Sustaining and spreading penicillin allergy delabelling: A narrative review of the challenges for service delivery and patient safety

Abstract

Many patients report allergies to penicillin, although in over 90% of these the label of penicillin allergy is shown to be incorrect following comprehensive testing. Inappropriate and inaccurate penicillin allergy labelling is a barrier to antimicrobial stewardship and can lead to patient harm. This review assesses an emergent evidence base and trend favouring delabelling using direct oral penicillin challenges following a stratified risk assessment of the likelihood and existence of true penicillin allergy, to identify and make recommendations for key components for implementation in standard practice. Research to date has focussed on the feasibility and clinical and financial outcomes of these direct delabelling strategies. There is a paucity of studies exploring the views and engagement of patients and healthcare professionals, and a gap in the evidence for prerequisites to safely deliver, sustain and spread the implementation of such services across health systems.

1. INTRODUCTION

Choice of antibiotic treatment depends on the infection and patient factors including their reported or documented allergy status. Penicillins are the first‐line antibiotics for many common infections and sepsis.1, 2 Six to 10% of the general population3 and 15–20% of hospital inpatients in the UK and USA carry a penicillin allergy (PenA) label, although emergent research shows that 90–95% of these labels are found to be incorrect following comprehensive allergy testing.2, 4, 5, 6, 7 Identification and removal of inaccurate and spurious PenA labels is referred to as delabelling.

Focus on antimicrobial stewardship (AMS) and concerns of inappropriate use of antimicrobials has led to greater interest in the impact of spurious PenA labels on clinical and operational outcomes, and a call for global action.8, 9 Inaccurate PenA labels are a major barrier to AMS and a patient safety concern.2, 4, 10 Large cohort studies from UK and USA show that PenA labels enhance the risk of serious hospital acquired infections such as methicillin‐resistant Staphylococcus aureus, vancomycin‐resistant Enterococci and Clostridioides difficile infections.3, 10, 11 Furthermore, PenA labels are associated with a higher risk of surgical site infections, lengthened hospital stay and greater use of more expensive antibiotics such as carbapenems and 6‐fluoroquinolones.11, 12, 13 The excess cost of alternative antibiotics per se in PenA patients has been reported at £250–500k per annum in a single National Health Service Trust in the UK14 and an estimated at US$64m attributed to longer hospital stay in PenA patients over a 3‐year period in the Kaiser Permanente Group of hospitals, south California, USA.11

Reports to the National Reporting and Learning System in the UK highlight an association between harm and allergy status, with nearly a third of all medication incident reports involving patients with known documented allergy to one or more medicine.15 Potential causative and contributory factors include the fact that the term allergy is often used interchangeably for intolerance, the diverse range of nonimmunological reactions that may occur and by errors and inadequacies in clinical documentation.16 Research has highlighted inadequacies in knowledge, skills and training amongst medical students and healthcare professionals in basic drug allergy history taking.17, 18

We posit that the gap between developing a PenA delabelling intervention and implementation into routine practice is likely to be significant. To embed, sustain and spread interventions, we need to understand not just whether interventions are effective, but also the prerequisites for their successful adoption and diffusion, taking into account behavioural and contextual factors.19 Therefore, effective PenA delabelling strategies require interventions that are sensitive to context. Whilst delabelling in specialist allergy clinics is established, there is currently little consensus on the ideal components of delabelling using oral challenges and associated implementation strategies. The aim of this review is to identify and assess current knowledge in relation to key components for oral delabelling challenges as reported in the literature.

