In this study we found that implementation of an inpatient proactive systematic penicillin allergy de-labelling program with oral amoxicillin challenge was associated with both increased removal of penicillin allergy labels at 30 days from admission, and a greater utilization of beta-lactams when patients were prescribed antibiotics within 6-months of the intervention period. Furthermore, there were no adverse events of the direct oral challenge reported from the original direct oral challenge intervention. Introduction of de-labelling programs in the adult inpatient setting appears safe, effective, and can help patients preferentially receive beta-lactam antibiotics, typically the first-line class of antibiotics.

Only three patients (1.5%) of those with reported penicillin allergy labels in the non-intervention group had their penicillin allergy label removed. From previous literature, this is much lower than the expected proportion of patients with histories that would qualify them as low risk and appropriate for direct oral challenge [2]. This is reflective of the low level of de-labeling based on passive assessment by the pharmacy team during medication reconciliation.

Our study complements other studies in the literature, showing the effectiveness, safety, and effect on antibiotic prescribing of an inpatient penicillin allergy assessment program [7,8,9,10,11,12,13]. Recent studies by Ramsey et al. [9] and Chua et al. [10] support the safety of this direct oral challenge approach, and it is the standard of care for assessment of low risk penicillin allergies outlined in the current North American drug allergy guidelines [1].

Few studies have evaluated the downstream consequences of de-labelling on antibiotic prescribing [3, 9, 17], and typically do so by evaluating only those outcomes in patients receiving the specific intervention. In this paper, we demonstrate a significant impact at the level of all patients with penicillin allergy label who were located at a hospital receiving the de-labelling intervention, and its expected benefit compared to a parallel non-intervention cohort. These findings provide compelling evidence for the broader adoption of these approaches and their potential to be implemented as a part of institutional antimicrobial stewardship programs [3].

Commensurate with the aim of improving beta-lactam usage, we did see a trend towards reduced proportional non-beta-lactam use in the intervention group compared to the non-intervention group in 6-months after admission, which fits with the expected replacement of non-beta-lactam antibiotics by beta-lactam antibiotics for those who were de-labelled. While there was no significant effect of the intervention on incidence of C. difficile infection in the 3 months following admission, the outcome was rare. Initial admission length of stay was not markedly impacted by the intervention, but may have a delayed impact on decreased length of stay in subsequent visits, and warrants further evaluation.

Our study has several limitations. First, the original de-labelling intervention excluded a large number of inpatients prior to assessment. Many of the potentially eligible patients were significantly unwell with 56% of patients being excluded from assessment based on pre-defined exclusion criteria. The primary reasons for exclusion were cognitive issues that would prevent accurate assessment such as delirium, dementia, or active suicidal ideation. In spite of these exclusions, we still found a marked beneficial impact of the intervention. A second limitation of the study was that it was retrospective in nature and without randomization. While this does pose some limitations to inference, the study did benefit by the ability to compare two parallel cohorts that changed intervention periods, which may yield some additional balancing of confounders beyond those measured and adjusted for in the analyses (see Additional file 3: Table S1 for covariates between intervention and non-intervention groups). We do not believe that there would be significant carry-over effects between the implementation periods.

In summary, our study supports the adoption of inpatient programs for penicillin allergy de-labelling with direct oral challenge to help reduce the number of inappropriate penicillin allergy labels in patients and enable improved future utilization of first line beta-lactam antibiotics. Future studies are needed to answer whether these interventions result in improvements across other patient outcomes including length of stay, antibiotic toxicities, and even infection-related morbidity and mortality.