During the study period, 179 of 2890 screened patients, admitted to the participating wards, reported a PenA during medication reconciliation, resulting in an overall prevalence of 6.2% for the included wards. The highest prevalence was detected in cardiac surgery (35/391, 9.0%), followed by radio-oncology (37/474, 7.8%). ENT wards and vascular surgery showed lower prevalence of reported penicillin allergy (ENT: 90/1564, 5.8%; vascular surgery: 17/461, 3.7%) (p = 0.005).

Reaction types assessed during allergy assessment are shown in Table 1. The most reported reactions to penicillin antibiotics were skin rash (108/179; 60.3%), followed by non-immune mediated reactions like isolated nausea (24/179; 13.4%) and respiratory symptoms (16/179, 8.9%). A minor proportion of patients reported diffuse body swelling (5/179, 2.8%), prior positive prick-test (2/179, 1.1%), angioedema (1/179, 0.6%) and anaphylactic shock (1/179, 0.6%). Twenty-four patients (13.4%) were unable to characterize the nature of their reaction. Combinations of reaction types were reported in 18 patients (10.0%).

Patients were risk stratified according to the local penicillin allergy management guideline. Sixty-two patients (34.6%) were classified as low risk, 100 patients (55.9%) as moderate risk, and 17 patients (9.5%) as high risk of true penicillin allergy (Fig. 2). Time since index reaction was more than five years ago in 138 patients (77.1%), 38 (21.2%) of the reactions occurred during childhood. Overall five patients (2.8%) could not remember the time of the allergy reaction. The PEN-FAST score supported the clinical risk stratification, details can be found in the supplemental material 6.

Overall, 54 of 162 eligible patients (33.3%, 95% CI 26.1–41.2%) were delabeled following the delabeling intervention (Fig. 2). Based on allergy history, 34 of 62 (54.8%) low risk patients were directly delabeled without further testing. The remaining 21 patients underwent DOC. Of these, six patients were classified as low risk and 15 as moderate risk according to the local guideline. DOC was well tolerated in all but one patient, who developed an immediate, non-severe rash not requiring treatment. This patient had been classified as moderate risk by the local guideline and following the reaction retained the allergy label. Delabeling outcomes were documented in discharge letters for 47 of 54 delabeled patients (87.0%).

Stratification by ward demonstrated differences in the relative use of delabeling strategies (Fig. 3). On ENT and radio-oncology wards, a larger proportion of patients were directly delabeled based on allergy history alone (ENT: 15/24, 62.5%; radio-oncology: 9/10, 90.0%), while in cardiac surgery a higher proportion of patients underwent direct oral challenge (8/14, 57.1%). In vascular surgery, both approaches were used less frequently, reflecting the smaller overall number of eligible patients. Delabeling rate was highest in cardiac surgery (14/35, 40.0%) and lowest in ENT (24/90, 26.7%). Differences in delabeling rates across participating wards were not significant (p = 0.473).

Risk stratification according to local guideline and delabeling outcomes. DOC: direct oral challenge; Directly delabeled: Delabeled by allergy history without testing; Sustained delabeling: Patients without penicillin allergy label in primary care follow up according to 13 returned questionnaires

Use of delabeling strategies stratified by ward. ENT: ear nose throat

Implementation of PenA delabeling was influenced by a combination of facilitating and limiting factors. All identified barriers, including reasons for unsuccessful delabeling, as well as facilitators are summarized and mapped to the CFIR framework in Table 2.

Fifteen healthcare professionals completed the questionnaire on barriers and facilitating factors. The most frequently reported facilitators included interdisciplinary collaboration (10/15, 66.7%), the beneficial role of pharmacists in direct delabeling and coordinating the process (6/15, 40.0%), and training for ward staff (4/15, 26.7%). The availability of a local guideline with well-structured workflows (3/15, 20.0%), and easy integration of delabeling into routine medication assessment conducted by clinical pharmacists (3/15, 20.0%) further facilitated the delabeling process. Recurring barriers identified by healthcare professionals were related to limited medical staffing resources (6/15, 40.0%), high workload (6/15, 40.0%), short length of hospital stays (8/15, 53.3%), and patient-related factors such as language barriers, uncertainty about undergoing PenA delabeling, or refusal of delabeling (6/15, 40.0%). Structural and process-related challenges (6/15, 40.0%), particularly during the early implementation phase, further affected feasibility.

Overall, 125 of 179 patients (69.8%) could not be delabeled. Seventeen of them (13.6%) were classified as high risk according to the internal guideline and therefore not eligible for delabeling. Reasons for unsuccessful delabeling varied between wards. On the ENT ward, delabeling most often failed due to short hospital stays with discharge before completion of evaluation, combined with a high patient turnover and overall high workload. In radio-oncology, delabeling was frequently not pursued for clinical reasons (palliative situation, systemic corticosteroid therapy, critical clinical condition). In addition, ward physicians were initially partly sceptical towards delabeling, citing inexperience with assessing and managing allergic reactions. Acceptance of delabeling increased among ward physicians after targeted education. Across all wards, however, the most frequent reasons preventing delabeling were high-risk classification and staffing shortages among the involved pharmacists and infectious disease specialists, which limited the ability to provide patient education and perform testing.

Out of 54 delabeled patients, 33 (61.1%) agreed to follow-up with their primary care physician after eight months. A total of 13 questionnaires (39.4%) from primary care physicians were returned to the study team. In 10 of 13 patients (76.9%), the penicillin allergy status was “no penicillin allergy” in the primary care documentation. In nine of these 10 cases (90.0%), the primary care physician previously did not know about the patient’s penicillin allergy. Three patients (23.1%) had already been prescribed a penicillin antibiotic after hospital discharge. One patient reported a mild allergic reaction characterized by unspecific body swelling. However, the patient stated that they would take penicillin in the future if medically indicated.

Nine of 13 (69.2%) primary care physicians stated that they would be willing to prescribe a penicillin antibiotic to the patient in the future. Regarding discharge documentation, 3 of 13 (23.1%) primary care physicians reported that the delabeling was documented in the hospital discharge letter, whereas five physicians (38.5%) reported not having received a discharge letter, and another five (38.5%) reported that the delabeling was not documented. On review, the absent of documentation regarding the delabeling intervention could be confirmed in only one hospital record, suggesting that four of the five physicians likely did not receive or read the discharge letter. One physician stated that the patient herself reported on the delabeling.