1.1. Allergy status in medical practice

Establishing and documenting information about an individual's response to therapeutic agents is a core component of Good Medical Practice and record keeping.20, 21 In particular, documentation of any adverse responses, either due to known extension of the pharmacological action of the drug, or unexpected, unpredictable reactions that may be genetically determined or immunologically mediated, is key to ensuring avoiding inappropriate re‐exposure, ensuring patient safety and optimising continuing care. The term allergy is commonly and nebulously used to refer to and record all adverse responses. With the increasing use and interoperability of electronic health records, any allergy status documentation on the patient's record will transfer across different healthcare settings as part of the core medical information, making accuracy essential. In the UK, national guidance has been issued to facilitate diagnosis and management of drug allergy, with recommendations for assessment, documenting and sharing information with other healthcare professionals, providing information and support to patients, and nonspecialist management and referral to specialist services.16 For the final element, the national guidance sets out the subset of patients, including those with PenA labels, who should be referred to specialist allergy services. Similar recommendations for allergy identification, management and documentation have been made in the USA and Australia.22, 23

1.2. PenA delabelling methods

The diagnosis and assessment process for PenA has historically involved a systematic clinical history, review of previous records, skin tests, and a supervised penicillin oral challenge test (if skin tests are negative). Skin tests are labour‐intensive, time‐consuming and require specialist input.24, 25 Given the burden of PenA and huge unmet demand for allergy services, PenA tests are not routinely available to hospitalised patients.26, 27 Recent studies have suggested that positive skin tests do not always predict outcomes of an oral penicillin challenge, which is considered the gold standard test to exclude an allergy and confirm clinical tolerance.24, 28, 29, 30 This has led to trials of direct oral penicillin challenge in low‐risk patients (those most unlikely to be allergic based on risk assessment and stratification), thus obviating the need for skin tests without compromising safety and creating opportunities for delabelling without direct specialist input.

Direct oral penicillin challenges to delabel have gained favour on the premise that a vast majority (95–99%) of PenA labels are spurious due to inaccurate and incomplete documentation by healthcare professionals or inadequate patient understanding of what constitutes an allergy.31, 32 The first stage of direct PenA delabelling involves a comprehensive, structured assessment of the clinical history to establish a level of certainty and likelihood of the reported allergy. Clinical algorithms adapted from expert opinion, published studies and guidelines, have been proposed to aid structured risk stratification by nonspecialists.5, 9, 33 Paper and computer‐based stratification tools have been developed and employed at various stages of the patient's journey by clinicians and pharmacists in hospitalised patients and for preoperative testing.5, 34, 35, 36, 37 Application of these tools results in 1 of 3 possible outcomes: removal of spurious PenA label; referral to specialist allergy assessment services for those deemed to be high risk; or confirmation of PenA status.

1.3. Models and outcomes of direct oral challenge delabelling

Recent studies of newer approaches of direct PenA delabelling using structured review and algorithms have primarily focussed on safety and clinical effectiveness (see Table 1). Those conducted in hospital settings have involved a multidisciplinary team as a part of AMS programmes37, 38; and outpatient delabelling has mainly involved allergy specialist clinics.39, 40 Patient partnership is key to the success of direct Pen‐A delabelling, however, some patients do not consent to participate and even when they do, are not comfortable with re‐exposure.35, 41

Table 1.

Overview of oral penicillin challenge studies in the last 5 years (2014–2019)

