Healthcare professionals’ experiences of de-labelling erroneous penicillin allergy records in general practice: a qualitative study

Participants

Of the 54 HCPs invited to participate in this study, 18 (33%) completed an interview. Of these, 13 were GPs, two were research nurses who consented patients to the trial, and three (2 = pharmacist, 1 = research nurse) performed drug provocation testing. The sample consisted of HCPs from general practices in Cornwall and across Yorkshire (reflecting the regions participating in the trial). Interviews were conducted between August 2023 and February 2024; this was towards the end of the trial when no more patients were being recruited. Interviews lasted 22–49 minutes (mean: 36 minutes).

First, it is noted, HCPs were encouraged to reflect on their experiences of each aspect of the trial (for example, identification of patient records, screening and eligibility, consent and randomisation, allergy testing, de-labelling, prescribing), including any challenges or areas for improvement. Participants were unanimous, and emphatic, in their praise for the trial, ‘I think this is probably the best study we’ve ever been involved with’ (P2, GP), and areas for improvement hinged on individual preferences (for example, video instructions for de-labelling in addition to written instructions). While we agree that ‘you don’t often hear the positives, do you; you just hear the negatives, so it is important to hear that, too’ (P1, GP), we elected to present findings that speak not only to participants’ views on the various trial processes, but also deepens our understanding of how the trial elicited behaviour change.

We present three themes that explore HCPs’ views on the reasons that led to the high prevalence of erroneous PenA labels, the patient and professional risks inherent in challenging labels without allergy testing, and finally, the ways in which ALABAMA mitigated these risks.

GPs’ beliefs about why patients have been incorrectly labelled: ’You just can’t make heads nor tails of it’

GPs reflected on the high prevalence of erroneous PenA labels and the ways in which these historical clinical decisions instil their current patient prescribing decisions with risk.

GPs believed they had ‘a fairly significant number of patients’ (P1, GP) who were incorrectly coded with PenA, often because side effects from penicillin-based medication or symptoms from their illness had been conflated with allergy, typically during childhood. However, as the primary event occurred years, often decades ago, the patient’s medical file did not provide their current GP with any context or rationale for the clinical judgement behind the PenA label:

‘But that’s [PenA label] based on a clinical decision and you have no idea about the clinician’s decision-making process, who made that decision, when it was made, maybe fifteen, twenty years ago.’ (P4, GP)

GPs explained this uncertainty was amplified among patients whose PenA label predated the digitisation of medical files as there was a paucity of detail in the paper records, or else they were illegible, ‘the paper records you just can’t read and there’s no detail (…) you just can’t make heads nor tails of it’ (P2, GP). This paucity of information made it challenging for GPs to assess whether the patient had a true PenA.

A smaller number of GPs also reflected on the ways in which GP understanding about penicillin had changed over time and that GPs themselves had conflated medication side effects with allergy in the past. One GP noted that their predecessors could not, or did not, anticipate the risks associated with such a liberal approach to PenA labelling based on scant clinical information, ‘documentation of penicillin allergy became an incredibly important thing and no one considered this could possibly have a downside to it’ (P11, GP). They went on to argue that all GPs were therefore ‘complicit’ in the preponderance of incorrect PenA labels in general practice.

Ultimately, GPs felt that there was no clear guidance on how to challenge a patient record with a historically ambiguous PenA label, or those presenting with non-anaphylactic reactions to a medication, ‘[if] they have anaphylaxis then yes, we will refer to the allergy clinic, but apart from that there’s no clear guidance to these allergy reactions’ (P8, GP). Consequently, GPs rarely, if ever, challenged patients’ PenA label and continued to avoid penicillin-based medications in favour of convenient second-line alternatives, ‘if there’s any doubt, we would just go with the alternative’ (P2, GP).

Penicillin allergy de-labelling without testing: ’If we’re feeling very, very brave, we might do a trial’

Indeed, most GPs reported being ‘very cautious’ about ‘trialling’ penicillin among those labelled as allergic. Some explained they ‘occasionally, if we’re feeling very, very brave, we might do a trial and see and get them to let us know if they were okay with it’ with the caveat that ‘that’s probably quite rare’ (P2, GP). However, others were much more reticent, ‘so most clinicians will be reluctant to prescribe against a coded allergy, if that makes sense. I probably don’t even question it very much’ (P4, GP).

The risk to patient safety when conducting ‘trials’ was the primary concern for GPs as they were ‘in the community with no backup’ (P14, GP) and feared the patient may experience (fatal) anaphylaxis. This was particularly poignant in one general practice that had a history of patient fatality attributable to PenA:

‘A long time ago we had a patient who suffered an allergic reaction to a penicillin and died because of that and that shadow is hanging over us, it’s more than 20 years since this incident (…) [but] we’re very risk averse.’ (P16, pharmacist)

One GP also emphasised the personal risk required to initiate such trials:

‘You know, we live in an environment where we have a vengeful regulator (…) no-one’s got any confidence in the GMC [General Medical Council] at all, that they’re not just gonna [sic] smash you. And so, everyone sort of feels in fear of their professional registration.’ (P11, GP)

