Authors' objectives: One in five inpatients carries a penicillin allergy label. However, 90–95% of labels are incorrect. Penicillin allergy labels lead to increased risk for serious hospital infections and longer hospital stay and are associated with higher estimated healthcare costs. Penicillin allergy testing is onerous and requires a specialist. Routine inpatient testing is not available. Recent evidence suggests that a direct oral penicillin challenge delivered by non-allergy specialists is safe in ‘low risk’ patients, who are highly unlikely to be allergic based on history. Penicillin allergy labels (PALs) occur in 6% of the population in England, and 15–20% of inpatients have a PAL in their records. Previous studies have shown that 90–95% of PALs are inaccurate. The current standard of care involves allergy specialist review, skin tests and a challenge procedure. There is an unmet need for allergy services and penicillin allergy testing is not routinely available. Penicillin allergy labels lead to use of alternative antibiotics which might be less effective, might enhance risk of antimicrobial resistance (AMR) and hospital-acquired infections, contribute to longer hospital stay, higher re-admission rates and mortality. Mis-labelling patients has been linked to suboptimal skills and knowledge gaps in allergy history taking and poor documentation in clinical records. Patients can be stratified into ‘low risk’ and ‘high risk’ based on a structured clinical history and scrutiny of clinical and prescription records. ‘Low risk’ patients are highly unlikely to have an immune-mediated reaction to penicillin. A direct oral penicillin challenge (DPC; without undertaking skin tests) has been shown to be safe and feasible and successfully delivered by non-allergy healthcare professionals (HCPs), such as clinical pharmacists and physicians. There is limited evidence from the UK, but there is a larger body of evidence from the USA and Australia regarding the success of DPC in mitigating the adverse impact of PALs. Safe implementation depends on an appropriate clinical setting and uptake amongst patients and relevant stakeholders. Primary To explore behaviour, attitudes and acceptability of patients, HCPs and managers regarding use of DPC in ‘low risk’ patients. To develop treatment pathways and a clinical governance framework for this service model.

Authors' results and conclusions: One thousand and fifty-four of 2257 screened patients were eligible, 270 of 643 approached patients consented (42%). Two hundred and fifty-nine patients were risk-stratified (155 ‘low risk’; 104 ‘high risk’). Of the 155 ‘low risk’ patients, 126 underwent direct oral penicillin challenge, 122 (97%) were de-labelled with no serious allergic reactions and 43 patients were interviewed. Low-risk patients accepted their allergy labels, had limited knowledge of the adverse impact and most were keen to have their labels reviewed. Healthcare professionals demonstrated a risk-averse approach, although would engage in the intervention with training, resource availability and a governance framework in place. The total costs of the direct oral penicillin challenge pathway were higher than the costs of direct oral penicillin challenge alone (£940 vs. £98–288 per patient). There were minimal expected savings in antibiotic and hospital costs in the short term and potentially large healthcare cost savings over 5 years. This first multicentre United Kingdom study showed that non-allergy specialist-led direct oral penicillin challenge is feasible in secondary care. A high proportion of direct oral penicillin challenges were successful, with positive feedback from patients. Majority of screened patients did not progress through the study pathway. Going forward, a multipronged approach is needed to enhance equitability of direct oral penicillin challenge in routine practice. Follow-up mechanisms to consider the intervention during a clinically stable state and a governance framework for those lacking capacity to consent are needed. The cost of delivering a direct oral penicillin challenge pathway in its entirety is significantly higher than the costs of performing direct oral penicillin challenge per se. Workstream 1 A total of 2257 patients (834 males and 1423 female) were screened across the 3 clinical settings and the 3 participating sites. One thousand two hundred and three (53.3%) were deemed ineligible based on study criteria at screening. A total of 1054 (46.6%) patients were considered eligible, 643 were approached and 412 not approached due to practical factors. Of 643 patients who were approached, 373 declined to participate and 270 (116 males and 154 female) consented. The overall conversion rate from screening to informed consent stage across all clinical settings was 12%. It was very low (3.3%) in the acute settings but greater in the elective settings at 17.7%. Progression in study pathway significantly was greater in Oxford (OR −2.06; p = 0.001), less in Leeds (OR −0.37; p 

