Health outcomes of penicillin allergy testing in children: a systematic review

Abstract

Background

Penicillin allergy labels are commonly acquired in childhood and lead to avoidance of first-line penicillin antibiotics. Understanding the health outcomes of penicillin allergy testing (PAT) can strengthen its place in antimicrobial stewardship efforts.

Objectives

To identify and summarize the health outcomes of PAT in children.

Methods

Embase, MEDLINE, Web of Science, Cochrane Library, SCOPUS and CINAHL were searched from inception to 11 Oct 2021 (Embase and MEDLINE updated April 2022). Studies that utilized in vivo PAT in children (≤18 years old) and reported outcomes relevant to the study objectives were included.

Results

Thirty-seven studies were included in the review, with a total of 8411 participants. The most commonly reported outcomes were delabelling, subsequent penicillin courses, and tolerability to penicillin courses. Ten studies had patient-reported tolerability to subsequent penicillin use, with a median 93.6% (IQR 90.3%–97.8%) of children tolerating a subsequent course of penicillins. In eight studies, a median 97.3% (IQR 96.4%–99.0%) of children were reported as ‘delabelled’ after a negative PAT without further definition. Three separate studies verified delabelling by checking electronic or primary care medical records, where 48.0%–68.3% children were delabelled. No studies reported on outcomes relating to disease burden such as antibiotic resistance, mortality, infection rates or cure rates.

Conclusions

Safety and efficacy of PAT and subsequent penicillin use was the focus of existing literature. Further research is required to determine the long-term impact of delabelling penicillin allergies on disease burden.

Introduction

Penicillin allergy is the most common drug allergy reported in children, with about 5% labelled as penicillin allergic. After appropriate testing, fewer than 5% of these can be confirmed as having clinically significant acute-onset IgE-mediated allergy or delayed-onset T cell-mediated hypersensitivity.¹ Children are often misdiagnosed as penicillin allergic because an allergic rash is often difficult to distinguish from viral exanthem.² In a large recent study, the median age of obtaining an allergy label was 1.3 years (IQR 0.9–2.3 years),¹ and in children, 75% of penicillin allergy labels (PALs) are obtained by the age of 3 years.³

Spurious PALs are a challenge to antimicrobial stewardship. PALs are associated with poor health outcomes, such as increased mortality, Clostridioides difficile and MRSA infections, treatment failure, increased length of stay (LOS) in hospital, and increased healthcare costs.^4–7 To combat this, penicillin allergy testing (PAT) and delabelling have been promoted by the WHO as an effective antimicrobial stewardship strategy.⁸ PAT has been shown to reduce glycopeptide and fluoroquinolone use in adult inpatients,⁹ but similar outcomes have not been documented in cohorts containing only children.

PAT has become a recent focus of antimicrobial stewardship programmes, so it is valuable to examine whether it produces long-term health benefits in children. Establishing this can accelerate delabelling programmes and build confidence in patients, carers and prescribers for using penicillin. The objective of this systematic review was to identify and summarize the health outcomes that have been reported after PAT in children.

Methods

The Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) was used for the systematic review.¹⁰ The protocol was published in advance (PROSPERO CRD42021290366).

Eligibility criteria

We included primary studies that utilized in vivo PAT in penicillin-allergic children aged 18 years or younger. In vivo tests were defined as any form of skin test or oral challenge test. We included studies that stated ‘beta-lactam allergy’ if more than 50% of that population were reported to be penicillin allergic. Studies were excluded if they stated ‘beta-lactam allergy’ but did not specify which β-lactams. Studies that recruited both adults and children were included if the results for children were separately reported. We only included studies that reported other outcomes in addition to the result of PAT (i.e. a positive or negative test). Studies that did not report other outcomes were excluded. Case reports and economic evaluations were excluded. Only English abstracts were included.

Search strategy

The search strategy was developed with the help of two health service librarians independently. Search terms were developed with the synonyms of the following concepts: children, penicillin (or beta-lactam) allergy and allergy tests. Search terms related to the study outcomes were not used in order to maximize sensitivity. The full search strategy is listed in Table S1. We searched Elsevier Embase, PubMed MEDLINE, Clarivate Web of Science, Wiley Cochrane Library, Elsevier SCOPUS and HDAS CINAHL from their database inception through to 11 Oct 2021. The search on Embase and MEDLINE was repeated on 11 April 2022. No restrictions were applied to publication status in the search phase. Targeted grey literature search for unindexed abstracts in the past 2 years was conducted on American Academy of Allergy, Asthma & Immunology, British Society for Allergy & Clinical Immunology, and European Academy of Allergy and Clinical Immunology.

