifying low-risk subjects will optimize antimicrobial use in individualized care and improve overall antibiotic use within the medical community. Future initiatives are needed to enhance patient and physician education on antibiotic allergies to prevent re-labeling of antibiotic allergies.²⁰

Appendix A. Antibiotic Allergy Questionnaire

Name
Date of Birth
Gender	□ Male
	□ Female
	□ Other _________
	□ Prefer not to disclose
Ethnicity

1. Have you ever experienced an adverse reaction to an antibiotic before? (If you have experienced a reaction to more than 1 antibiotic, please fill out 1 form, questions 1–11, per antibiotic reaction)

   • □ Yes
   • □ No
2. Which antibiotic did you take? Please select only one.

   • □ Penicillin
   • □ Amoxicillin
   • □ Azithromycin (ZPak)
   • □ Amoxicillin/Clavulanate (Augmentin)
   • □ Clindamycin
   • □ Cephalexin (Keflex)
   • □ Ciprofloxacin (Cipro)
   • □ Sulfamethoxazole/Trimethoprim (Bactrim)
   • □ Metronidazole
   • □ Levofloxacin (Levaquin)
   • □ Doxycycline
   • □ Other: _________________
3. Date of adverse reaction (approx. okay)

   • □ Within the past year
   • □ 1–5 years
   • □ 5–10 years
   • □ 10+ years
   • □ Unknown
4. Why were you taking an antibiotic?

   • □ Upper respiratory infection (throat infection)
   • □ Lower respiratory infection (pneumonia, bronchitis)
   • □ Ear infection (otitis media)
   • □ Skin infection
   • □ Urinary tract infection
   • □ Other: _________________
5. How were you taking the antibiotic?

   • □ Oral (by mouth)
   • □ Intravenous
   • □ Topical
   • □ Intramuscular
6. What was your reaction to the antibiotic? Please check all that apply.

   • □ Nausea
   • □ Vomiting
   • □ Abdominal pain
   • □ Constipation
   • □ Diarrhea
   • □ Blood in the stool
   • □ Heartburn
   • □ Difficulty Swallowing
   • □ Black stools
   • □ Rash
   • □ Hives
   • □ Itchy skin
   • □ Skin ulcers
   • □ Shortness of breath
   • □ Asthma/wheezing
   • □ Cough
   • □ Spitting up blood
   • □ Joint pain or stiffness
   • □ Muscle pain
   • □ Muscle spasms or cramps
   • □ Swelling
   • □ Facial swelling (angioedema)
   • □ Fatigue
   • □ Headache
   • □ Other (please describe below): ___________

7. How soon after starting the medication did the reaction happen?

   Hours Days Weeks Months

8. Were you hospitalized from the reaction?

   • □ Yes
   • □ No

9. What happened after you stopped taking the antibiotic?

   ________________________________________

   ________________________________________

   ________________________________________

10. Have you undergone skin prick testing and/or oral challenge to the antibiotic?

    • □ No
    • □ Yes
    • □ Unsure
11. Please indicate which antibiotics you have taken since the time of the reaction (to the best of your memory)

    • □ Penicillin
    • □ Amoxicillin
    • □ Azithromycin (ZPak)
    • □ Amoxicillin/Clavulanate (Augmentin)
    • □ Clindamycin
    • □ Cephalexin (Keflex)
    • □ Ciprofloxacin (Cipro)
    • □ Sulfamethoxazole/Trimethoprim (Bactrim)
    • □ Metronidazole
    • □ Levofloxacin (Levaquin)
    • □ Doxycycline
    • □ Other: _________________