Authors' methods: This study (1 May 2021–30 April 2023) involved delivery of direct oral penicillin challenge by non-allergy specialists across three clinical settings (medical/infectious diseases wards, presurgical and haematology-oncology units) at three hospitals. The study had three workstreams: Workstream 1: Screening for potential suitability. Patients were stratified into ‘low risk’ and ‘high risk’. ‘Low-risk’ patients underwent direct oral penicillin challenge. Workstream 2: One-to-one semistructured interviews with patients (N = 43) and focus group (N = 28) discussions with stakeholders. Workstream 3: Care pathway mapping, decision-analytic modelling and value of information analysis were carried out to determine potential cost-effectiveness of direct oral penicillin challenge. Relatively small sample size for direct oral penicillin challenge, poor conversion rate, particularly in acute settings, patients with limited English language proficiency could not be included and the study was not sufficiently powered and controlled to conduct a cost-effectiveness evaluation. This study spanned 24 months and involved three clinical settings [acute medical units (AMUs)/infectious diseases (ID) units, presurgical units and haematology-oncology units] in three hospitals including University Hospitals Birmingham NHS Foundation Trust, Leeds Teaching Hospitals and Oxford University Hospitals NHS Foundation Trust. The study included three workstreams (WS): Workstream 1 Patients with a PAL were identified and screened from respective Trust clinical records using a structured proforma employing predetermined study criteria. Clinically unstable patients, pregnant or breastfeeding patients, those with current COVID-19 infection, patients lacking capacity to give informed consent or taking part in another research intervention study were not approached. Those potentially suitable were approached for expression of interest. Informed consent was obtained from interested participants and a risk stratification process was conducted based on pre-determined study criteria. Participants were classified as ‘low risk’ and ‘high risk’. Risk stratification was conducted by non-allergy specialist HCPs (Research Pharmacists at Birmingham and Oxford and Research Nurses at Leeds) and a non-allergy specialist study consultant provided clinical supervision for the entire process. ‘Low risk’ patients were those who were highly unlikely to have a true allergy and ‘high risk’ patients gave a history suggestive of an immune-mediated reaction and/or had concomitant comorbidity, such as severe asthma or cardiac disorder. ‘Low risk’ patients were offered a DPC. This involved administration of a single dose of 500 mg amoxicillin and 1-hour observation to monitor for an immediate hypersensitivity reaction (HSR) followed by 250 mg twice daily for 3 days to assess for a delayed HSR. All patients were reviewed on day 5 to check for their clinical tolerance. One hundred and twenty-six DPCs were conducted [119 ‘opportunistic’ and 7 ‘therapeutic’ (to treat current bacterial infection)]. Data analysis Descriptive data were generated for all clinical settings across the three sites. To aid in the detection of potential associations, continuous variables were compared between the hospital/clinical setting groups using Kruskal–Wallis tests. Similarly, categorical variables were compared across hospital/clinical settings groups using Pearson chi-squared tests. Detailed analysis was provided to compare outcomes, such as de-labelled ‘Yes/No’, with the use of logistic regression. First, the outcome was regressed upon each potential ‘risk factor’, such as gender, producing an unadjusted odds ratio (OR; by exponentiation of the fitted parameter), and then a multivariable logistic regression used to yield adjusted ORs.

Authors' identified further research: Value of information analysis suggests that a randomised controlled trial with long-term follow-up for up to 5 years is needed to determine the cost-effectiveness of DPC by capturing relevant data relating to GP consultations for infections, hospital admission rates and duration, AMR, hospital-acquired infections and mortality. Future studies with opportunistic de-labelling may offer better value for money by designing the intervention as a non-allergy specialist pharmacist-led DPC along the lines of the model implemented at the Oxford study site.

Project Status: Completed

URL for project: https://www.journalslibrary.nihr.ac.uk/programmes/hsdr/NIHR129069

Year Published: 2025

URL for published report: https://www.journalslibrary.nihr.ac.uk/hsdr/MTYW6557

URL for additional information: English

English language abstract: An English language summary is available

Publication Type: Full HTA

Country: United Kingdom

DOI: 10.3310/MTYW6557

Organisation Name: NIHR Health and Social Care Delivery Program

Contact Name: Rhiannon Miller

Contact Email: rhiannon.m@prepress-projects.co.uk