Author, year	Patients	Setting and country	Intervention	Context details	Consent	Patient perceptions	Staff perceptions	Safety follow up outcomes
Savic et al 201934	Adults 119/219 patients stratified as low risk	Presurgical assessment UK	Risk‐stratified screening questionnaire Direct oral challenge—10%, 50% and 100% (500mg) amoxicillin and 3‐day course at home Hospital record updated; letter to general practitioner 5–7‐day postclinic follow up for delayed symptoms 3‐month follow up to check GP record	Dedicated delabelling clinic Facility to test for alternatives Full resuscitation equipment and personnel available 20 min between increments 1 h observation afterwards	163/219 agreed to testing, of whom 98/119 were classified low risk	For the 55 successfully delabelled patients 35/43 no anxiety on day 30/43 not happy with removal without testing 56 patients declined testing 25 never take whatever the result 11 not happy to take part in research 8 not enough time 12 other/ unknown	Not assessed	56 underwent challenge 1 urticaria after second dose 4 mild nonallergic symptoms during 3‐day course but completed course 2 patients penicillin avoided for surgical prophylaxis despite negative challenge 47/55 GP record correct; 3/55 retained allergy label
du Plessis et al 201941	Adults 250 eligible hospitalised patients	Tertiary hospital New Zealand	Electronic and manual review of allergy status by pharmacists Interview undertaken by pharmacist with outcomes of Delabel without challenge Oral challenge* under supervision Referral to immunology clinic *placebo, placebo, 5 mg, 50 mg, 500 mg (all suspension, in yoghurt) Patient education irrespective of outcome; information about applying for medic‐alert bracelet Letters to patients and primary care practitioners with outcome of interview and any intervention Electronic medical records updated after interventions 1‐month and 1‐year telephone interview	Exact location not specified Supervision by the primary treating team Pharmacist trained in preparation and administration of oral challenges at a local immunology clinic Doses given 30 min apart and for 24 h afterwards, unless a full course was indicated	3 declined 250 included	At discharge 119/199 delabelled patients happy to take again 57 only if there was no option 23 still not comfortable At 1 year 159/186 agreeable to taking	Not assessed	199/250 delabelled: 160 with no challenge; 31 after oral challenge; 8 referred to clinic 51 label confirmed: 24 with no challenge 3 with challenge (rash with or without itchiness at 27, 29 and 42 h postdose) 24 referred 23 lost to follow up (13 delabelled; 10 confirmed allergy) 3/186 delabelled patients were re‐labelled due to delayed reactions after re‐exposure
Kuruvilla et al 201953	Adults 50 patients with penicillin allergy labels out of 355 seen in an allergy clinic	Outpatient allergy clinic USA	Review of electronic medical record to identify patients Algorithm for risk stratification Delabelling without oral challenge if reaction was gastrointestinal upset or had received penicillin after the original label Direct oral challenge for those with penicillin exposure >12 months ago and lower risk using single dose of 500 mg Electronic medical record updated after intervention	Allergy clinic Baseline monitoring of vital signs and every 30 min for 60 min after therapeutic dose	20/38 who met criteria consented 18/38 declined; 9 due to apprehension about recurrent reactions	Only assessed in 9 of 18 patients who declined	Not assessed	4 delabelled with no oral challenge 3 patients developed subjective reactions not considered positive challenges: Diffuse pruritus, chest tightness and dizziness No reports of delayed reactions
Trubiano et al 201854	Adults 98 of 195 inpatients and outpatients with penicillin allergy considered low risk	Cancer patients Australia	Electronic medical record to identify patients Algorithm for risk stratification Low risk patients given oral challenge: Either oral penicillin VK 250 mg or amoxicillin 250 mg with prolonged 5‐day challenge (250 mg twice a day) for those with a history of delayed reactions	Infectious diseases and antimicrobial stewardship services and outpatient antimicrobial stewardship led allergy testing service Service provided by allergy nurse and infectious disease physician Observed for 2 h and followed up for 5 days	2 declined 46 consented 50 did not meet inclusion criteria	Not assessed	Not assessed	All patients delabelled with no adverse drug reactions in the 90 days after oral challenge
Arnold et al 201955	Paediatrics 176 children assessed for β‐lactam allergy	Tertiary paediatric hospital Australia	Retrospective review of standard care of direct oral penicillin challenge only or direct oral penicillin challenge with skin testing (if skin testing negative) depending on preference of person treating Oral penicillin challenge with suspected culprit antibiotic by administering 1/10 and then a full dose of the antibiotic 30 min apart if there was no reaction to the first dose 5‐day extended course for successful oral penicillin challenge	Allergy specialist/ immunologists service Observations for 1 h after challenge	Not known as retrospective study of those who had consented to attend allergy clinic	Not assessed	Not assessed	Oral challenge only—3 reacted Oral challenge after negative skin testing—4 reacted 3 of the 7 who reacted experienced anaphylaxis 6/132 children with negative oral penicillin challenge reacted to extended course