Study selection

Duplicate results were removed, and M.K. screened all abstracts against the eligibility criteria. A random one-third of abstracts were screened by a second reviewer (A.M. or R.P.). Full texts were then retrieved for all records that met the eligibility criteria and were all independently screened by two authors. Disagreements were resolved by consensus or discussion with a third author if required.

Assessment of methodological quality

Quantitative studies were assessed for methodological quality using the National Heart, Lung, and Blood Institute Study Quality Assessment Tools¹¹ by one author (M.K.). A separate assessment tool was used for each study design—controlled intervention studies, observational studies, and pre–post studies. A third of the quality assessments were double checked by a second reviewer (A.M.). Each study was given an overall quality rating of ‘good’, ‘fair’ or ‘poor’ based on their methodology and potential for bias. The studies were additionally given a rating (1–5) based on the Quality Rating Scheme for Studies and Other Evidence. Studies were not excluded based on their quality.

Data extraction

Data were extracted by one reviewer (M.K.), and one-third of extractions were double checked by a second reviewer (A.M.). Full texts in foreign languages were translated using Google Translate. The data extracted included study design, setting, sample size, type of allergy challenge, duration of follow-up period, number/proportion of patients delabelled, patient/carer acceptability of testing or penicillins, subsequent exposure and tolerability to penicillins, effect on hospitalization or LOS, healthcare-associated infections, and any other relevant health outcomes. Two attempts were made at contacting authors to clarify relevant data by e-mail.

Data synthesis

We utilized the synthesis without meta-analysis (SWiM) reporting guidelines.¹² We grouped the narrative synthesis based on type of outcome reported. The effect size for dichotomous outcomes was calculated as median with IQR. Heterogeneity was not assessed statistically, but we examined the comparability between studies based on their outcome measurement.

Results

Our search identified 1363 unique articles after removal of duplicates. We evaluated the full text of 101 articles and included 37 articles in this systematic review. There was a 98% and 97% interrater agreement rate during the abstract and full-text evaluation respectively. The PRISMA flow chart can be found in Figure 1. No amendments to the extracted data were required after it was checked by a second reviewer.

Figure 1.

PRISMA flow diagram.

There were 25 full manuscripts and 12 conference abstracts. Of these, there was one randomized controlled feasibility trial (82 participants),¹³ one pre–post study with no control group (1028 participants),¹⁴ 34 observational studies (7287 participants)^2,15–46 and one qualitative study (14 participants).⁴⁷ In the observational studies, there was a median of 111 participants (range 8–1914). The randomized controlled feasibility trial¹³ compared the effect of oral challenge (OC) versus no OC on the LOS in the paediatric emergency department where the PAT took place. A summary of results table for all included studies can be found in Table S2.

Methodological quality

The full results of the assessment of methodological quality can be found in Table S3. There was 1 study rated ‘good’, 31 rated ‘fair’ and 4 rated ‘poor’. Studies were rated poor because they had inconsistent exposure or outcome measurement and low participation rates. Only two studies included justification for their sample size.^43,44 The qualitative study was not assessed for methodological quality.

Study outcomes

Study outcomes were collated and categorized into four aspects: (i) penicillin allergy labels; (ii) subsequent exposure and tolerability to penicillins; (iii) patient and carer acceptability; and (iv) disease burden.

Penicillin allergy labels

Delabelling was reported in 14/37 studies.^{13–16,18,23,25,27,28,30,43–46} Studies had different definitions of ‘delabelling’, and outcomes that were expressed as proportions had different denominators. Studies were divided into three groups based on outcome measurement. The results are presented in Table 1.

Table 1.