GPs also felt patient beliefs about their medical history, particularly if they had been labelled as penicillin allergic for many years, would lead to reticence about trying penicillin or being de-labelled without undergoing allergy testing:

‘A lot of patients that think they’re allergic, that have been told for forty years that they’re allergic, are very resistant when you then say to them, "actually, if it was just that you had a tummy upset, that isn’t really an allergy"’. (P1, GP)

These risks notwithstanding, some GPs went on to discuss the conditions required to remove PenA labels without drug provocation testing. First, GPs reported that if the patient’s medical history revealed they had previously been prescribed and tolerated a penicillin, this would be sufficient for them to remove the label, ‘I mean sometimes you do have patients who say "well, I had a different penicillin a year ago and I didn’t have any reactions" and if I can see that then I will take allergies off, I will remove it’ (P7, GP). In such cases GPs interpreted this historical tolerance of penicillin as an indicator the patient was not truly allergic and they were therefore confident that a dangerous reaction was unlikely.

Second, in order to open up the idea of trialling, GPs emphasised the importance of ‘drilling down into the [patient’s] record’ to better understand the context of the primary event that gave rise to the label, and any subsequent events, ‘you say "well what actually happened?" and "oh when I was a child my mum said I just got sick with this one" and then you think, well, actually that’s probably not an allergy’ (P9, GP). GPs explained that this then opened up a ‘discussion with the patient’ (P3, GP) about their treatment options, which necessitated getting ‘them [the patient] to help assume some of the risk as well, like get them agreed towards a trial of that particular antibiotic if it was only a very mild allergy’ (P3, GP). But again, GPs emphasised theirs was an inherently cautious profession, and in practice, it was rare that ambiguous PenA labels were challenged.

Ways ALABAMA mitigated risk: ’Being [in a] protocolised and a protected supported bubble; makes you much safer to do it’

This theme examines the ways in which risk was mitigated through participation in the ALABAMA trial. Central to this theme was a guidance document, developed by the trial team, which enabled HCPs to check patient eligibility, obtain informed consent, and complete electronic referral to the trial where participants would then be randomised. HCPs were also reassured that each patient would be assessed at multiple timepoints within the trial (that is, eligibility screening, referral to PAAP) using these robust tools. It was through this, and other supporting guidance documents, that clinicians felt confident to identify low-risk patients, recommend allergy testing and refer them to the trial, and later, de-label patients who had tested negative.

As an exemplar, the guidance document provided to HCPs to aid patient eligibility screening has been reproduced in Table 1, but note, this is but one section of a larger instructional document.

Across all participants a central strength of the trial was that it was ‘totally protocolised (…) so once you’ve got the allergy history, you’re not left working out is that low risk? or high risk? (…) there’s a flow chart that leads you to the right answer’ (P17, pharmacist). Embedded within this ’algorithm‘ was a referral pathway to more senior clinicians for ambiguous or contradictory cases, ‘some were not straightforward so I had to speak to the clinician to find out if they were potentially eligible’ (P19, research nurse). While this could be frustrating as it delayed patient recruitment, particularly in sites where patients could not recall their ‘allergic reaction’ or reported vague symptoms such a rash, most were grateful for this added rigour in the consent procedure.

Additionally, the procedures for identifying and assessing patient risk were repeated at key points throughout the trial, including the identification of patients from primary care records, consenting patients, and at the test clinic. Again, this provided HCPs with assurance that only patients identified as ‘low risk’ would be tested, ‘patients have been stratified by the time they get to us, the GP’s gone through their record and so we’re only getting the low-risk patients anyway’ (P16, pharmacist, test clinic). HCPs felt confident that patients were being ‘properly tested in a safe environment’ (P2, GP) and emphasised that ‘as medics, we work in an environment where we test all the time, and if somebody devises a test to check for penicillin allergy and it comes back negative, you’ve got to trust that, haven’t you?’ (P6, GP).

Clinicians at test centres reported patients had been ‘well prepared during the early parts of the study’ (P16, pharmacist) but nevertheless ‘we probably went through in quite detail about what the testing would involve’, particularly as it related to whether patients would go straight to oral challenge test or have skin testing first. Consequently, few patients registered any questions or concerns about the test procedures or their personal risk being tested, and most accepted whichever testing pathway they were offered. Among the very small number of patients who HCPs reported had anxiety about being tested, most were reassured by clinicians’ two-pronged approach to first explain only those who are ‘very low risk get involved in the trial’ and second, ‘you are coming to a clinical environment which is very controlled’, and well equipped to manage emergency situations should they arise (P18, research nurse).

Four of the GPs in the study had consulted with patients who had been de-labelled as part of the trial. They characterised prescribing penicillin-based medications as unproblematic, ‘I have prescribed antibiotics which were penicillin based (…) and they [the patient] were pretty happy with it. They didn’t question me about it. They took it’ (P14, GP). Those without direct experience hypothesised they ‘wouldn’t hesitate to decide that’s the best antibiotic for them’ (P6, GP). Some GPs said they would spend additional time with patients to ensure they understood the change in their allergy status, were well informed about the potential side effects of the medication, and that they ‘were happy’ (P6, GP).