Lachover‐Roth et al 201956	Adults and paediatrics 741 of 784 ambulatory patients evaluated for penicillin allergy	Outpatient allergy unit Israel	Retrospective review Oral challenge test for 5 days following a skin test Medical records review to assess antibiotic purchase after allergy evaluation Phone survey to determine re‐exposure after allergy evaluation, reactions and perceptions to re‐exposure	Allergy and clinical immunology unit	Not known as retrospective study of those who had consented to attend allergy clinic	Yes—579 patients surveyed 96 would be willing to use penicillin 163 refused to use Lack of conviction of safety Inadequate understanding of results Refusal of family physician to prescribe	No, but patient survey indicated that a number of family physicians refused to prescribe	53/741 reacted during oral challenge test 19/344 survey patients reported adverse reactions 366/654 who were delabelled still had a penicillin allergy label on their electronic medical record, with 238 patients having purchased or been prescribed penicillin regardless
Moussa et al 201857,58	Adults 190 of 194 preoperative patients assessed for β‐lactam delabelling	Preoperative patients Canada	3‐step process Allergy unit consultation to determine likelihood of allergy Risk assessment Testing with skin testing followed by oral challenge ‐ single dose of 300 mg penicillin V or 500 mg amoxicillin for low risk patients ‐ graded challenge of same drugs at 10%, 30% and full dose for high risk patients Patients called 24 h post‐testing to report delayed reactions Electronic medical records updated	Preoperative staff involved in referral Experienced clinical staff performed clinical evaluations and testing Tests performed in interventional allergy care unit Allergist supervised for up to 2 h after last test dose Basic monitoring for 1 h after single dose Intensive supervision for graded challenge: Recliner chair, intravenous access and frequent vital sign and pulmonary function monitoring	All	Not assessed	Not assessed	44 patients delabelled without oral challenge based on skin test results and history 7 confirmed allergic by oral challenge
Vyles et al 201759,60 and 201858	Paediatrics 100 of 352 children with low risk symptoms	Paediatric emergency department USA	Risk assessment using penicillin allergy questionnaire 3 tier penicillin testing: Skin testing Oral challenge Single dose of 500 mg amoxicillin if negative skin test Graded dosing if positive skin test Electronic medical record updated Follow up with parents and primary care provider	Testing by paediatric emergency medicine or allergy and/or immunology fellows who were trained in allergy testing by a board‐certified allergist	82/434 classified low risk not interested	81/100 parents surveyed 90% aware of child being delabelled 59 would be comfortable to re‐expose to penicillin 19 somewhat comfortable and 3 not comfortable as fearful of repeat reaction	No assessment of perceptions but 98/100 primary care physicians surveyed 82 informed by patient families of delabelling 51 still had allergy label in medical record	100 patients delabelled 36 required antibiotics in follow up period, received 13 prescriptions of azithromycin, 26 prescriptions of penicillins and 7 of cephalosporins
Sundquist et al 201729	82 adult and paediatric patients with penicillin allergy listing	Allergy and immunology practice USA	Electronic health record identification Review by allergist for exclusions 3‐step allergy testing process: 2 skin tests, followed by oral challenge using 250 mg amoxicillin in those with negative skin tests Patient counselling including information about adverse drug reactions, that would not be considered allergy Letter for patient and primary care physician	Dedicated clinic Monitored for 60 min after oral challenge	12/82 declined 7/82 agreed but did not attend 1/37 who were skin tested opted out of oral challenge	1‐week and 6‐month follow up 28/31 who were followed up at 6 months would take penicillin/ amoxicillin in the future if prescribed All 31 thought penicillin allergy testing provided important medical information	7/8 referring physicians completed an online survey Estimated that of 50% of their patients with allergy who were asked to participate, <50% agreed Perceived barriers to recruitment (scored 1–10 where 10 is most important) Patient did want to take time (9.43) Physician lacked time to discuss testing with patient during the visit (7.86). Patient not wanting to risk having a reaction (5.43) or taking part in research (5.14) Physician forgot to discuss (5.43) or did not know patient had an allergy (4.14)	None tested positive to oral challenge 2 reported delayed nonallergic reactions 3/11 who were subsequently prescribed antibiotics received penicillin/ amino‐penicillin antibiotic
Chen et al 201736	Adults 252/1203 patients with a penicillin allergy flag	Multidisciplinary inpatient allergy service in large academic hospital USA	Electronic health record‐associated algorithms for identifying and prioritising patients Review by pharmacist screening for testing Oral challenge to amoxicillin 500 mg orally if skin tests were negative Removal of allergy label and results in notes Physicians and patients individually informed and counselled about the results and implications for future penicillin use	Multidisciplinary team; pharmacist‐led screening with allergist on‐call to address queries Testing materials streamlined An emergency reaction kit (epinephrine and diphenhydramine) carried by pharmacists Referrals through use of electronic algorithm or direct referral Patients monitored for 60 min after challenge	Not reported	Not assessed	Not assessed	252 evaluated, of whom 5 delabelled during interview as previously tested 1 patient developed urticaria within 1 h of oral challenge 16 relabelled despite successful delabelling documentation, education and counselling