Delabelling outcomes

	Outcome reported (numerator)	Number with outcome	Sample size	Definition of sample size (denominator)	Proportion with outcome (%)
Group 1: Studies that reported delabelling immediately after PAT
ȃAllen, 2020	Delabelled	99	102	All children who completed PAT	97.1
ȃBauer, 2021	Delabelled	51	52	All children who completed PAT	98.1
ȃHershkovich, 2009	Letter given to parents	96	98	All children who completed PAT	98.0
ȃLangley, 2002	Delabelled	66	69	All children who completed PAT	95.7
ȃLoprete, 2020	Delabelled	28	35	All children who completed PAT	80.0
ȃVyles, 2017	Delabelled	100	100	All children who completed PAT	100.0
ȃVyles, 2020	Delabelled	36	37	All children who completed PAT	97.3
ȃWang, 2020	Delabelled	53	53	All children who completed PAT	100.0
					Median of group 1: 97.3% (IQR 96.4%–99.0%)
Group 2: Studies that verified delabelling outcome
ȃChigullapalli, 2021	Delabelled in primary care medical records	Uncleara	Uncleara	Uncleara	56.0
ȃKleris, 2017	Delabelled in electronic medical record	28	41	PAT-negative children	68.3
ȃVyles, 2018	Delabelled in primary care medical records	47	98	PAT-negative children	48.0
Group 3: Studies that reported delabelling outcomes as part of a delabelling pathway
ȃLecerf, 2020	Delabelled	22	114	Children who were assessed using a nursing questionnaire	19.3
ȃSearns, 2020	Delabelled (post-intervention)	47	345	All children admitted with penicillin allergy	13.6
ȃWong, 2018	Delabelled	11	32	Children who were assessed using a delabelling algorithm	34.4

Eighteen children were eligible for PAT in the study. Percentage results were reported thereafter, but the denominator for each percentage was unclear.

In Group 1 (8/14 studies), delabelling was assumed upon non-sensitization after PAT. A median 97.3% (IQR 96.4%–99.0%) of children were delabelled in Group 1. In Group 2 (3/14 studies), delabelling was defined as an update to drug allergy records in electronic health records or primary care records. A range of 48.0%–68.3% of children were delabelled in Group 2. In Group 3 (3/14 studies), the definition was a combination of Group 1 and 2, but the outcomes were reported as a proportion of all children who underwent a delabelling pathway or nursing assessment, so the denominator was not comparable to the other groups. A range of 13.6%–34.4% of children were delabelled in Group 3.

The uncontrolled, pre–post study found a difference in the rate of delabelling between pre- and post-implementation of a clinical pathway (2.6% versus 13.6%, P < 0.0001).¹⁴ Three studies reported the use of letters to communicate the results of negative PAT.^18,23,46 One study asked primary care providers whether parents had informed them of negative PAT. Eighty-four percent of primary care providers said they were not notified of the negative PAT results, even though 90% of carers said they had done so.⁴⁴

Five studies examined whether children were relabelled with penicillin allergy. The rate of relabelling ranged from 0% to 3.3%. Two children³² were relabelled after reporting a reaction followed by positive repeat PAT, three^14,23,44 reported reactions but were not re-evaluated, and one¹⁶ had no documented reason for relabelling. The follow-up period for this outcome varied between studies; the shortest was at the point of discharge from hospital,¹⁶ and the longest was 112 months after negative PAT.²³

Subsequent exposure and tolerability to penicillins

Subsequent exposure to a treatment course of β-lactams was reported in 25 studies.^{2,16–24,26,29,31–42,48} The results for exposure and tolerability are presented in Table 2, where outcomes are grouped by the antibiotic measured. Penicillin-specific data were reported in 10/25 studies. The remainder of the studies reported a combination of penicillins and other β-lactams. Overall, the median tolerability to a subsequent course of β-lactams was 94.8% (IQR 91.6%–98.2%), and for penicillins 93.6% (IQR 90.3%–97.8%). Most intolerance reactions were mild, such as urticaria or maculopapular rash. One child, whose index reaction was anaphylaxis to co-amoxiclav, was oral co-amoxiclav PAT negative but developed anaphylaxis (urticaria, dyspnoea and gastrointestinal symptoms after 1 h of drug ingestion) to a subsequent course of co-amoxiclav.³²

Table 2.

