Removing an inaccurate penicillin allergy label from a hospitalized patient’s chart can unlock first-line antibiotics that the label would otherwise block, but a new systematic review finds that how well such delabeling works depends dramatically on the method used. The review, published in the journal Drugs – Real World Outcomes, synthesized nine hospital-based studies, three randomized controlled trials and six nonrandomized interventional studies, and concluded that hospital-based penicillin allergy delabeling interventions appear safe and locally effective when embedded in structured clinical pathways. Yet the evidence base is so heterogeneous that the authors say universal hospital protocols cannot yet be formulated.
The stakes are considerable. Self-reported penicillin allergy is present in roughly 3 to 10 percent of hospitalized patients, but available evidence indicates that up to 90 percent of these labels are incorrect, according to background cited in the review. Such labels push clinicians toward second- or third-line antibiotics and are associated with increased adverse events, prolonged hospital stays, higher healthcare costs, and greater selection pressure for antimicrobial resistance. In critical care settings especially, antibiotic allergy labels often preclude first-line therapies entirely, promoting broader-spectrum regimens with downstream consequences for stewardship and resource use.
The review team searched PubMed/MEDLINE, CENTRAL, Scopus, and Web of Science up to January 2026, with additional records from ClinicalTrials.gov and citation tracking. Eligible studies enrolled hospitalized patients with documented penicillin or beta-lactam allergy labels who underwent structured delabeling interventions and included a comparator group receiving usual care. From 1848 records identified, 992 unique records were screened, 44 assessed in full text, and nine studies published between 2021 and 2025 made the final qualitative synthesis. Because of extreme heterogeneity in design, interventions, and outcomes, the authors judged a statistical meta-analysis unfeasible and synthesized the findings narratively instead.
The headline numbers vary enormously. Delabeling success rates ranged from approximately 4 percent to over 95 percent, depending on the intervention approach, risk of bias, and local workflows. The highest success was consistently observed in interventions using direct oral challenge, in which low-risk patients swallow a supervised dose of penicillin under monitoring. In such settings, negative challenge rates and successful label removal frequently exceeded 80 to 95 percent, particularly in controlled inpatient environments and randomized trials. The review’s authors point to the ORACLE trial by Rose and colleagues, in which amoxicillin challenges in critically ill patients were highly tolerated, yielding a negative challenge rate of 98.0 percent, or 39 of 40 patients, with zero antibiotic-associated serious reactions.
Pharmacist-led and antimicrobial stewardship-based interventions also performed well, effectively identifying inaccurate allergy labels and supporting safe delabeling through structured allergy assessment and protocolized evaluation. At the opposite end of the spectrum, passive system-level approaches, such as electronic health record-based label review and clinical decision support alerts, achieved superior institutional reach across thousands of patient-days but produced significantly lower absolute rates of confirmed label removal. The review attributes this to low clinician engagement, alert fatigue, and the absence of immediate diagnostic confirmation. In other words, digital prompts can flag thousands of candidate labels, but without a clinician actually testing the patient, the label often stays.
Safety data were reassuring across the board. Adverse events were uncommon and predominantly mild, and serious adverse reactions were rare. In the multicenter study by Drummond and colleagues, new adverse drug reactions occurred in only 0.9 percent of patients within 90 days. In the whole-of-hospital cohort evaluated by Chua and colleagues, a positive oral challenge was verified in just 3 percent of patients, 6 out of 200 tested individuals, presenting exclusively as mild, self-limiting cutaneous or localized hypersensitivity signals without any serious adverse events. In direct oral challenge pathways generally, severe systemic manifestations or anaphylaxis were virtually absent across the audited cohorts.
Beyond label removal itself, delabeling was associated with improved prescribing practices, including greater use of narrow-spectrum beta-lactams and reduced reliance on broad-spectrum or restricted antibiotics. Longitudinal studies suggested sustained effects on prescribing and label accuracy over extended follow-up. The review also highlights a subtle threat to durability: so-called label resurrection, in which a verified, successfully removed allergy label is accidentally re-entered into the electronic health record during subsequent admissions, often driven by automated software rollbacks or patient self-reporting. Despite low resurrection rates, tracking over 6-to-12-month horizons showed that maintaining an unverified status correlated with sustained increases in beta-lactam utilization and declines in restricted broad-spectrum antibiotic exposure.
The review also scrutinized its own evidence quality. Randomized trials generally showed low or some-concerns risk of bias, with concerns mainly around deviations from intended interventions. The nonrandomized studies, mostly quasi-experimental pre-post designs, carried higher risk, primarily from confounding and participant selection. Notably, the authors observe that studies with serious risk of bias, such as those by Drummond and Turner, often reported the highest success rates, reflecting selection biases in which non-blinded clinicians channel ideal, lowest-risk candidates into testing tracks. By contrast, the most methodologically robust randomized trial, the PROPEL trial by Stone and colleagues, showed far more modest real-world success at approximately 4.4 percent. The wide variation, the authors argue, is a direct product of protocol design rather than a uniform clinical truth.
On risk stratification, the controlled trials relied on localized, validated instruments, chiefly PEN-FAST and the Antibiotic Allergy Assessment Tool. PEN-FAST, which uses simple point-based parameters such as hypersensitivity history occurring over five years ago, anaphylaxis or angioedema, severe cutaneous reactions, and treatment requirement, showed the highest clinical feasibility and predictive accuracy for non-allergist staff, functioning as an easily implementable bedside tool with a high negative predictive value. The AAAT offers comprehensive multidimensional mapping but remains constrained to specialized or high-resource stewardship workflows, limiting rapid nonspecialist bedside use on standard medical floors.
The authors’ bottom line is a call for standardization rather than a universal protocol. The evidence base is structurally fragmented, they write, and extreme clinical and methodological heterogeneity limits the formulation of universal hospital protocols. They urge standardized, multi-center randomized controlled trials to establish uniform risk stratification algorithms and outcome metrics, including a shared definition of delabeling success that distinguishes a negative clinical challenge from formal electronic health record erasure and confirmed long-term retention. Until then, the review suggests, hospitals can implement structured, risk-stratified delabeling, ideally combining clinical assessment with direct oral challenge for low-risk patients, as a locally effective and apparently safe component of antimicrobial stewardship, while recognizing that results must be contextualized within their localized patient acuity boundaries and study design limitations.
Subject of Research: Effectiveness and safety of hospital-based interventions for removing inaccurate penicillin allergy labels in hospitalized patients
Article Title: Effects of Interventions for Delabeling Penicillin Allergy in Hospitalized Patients: A Systematic Review
Article References: Jovčić, F. Z., Milosavljević, M. N., Cmiljanić, M., Pejčić, A. V., & Stefanović, S. M. (2026). Effects of Interventions for Delabeling Penicillin Allergy in Hospitalized Patients: A Systematic Review. Drugs – Real World Outcomes. https://doi.org/10.1007/s40801-026-00594-2
Image Credits: AI Generated
DOI: 10.1007/s40801-026-00594-2
Keywords: penicillin allergy, delabeling, hospitalized patients, direct oral challenge, antimicrobial stewardship, systematic review, beta-lactam antibiotics, electronic health records, risk stratification, drug safety, Effects, Interventions
News Source: Ophelia Keating. (October 28, 2026). Hospital Penicillin Allergy Delabeling Works Best With Direct Oral Challenge, Review Finds, But Evidence Is Fragmented. Scienmag.