Whilst these studies have generated proof of concept in favour of a direct oral penicillin challenge procedure for PenA delabelling, they were limited due to: relatively small sample size; little or no assessment of views and perspectives of healthcare professionals42 and patients regarding their confidence in embedding such an approach into routine clinical practice; lack of exploration of reasons for patients' unwillingness to consent to direct oral challenge or re‐expose to penicillins; and failure to update medical documentation with the outcome of the direct oral challenge. Although most studies have shown direct oral challenges to be safe (no documented anaphylaxis or serious delayed reactions), relatively mild cutaneous reactions after a direct oral challenge30, 40, 43 occur, justifying a place for such an intervention in acute care hospitals with an immediate access to management of allergic reactions.2, 44 Caution and concern about potential false negative tests for those patients where the index drug is amoxicillin–clavulanic acid or flucloxacillin has also been raised, unless these antibiotics are used for the confirmatory challenges.45, 46

Thus, there is a notable knowledge gap in respect of the requirements new service models and interventions place on the patients, healthcare professionals and organisations to implement and sustain change. Insights from the implementation literature suggests the need for targeted, theoretically informed interventions to promote change in healthcare professional behaviour and address organisational impediments to adoption.47, 48 Importantly, PenA delabelling studies have not yet addressed prerequisites with respect to clinical governance frameworks that are likely to vary between health services in different countries.

1.4. Challenges of spread and sustainability

With the growing global interest in PenA delabelling services to promote AMS and proven benefits in terms of clinical outcomes, one of the challenges is in moving from isolated trials of delabelling to establishing and spreading this as a model of care within and across different care settings. Clearly it is important to involve patients in clinical decisions prior to undertaking PenA delabelling, and yet there is little in the published literature to suggest that their perceptions and concerns have been addressed. Understanding and responding to patient perceptions of risk and reward is crucial to enable high uptake of delabelling programmes. Evidence indicates that proven treatments can take several years to become embedded into clinical practice.49 Application of improvement and implementation science approaches to focus on core elements of facets that lead to successful sustenance and spread of such interventions may help.50 A fundamental aspect of these is a better understanding of not just the intervention, but the contextual and infrastructural aspects that leads to successful improvements, with attention to beliefs and behaviour of patients and healthcare professionals.51

The evidence to date for direct PenA delabelling strategies has focussed on aspects of individual practice and pathways, such as risk stratification, importance of information accuracy and flow, interprofessional communication and training. Longer‐term outcomes, as well as broader aspects that are key to implementation spread and sustainability, such as wider organisational determinants and incentives, organisational responses to risk and psychological factors at the patient and physician level, are less well researched.