Exposure and tolerability to β-lactams after negative PAT

Study	Antibiotic exposure measured	Follow-up period	Number exposed to antibiotic	Number of PAT-negative children followed up	Number tolerated (% of children exposed)
Penicillin re-exposure
ȃBauer, 2021	Penicillins	On discharge	42	82	NR
ȃChigullapalli, 2021	Penicillins	‘Few months’	Uncleara (31% of sample)	Uncleara	Uncleara (100)
ȃCorless, 1970	Penicillins	Not stated	3	8	3 (100)
ȃD’Netto, 2021	Penicillins	Not stated	231	452	219 (94.8)
ȃExius, 2021	Amoxicillin	5 years	265	890	226 (85.3)
ȃGraff-Lonnevig, 1988	Penicillins	1–4 years	110	198	103 (93.6)
ȃLabrosse, 2018	Penicillins	2 years	67	114	64 (95.5)
ȃMill, 2016	Amoxicillin	3 years	55	250	49 (89.1)
ȃPicard, 2012	Penicillins	‘Few years’	59	170	54 (91.5)
ȃSchechter, 2019	Penicillins	6 months	82	402	75 (91.5)
Median penicillin tolerability: 93.6% (IQR 90.3%–97.8%)
Penicillin and cephalosporins re-exposure
ȃPichichero, 1998	Penicillins and cephalosporins	6 months—8 years	163	163	160 (98.2)
β-Lactam re-exposure
ȃHershkovich, 2009	β-Lactam	44–112 months	59	71	58 (98.3)
ȃLobato, 2019	β-Lactam	Not stated	36	61	NR
ȃMisirlioglu, 2014	β-Lactam	At least 3 months	60	CDb	56 (93.3)
ȃMonteiro, 2018	β-Lactam	Not stated	19	86	17 (89.5)
ȃPaulino, 2022	β-Lactam	1–5 years	46	101	46 (100)
ȃPentland, 2019	β-Lactam	Not stated	72	153	72 (100)
ȃRegateiro, 2019	β-Lactam	6 months—10 years	104	212	99 (95.2)
ȃTonson la Tour, 2018 (PAT negative)	β-Lactam	Not stated	122	172	118 (96.7)
Culprit β-lactam re-exposure
ȃCelik, 2020	Culprit β-lactam (83% penicillinsc)	6 months	179	255	175 (97.8)
ȃColli, 2021	Culprit β-lactam (95% penicillinsc)	1 year	20	43	15 (75.0)
ȃJuchet, 1994	Culprit β-lactam (67% penicillinsc)	Up to 23 months	12	22	11 (91.7)
ȃPonvert, 2007	Culprit β-lactam (68% penicillinsc)	Not stated	93	141	86 (92.5)
ȃPouessel, 2019	Culprit β-lactam (85% penicillinsc)	11 months (mean)	30	78	28 (93.3)
ȃThimmesch, 2021	Culprit β-lactam (95% penicillinsc)	1 year	56	117	55 (98.2)
ȃTonson la Tour, 2018 (initial PAT positive, repeat PAT negative)	Culprit β-lactam (89% penicillinsc)	Not stated	11	14	11 (100)
Median tolerability for all studies: 94.8%d (IQR 91.6%–98.2%)

NR, not reported. CD, cannot determine.

Percentages were reported in the study with an unclear denominator.

Follow-up numbers included those with allergies other than to β-lactams. Only exposure/tolerability results were separately reported for β-lactams.

Percentage of the baseline sample size (not provided in this table).

Median of studies that reported tolerability outcome.

Twelve studies reported a breakdown of all antibiotic exposure in PAT-negative children. The results are shown in Figure 2. In children who were exposed to any antibiotics during the follow-up period, a median of 86.9% (IQR 78.0%–100%) received a course of β-lactams. The remainder received non-β-lactam antibiotics. However, in one of these studies (n = 130), 46.2% of respondents were unsure what class of antibiotic was prescribed subsequently.³⁶

Figure 2.

β-Lactam exposure in children who had at least one course of antibiotic and reasons why β-lactams were avoided.

Reasons for avoiding β-lactams were explored in 4/12 studies.^34,36,39,40 Most of the avoidance was due to patient or physician refusal, followed by reasons not related to penicillin allergy.

Patient and carer acceptability

Acceptability of PAT was explored in one study: 98% of parents were satisfied with the process, and 85% were confident to undergo a similar process in primary care.¹⁵ Four studies explored the level of comfort or confidence in using penicillins in the future, with 72.8%–100% of carers feeling comfortable.^15,18,34,44 One study found that the prevalence of parents feeling high or extreme anxiety towards amoxicillin decreased from 47.4% pre-PAT to 1.7% post-PAT.²⁶ The qualitative study found many families felt comfortable with resuming penicillins after negative PAT, and that delabelling brought relief as they no longer had to be afraid of certain medicines.⁴⁷

Disease burden

The randomized controlled feasibility trial reported a significantly increased LOS in the paediatric emergency department (ED) for children who receive PAT in the ED (n = 37) compared with children who did not receive PAT (n = 42) (216 versus 151 min, P < 0.01). The authors hypothesized that eliminating the time required for research consent process and randomization could eliminate most of the increased LOS.¹³

None of the included studies reported on any other outcomes related to disease burden, such as antibiotic resistance rates, mortality, infection rates or cure rates.

