Development and Validation of a Sulfa Antibiotic Allergy Clinical Decision Rule

Trimethoprim-sulfamethoxazole is first-line treatment for many infections; however, use is limited by sulfa allergy.¹ Use of trimethoprim-sulfamethoxazole is frequently required by antimicrobial stewardship programs to prevent use of more restricted antibotics.² Studies with nonstandardized challenge criteria suggest that those with low-risk allergy phenotypes can safely undergo direct oral challenges (OCs); however, no current risk-stratification tool exists to guide challenges.^3,4 We sought to adapt PEN-FAST, a penicillin allergy clinical decision tool, for trimethoprim-sulfamethoxazole allergy.⁵

PEN-FAST (eFigure in Supplement 1)⁵ was adapted as a trimethoprim-sulfamethoxazole allergy clinical decision rule (SULF-FAST) for use and validation in 2 data sets (in Australia and the US). Patients aged 18 years or older with a trimethoprim-sulfamethoxazole allergy referred to drug allergy services in Melbourne, Australia (Austin Health, Peter MacCallum Cancer Centre; November 1, 2015, to July 31, 2022), or Nashville, Tennessee (Vanderbilt University Medical Center; October 1, 2015, to February 28, 2019), were prospectively assessed.^4,5 Patients with a nonsevere sulfa or trimethoprim-sulfamethoxazole allergy (ie, excluding anaphylaxis within 5 years and severe cutaneous adverse drug reaction),⁴ provided written consent to undergo OC at clinician discretion (eTable in Supplement 1). A positive test result was defined as a positive patch test (PT) result or a clinician-observed or patient-reported presumed immune-mediated reaction after the challenge. A PEN-FAST score (eFigure in Supplement 1) and its diagnostic performance were calculated for each cohort and allergy phenotype subgroup. Statistical analysis is detailed in the eMethods in Supplement 1. This study was approved by the Vanderbilt University Medical Center institutional review board and the Austin Health human research ethics committees. This study followed the STROBE reporting guideline.⁶

In this study, the Australian (n = 116) and US (n = 204) cohorts had a similar age distribution (median age, 64 years [IQR, 54-74 years] and 62 years [IQR, 48-70 years], respectively) (Table 1). Prevalence of a positive trimethoprim-sulfamethoxazole allergy test result was 5.2% (95% CI, 1.9%-10.9% [6 of 116]) in Australia and 6.4% (95% CI, 3.4%-10.7% [13 of 204]) in the US.

The PEN-FAST tool that applied to the Australian cohort showed good discrimination in determining true allergy, with an area under the curve (AUC) of 0.86. A low score (<3) carried low allergy risk (<5%), while a score of 3 or more had high allergy risk (>20%). A cutoff of less than 3 points classified 108 patients as low risk, with 2 having a positive result for 1 or more allergy tests (1 morbilliform drug eruption after OC, 1 positive PT). Sensitivity to identifying this allergy using this cutoff was 66.7% (95% CI, 22.3%-95.7%), with a specificity of 96.4% (95% CI, 91.0%-99.9%), positive predictive value (PPV) of 50.0% (95% CI, 15.7%-84.3%), and negative predictive value (NPV) of 98.1% (95% CI, 93.5%- 99.8%) (Table 2). Subgroup analysis showed good performance with delayed phenotype (AUC, 0.78 [95% CI, 0.56-0.99]); however, limited cohort numbers prevented performance assessment for immediate and unknown phenotypes.

The AUC for the US cohort was 0.67, with 179 patients scored as low risk, with 8 (4.5%) testing positive for trimethoprim-sulfamethoxazole allergy (all during direct OC; Table 1). A cutoff of less than 3 points showed sensitivity of 38.5% (95% CI, 13.9%-68.4%), specificity of 89.5% (95% CI, 84.3%-93.5%), PPV of 20.0% (95% CI, 6.8%-40.7%), and NPV of 95.5% (95% CI, 91.4%-98.1%) (Table 2). In subgroup analysis, performance for immediate reaction was good (AUC, 0.71 [95% CI, 0.37-1.00]), with lower performance in delayed phenotype (AUC, 0.61 [95% CI, 0.44-0.78]). Of 115 patients with 3- to 72-month prescribing data, 45 (39.1%) were administered a trimethoprim-sulfamethoxazole course (>1 dose), with no adverse reactions.

Application of PEN-FAST criteria provides a good initial framework for identifying appropriate challenge candidates among patients with low-risk trimethoprim-sulfamethoxazole allergy. Although sensitivities of 66.7% and 38.5% for the challenge reaction were lower than with PEN-FAST (70.7% [95% CI, 57.3%-81.9%]),⁴ the tool’s safety is reinforced by high specificity and NPV within both cohorts. Although our study was limited by the small cohort and single-dose challenge, SULF-FAST has potential utility for identifying individuals at low risk with low pretest probability of true allergy who could proceed to OC as a delabeling strategy, especially among high-yield populations, such as those undergoing a planned transplant or immunocompromised hosts for whom trimethoprim-sulfamethoxazole is first-line treatment. Although single-dose challenge was protocolized, the high rate of tolerance of multiple extended trimethoprim-sulfamethoxazole treatment doses is reassuring for excluding delayed hypersensitivity. Further validation is required for widespread implementation.

Accepted for Publication: April 20, 2023.

Published: June 5, 2023. doi:10.1001/jamanetworkopen.2023.16776

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2023 Waldron JL et al. JAMA Network Open.

Corresponding Author: Jamie L. Waldron, MD, Austin Health, Department of Infectious Diseases, 145 Studley Rd, Heidelberg, VIC 3084, Australia (jamie.waldron@austin.org.au).

Author Contributions: Dr Waldron had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Waldron, Rose, Krantz, Phillips, Trubiano.

Acquisition, analysis, or interpretation of data: Waldron, Rose, Vogrin, Krantz, Bolotte, Phillips.

Drafting of the manuscript: Waldron, Bolotte, Trubiano.

Critical revision of the manuscript for important intellectual content: Waldron, Rose, Vogrin, Krantz, Phillips.

Statistical analysis: Waldron, Vogrin, Bolotte, Phillips, Trubiano.

Obtained funding: Trubiano.

Administrative, technical, or material support: Waldron, Krantz, Phillips, Trubiano.

Supervision: Krantz, Phillips, Trubiano.

Conflict of Interest Disclosures: Dr Phillips reported receiving personal fees from UptoDate, Janssen, Biocryst, Verve, Regeneron, and Novavax and grants from the National Institutes of Health and Australia’s National Health and Medical Research Council outside the submitted work. No other disclosures were reported.

Data Sharing Statement: See Supplement 2.