1.5. A way forward

When designing individual‐level interventions to change healthcare professional behaviours, 4 sets of tasks need to be completed: identifying barriers, selecting intervention components, using theory, and engaging end‐users.48 To sustain evidence‐based interventions, multiple facilitators, such as adaptation and alignment, and barriers, such as limited funding and limited resources, have been reported.51 These elements were reflected in our analysis of the evidence for direct Pen‐A delabelling interventions. We recommend that in order to design, develop, sustain, and spread safe and efficient delabelling interventions the following basic elements and prerequisites (Figure 1) should be considered and evaluated.

• Accurate risk stratification: Several studies52 have shown this to be feasible and successful as discussed above. National guidelines have been published in some countries to support the collation of relevant details about adverse responses and reactions on a prospective basis, but do not necessarily lead to a confirmed outcome. Combining these details through electronic health records with validated, structured algorithms would enable standardisation of risk stratification.

• Safe clinical environment: Few studies define the optimal setting and set‐up (monitoring protocol, rescue medication requirements) of the clinical environment in which direct oral penicillin challenges should be conducted. This information is essential for the sustainability and spread, as well as the development of business models to commission and deliver services.

• Multidisciplinary team: The involvement of a multidisciplinary team in identifying patients and managing treatment as well as updates to medical records is acknowledged in all studies.

• Trained staff: Most of the studies have involved individuals with a special interest or expertise in allergy; details of additional training for nonspecialists to deliver delabelling interventions are rarely reported. With the multidisciplinary and multiagency nature of healthcare provision across health and social care sectors, training for all relevant stakeholders and professionals needs to be considered.

• Defined governance framework: Few studies have explicitly considered governance frameworks in delabelling services. This is crucial to all stakeholders involved in such an intervention due to concerns regarding potential harm to patients and downstream medico‐legal consequences.

• Counselling and education tools: The high rate of safe delabelling without the need for skin tests indicates that patient understanding of allergy and the implications of a PenA label is an area that requires further attention. Similarly, exploring and enhancing healthcare professionals' knowledge, understanding and confidence in communicating with patients about allergies and the role of artificial intelligence systems to support risk stratification also requires further study.

• Updating electronic health records and communication with healthcare professional: Accuracy and completeness of documentation of suspected and confirmed allergy status may be a contributory factor in the overinflated reporting of PenA. There is little evidence of the role of intra‐operability between health IT systems in the transfer of allergy‐related information across different healthcare settings.

Importantly, future antibiotic use and antibiotic associated adverse reactions should be monitored to determine the sustained effectiveness of the overall delabelling program.

Figure 1.

Proposed prerequisites of a penicillin allergy oral challenge delabelling programme

2. CONCLUSION

Whilst strategies for direct PenA delabelling are being developed and tested, information on the behavioural insights and contextual requirements for successful implementation is scarce. The elements required for the sustainability and spread of such initiatives have resource and infrastructure implications. Despite health economic projections regarding clinical and cost‐effectiveness through reduction in use of high‐cost second line antibiotics, improved clinical outcomes and reduced length of stay, longer term safety outcomes and the business model for the commissioning and design of such services has rarely been reported. Similarly, the factors that influence individual patient and healthcare professional behaviours, and involvement of managerial and operational stakeholders in organisations are poorly understood. Future research and implementation strategies should therefore build on the work to date to address these gaps.

COMPETING INTERESTS

There are no competing interests to declare.

Jani YH, Williams I, Krishna MT. Sustaining and spreading penicillin allergy delabelling: A narrative review of the challenges for service delivery and patient safety. Br J Clin Pharmacol. 2020;86:548–559. 10.1111/bcp.14190