Discussion

Penicillin allergy labels

Delabelling is a complex process. Patient records should be updated to clearly communicate that a patient can receive penicillin antibiotics, whilst also preserving the historic PALs in their medical records. Delabelling a person does not mean they will not experience a hypersensitivity reaction in the future. This was demonstrated in one of the included studies where a child developed anaphylaxis upon re-exposure to penicillin antibiotics after initial delabelling.³²

There was a stark contrast in how studies reported delabelling. Although a median 97.3% of children were delabelled in Group 1, the studies in this group did not define what delabelling consisted of. It appears that all PAT-negative children were reported as delabelled, but the results do not agree with Group 2, where health records were checked to verify that allergy status was updated. The lower delabelling rate in Group 2 provides a reflection of current practice, where similarly, a previous systematic review of PAT in hospital inpatients found that 22.23%–92% of adults who were PAT negative were not delabelled on medical records.⁹ Without updating health records, children may not be able to receive first-line penicillin antibiotics where they are normally indicated.

The relabelling rate (0%–3.3%) identified in this review is much lower than that of the 5%–7% rate in previous published studies.^49,50 The five included studies had a wide range of follow-up period, which may skew the results. Additionally, most studies were not designed to detect this outcome and did not report this outcome, which may have underestimated the relabelling rate.

Although the evidence is drawn from a small number of studies, this review found that there is a gap between negative PAT and true delabelling from health records, highlighting the importance of penicillin allergy assessment pathways that include all aspects of the process, including updating of medical records. This is especially important for primary care providers, as they are responsible for at least 70% of all antibiotic prescriptions in England.⁵¹ This review identified several barriers for delabelling.

Firstly, the communication between healthcare providers was lacking or ineffective. For example, Vyles et al.⁴⁴ used the family to communicate negative PAT outcomes to their primary care providers, and Chigullapalli et al.¹⁸ sent letters. However, not all PAT-negative patients were successfully delabelled. On the other hand, some studies did not report how they relayed the delabelling instructions at all. Whilst the presence of formal communication such as clinic letters is important, the message included can influence delabelling as well. A survey of 86 surgeons and anaesthetists proposed two key elements of communicating pre-operative PAT results: (i) respondents valued the opinion of the allergist over the results of PAT, and (ii) respondents preferred concise and definitive statements that removed ambiguity and fear of allergic reactions.⁵² Future studies should endeavour to incorporate the most effective message that will encourage delabelling.

Secondly, a barrier exists in medical records systems. PALs appear to be recorded in different systems, such as paper notes, hospital electronic medical systems and primary care records. Delabelling on all systems can prove a challenge. Future research should focus on collaborative efforts between healthcare providers to improve effective communication and ensure PALs are adequately addressed in PAT-negative children. For example, interfacing health records between health systems or joint primary/secondary care records can be explored, so that a single record of delabelling can be viewed across systems.

Exposure, tolerability and acceptability

Antibiotic exposure was reported in many studies, but most of them were limited to the culprit allergen, penicillins, or β-lactams. Very few studies reported on non-β-lactam antibiotic usage. With the limited data available, we found that most PAT-negative children who required antibiotics were given β-lactams (Figure 2). This suggests PAT can be effective in shifting antibiotic prescribing trends towards penicillins.

In our review, there was a notable lack of comparator groups to establish whether there was a difference in prescribing trends in children who had PAT. In adults, case–control studies have shown PAT significantly increased penicillin use alongside decreased clindamycin and macrolide use,⁵⁰ and a reduction in infective readmissions.⁵³ Future studies should examine whether delabelling affects antibiotic exposure trends in children, particularly in a reduction of ‘watch’ and ‘reserve’ WHO classification antibiotics.

