Fewer than 5% of people labelled allergic to penicillin actually are

A penicillin allergy label on a medical record is usually wrong. The PALACE trial's opening statement of the problem is that fewer than 5% of patients labelled with a penicillin allergy are truly allergic [s3]. A Lancet review of antibiotic allergy puts the same point in more detail: most patients labelled allergic to penicillins are not allergic when they are appropriately risk-stratified, tested and re-challenged [s1].

This matters more than a wrong entry in a chart usually does, because the label changes what antibiotic a person gets for the rest of their life.

Where the labels come from

The Lancet review sets out what these reactions actually were. Many reactions documented as allergies were unknown to or not remembered by the patient; others were cutaneous reactions unrelated to drug hypersensitivity, drug-infection interactions, or drug intolerances rather than immune-mediated allergy [s1]. That last category — a reaction produced by the interaction of a drug and an infection rather than by hypersensitivity to the drug — covers a common route to a lifelong label: the reaction was real, the attribution was never tested, and nothing removed it afterwards [s1].

Antibiotics are genuinely the commonest cause of life-threatening immune-mediated drug reactions, including anaphylaxis and severe cutaneous adverse reactions [s1]. That is why the labels get applied readily. The problem is not the caution; it is that nothing removes the label once the episode passes.

What the label costs

A penicillin allergy label is associated with increased use of broad-spectrum and non-beta-lactam antibiotics, which the review links to more adverse events and more antibiotic resistance [s1]. It describes these labels, collectively, as a global threat to public health [s1].

A population-based matched cohort study in UK general practice put numbers on that. It followed 301,399 adults with no previous MRSA or C. difficile in the Health Improvement Network database between 1995 and 2015: 64,141 with a penicillin allergy label and 237,258 comparators matched on age, sex and study entry time [s2]. Over a mean 6.0 years of follow-up, 1,365 developed MRSA and 1,688 developed C. difficile [s2].

Among people with the label, the adjusted hazard ratio was 1.69 for MRSA (95% CI 1.51 to 1.90) and 1.26 for C. difficile (1.12 to 1.40) [s2]. Their antibiotic use differed sharply: adjusted incidence rate ratios of 4.15 for macrolides (4.12 to 4.17), 3.89 for clindamycin (3.66 to 4.12) and 2.10 for fluoroquinolones (2.08 to 2.13) [s2].

The mediation analysis is the part that makes this more than a correlation. Increased use of beta-lactam alternative antibiotics accounted for 55% of the increased MRSA risk and 35% of the increased C. difficile risk [s2] — that is, most of the MRSA excess ran through the prescribing change the label caused, which is the mechanism you would predict if the label, rather than something about the patients, were doing the work. It is an observational study, so residual confounding remains possible, and the authors frame the conclusion as an association mediated by antibiotic choice [s2].

Why removing the label has been slow

The standard way to remove a penicillin allergy label in adults is skin-prick and intradermal testing followed by an oral challenge with penicillin [s3]. That is resource-intensive, confines the procedure to specialist-trained physicians, and restricts the global population who could be delabelled [s3]. In practice, most people with the label never get near the test.

The trial that tested a shortcut

PALACE was an open-label, multicentre, international randomised trial at six specialised centres — three in North America and three in Australia — running from 18 June 2021 to 2 December 2022 [s3]. Eligible adults had a PEN-FAST score lower than 3, PEN-FAST being a prospectively derived and internationally validated clinical decision rule for point-of-care risk assessment in adults reporting a penicillin allergy [s3].

Participants were randomised either to a direct oral challenge with penicillin, skipping the skin testing, or to the standard of care: skin testing followed by oral challenge [s3]. The primary outcome was a physician-verified positive immune-mediated oral penicillin challenge within one hour, in the intention-to-treat population, with a noninferiority margin of 5 percentage points [s3].

Of 382 adults randomised, 377 were analysed — median age 51 years (IQR 35-65), 247 (65.5%) female — with 187 in the intervention group and 190 in the control group [s3]. Most had a PEN-FAST score of 0 or 1 [s3]. The primary outcome occurred in exactly one patient in each group, 0.5% versus 0.5%, a risk difference of 0.0084 percentage points (90% CI -1.22 to 1.24) [s3]. The one-sided 95% confidence interval sat below the noninferiority margin [s3]. In the five days after the challenge there were 9 immune-mediated adverse events in the intervention group and 10 in the control group (risk difference -0.45 percentage points, 95% CI -4.87 to 3.96) [s3]. No serious adverse events occurred [s3].

The limits of what PALACE shows

The trial enrolled adults selected as low risk by a validated decision rule, and most of them scored at the bottom of it [s3]. Its finding is that in that specific group, direct oral challenge is noninferior to skin testing first — not that skin testing is unnecessary in general, and not that anyone with a penicillin allergy label can be challenged. Both trial arms were run in specialised centres [s3]. The trial was also open-label, though the primary outcome was physician-verified [s3].

What it changes is the resource argument. If the low-risk group can be delabelled with a single supervised oral challenge, the bottleneck stops being specialist skin-testing capacity, which is the reason most labels have never been examined [s3].

Risk stratification, testing and any penicillin challenge are supervised medical procedures. This article describes what the trials and cohort studies found; whether a particular label is safe to question is a clinical assessment, not a self-assessment.

Sources

• Antibiotic allergy — The Lancet, 2018-12-14
• Risk of meticillin resistant Staphylococcus aureus and Clostridium difficile in patients with a documented penicillin allergy — BMJ, 2018-06-27
• Efficacy of a Clinical Decision Rule to Enable Direct Oral Challenge in Patients With Low-Risk Penicillin Allergy: The PALACE Randomized Clinical Trial — JAMA Internal Medicine, 2023-07-17