Outcomes of a penicillin allergy de-labelling implementation intervention in a UK hospital: a prospective, single centre, type 2 hybrid effectiveness-implementation study evaluated using mixed methods

Background Non-allergy specialist healthcare workers removing low-risk penicillin allergy (penA) records, enabling patients to receive penicillin antibiotics, is safe but not common practice. We developed a behavioural change implementation intervention to support non-allergist healthcare workers to remove incorrect penA records from adult inpatients (penicillin allergy de-labelling; PADL). This study aimed to evaluate the effectiveness of the implementation intervention strategy and its impact on delabelling and antibiotic prescribing.

Methods The implementation intervention was launched 10th June 2024. This prospective, single centre, type 2 hybrid effectiveness-implementation study was evaluated using mixed methods: process evaluation, qualitative and quantitative study designs. The proportion of patients delabelled was measured. Antibiotics were grouped into the three WHO AWaRe categories with differences in antibiotic use between the de-labelled and non-de-labelled measured. Pearson chi-squared tests, Welch’s t-tests or Wilcoxon matched-pairs signed-rank tests were used.

Results Between 10th June and 13th December 2024, the mean number of penicillin allergy records removed from the inpatient electronic prescribing system increased from 18.2/month pre-intervention to 42.7/month post intervention. Of 186 patients de-labelled with full data capture, the majority 126 (67.7%) were de-labelled by the antimicrobial pharmacists and the remainder by ward doctors or ward pharmacists across 21 clinical specialties. Antibiotic exposure, when grouped by WHO AwARe category, was not significantly different between de-labelled and not de-labelled groups. Of 138 patients with a penA record in their primary care records, there was evidence of communication with the GP to amend the primary care penA record for 94 (68.1%) patients of which 42 (44.7%) were actioned by GP surgeries. The PADL intervention was accepted by most patients, who considered hospital a safe place to be tested. Awareness of the PADL intervention was generally high across all healthcare workers. The PADL process was acceptable to all interviewed healthcare workers and aligned with staff roles and hospital processes but competing priorities were commonly cited as a barrier.

Conclusions The PADL implementation intervention successfully increased non-specialist workforce PADL of low-risk patients. There is opportunity to further optimise PADL but competing priorities are a challenge. We found non-statistically significant impact of PADL on antibiotic use.