Where β-lactams were indicated but not prescribed, the biggest contributor to avoidance was carer refusal. We found that approximately 10% of carers remained uncomfortable with future penicillin use after negative PAT. Picard et al.³⁶ found that carers’ subjective fears were predictors of β-lactam refusal, whereas the severity of the initial reaction was not a predictor. Although PAT alone can reduce anxiety towards penicillins,²⁶ there was minimal focus on education and reassurance in our included studies. It is also possible that their initial experiences during allergy diagnoses can influence their level of fear. Protudjer et al.⁴⁷ highlighted that children experienced a rapid allergy diagnosis upon presentation of a red rash without allergy assessment. Excerpts from carers’ interviews suggested clinicians tended to avoid the worst-case scenario, and by doing so instructed carers to avoid penicillin.

Clinicians refusing to prescribe penicillins contributed to the remainder of cases. This reiterates the importance of sufficient communication with primary care providers. Negative PAT should be followed by specific reassurance that these children can be safely prescribed penicillins and instructions to address PALs in health records.

A review of patient perspectives of PAT and subsequent antibiotic use produced similar findings, where 7%–41% of patients continued to avoid penicillins after a negative PAT.⁵⁴ The authors found several contributors to penicillin avoidance such as patients’ lack of confidence in test results and not knowing which antibiotics were safe to take. This led to the development of a behavioural intervention package by Santillo et al.⁵⁵ to support clinicians and patients. Similar behavioural interventions should be trialled in children and carers to promote penicillin acceptability. Additionally, studies should explore how to support clinicians in the management of children who present with penicillin-allergy-like reactions, such as addressing their concerns or introducing educational interventions.

For delabelled children who were subsequently exposed to β-lactams, this review found that a high percentage (94.8%) of children tolerated it. A recent meta-analysis of PAT in children⁵⁶ showed that the rate of false negatives (positive second challenge after a negative first challenge) was 2.8%. Although our review found that a median 5.2% of children did not tolerate a subsequent course of penicillin, it may be an overestimation as this outcome was patient-reported or carer-reported in most studies.

Courses of penicillin appear safe in the vast majority of delabelled children. We noted one PAT-negative child, whose index reaction was anaphylaxis, developed anaphylaxis again on re-exposure.³² Indeed, the state of desensitization does not last a lifetime, and each exposure to penicillin carries a small risk of an allergic reaction. Recent literature found that for each exposure to a course of penicillin, 1 in 255 320 (0.0003%) oral courses is associated with anaphylaxis, and for parenteral exposure the incidence was 1 in 123 792 (0.0008%). Out of penicillin-associated allergy reports, 1 of 1543 (0.065%) oral-associated allergy reports and 1 of 1030 (0.097%) parenteral-associated allergy reports were confirmed to be anaphylaxis.⁵⁷ Published risk stratification tools^58,59 may be used to minimize this risk, but they have not been validated in children. Incorporating PAT into an antimicrobial stewardship programme will require a pragmatic approach. With a target to maximize delabelling whilst minimizing risks during exposure, children with non-severe, non-immediate reactions should be priority targets.

Limitations

Comparison between groups was not possible as most of the studies were single-arm observational studies with no comparator group. Children with general β-lactam allergies were included in this systematic review, which was a broader inclusion criterion than purely penicillin allergies. The inclusion of non-penicillin β-lactam allergies was not ideal, as penicillin allergies may drive the use of non-penicillin β-lactams. Some studies provided percentage results without reporting the numerator or denominator, so the data could only be analysed by the percentage provided. Only one study was given a rating of ‘good’ for methodological quality, although 12 studies were abstracts, which limited the amount of information the authors could convey.

Conclusions

Safety and efficacy of PAT and subsequent penicillin use were the focus of existing literature. The results suggested PAT led to penicillin use that was well tolerated. Penicillin avoidance was caused by both clinician and carer refusal. Allergy records were not updated in all health records despite initial delabelling, indicating a need for robust communication of test results and instructions to remove allergy labels from all records. However, long-term health outcomes such as disease burden were not reported in any of the included studies. Further research is required to determine the long-term impact of removing spurious allergies.

Author contributions

M.K., S.T. and J.Sc. designed the review protocol. R.P. reviewed abstracts. A.M. reviewed abstracts and full texts, and assessed the risk of bias for included studies. M.K. drafted the manuscript, which was reviewed and approved by all co-authors.

Funding

This work was supported by the Health Education England Research Internship for Pharmacists (Wessex Region). The funder was not involved in the development of the protocol and the manuscript.

Transparency declarations

None to declare.

Supplementary data

Tables S1 to S3 are available as Supplementary data at JAC Online.