Inconsistencies in penicillin allergy labels in hospital and primary care after allergy investigation.

Penicillin allergy is reported by 10% of hospital inpatients but up to 90% of penicillin allergy labels can be removed after drug allergy investigation.1, 2 However, if labels are not correctly updated patients with confirmed allergy may be at risk of subsequent severe allergic reactions and patients where allergy is disproved might receive suboptimal treatment potentially leading to longer hospital admissions and increased risk of acquiring antibiotic resistant infections.3 Up to 50% of patients have an incorrect penicillin allergy label in the electronic medical record (EMR) after penicillin allergy investigation.4, 5 In Denmark, hospital and primary care sectors have separate, non-compatible EMR systems and there is no automatic synchronisation between systems. Information is shared through electronic letters and allergy labels are updated manually. The primary aim of this study was to investigate the degree of match/mismatch between penicillin allergy labels in hospital and primary care EMR systems in patients who had undergone penicillin allergy investigation in a hospital allergy clinic. The secondary aim was to investigate whether age, sex, type of reaction at index reaction and DPT, change of general practitioner (GP) and the result of allergy investigation were associated with allergy label mismatch. The study population comprised patients ≥18 years who were investigated for penicillin allergy during 2017–2019 at the Allergy Clinic, Gentofte Hospital. Patients were identified from the department database. The database has been described in previous publication.2 All patients had undergone a drug provocation test (DPT) with a penicillin, prolonged for 3–10 days (depending on time before symptoms onset at the index reaction). Immediate reactions were defined as onset within 2 h of first dose and delayed reactions as reactions with onset after 2 h.2 Most patients had a DPT with phenoxymethylpenicillin (36%) or amoxicillin (41%). After DPT, the hospital EMR should be updated with the allergy test result. The GP should receive an electronic discharge letter with the results. The information and format of discharge letters and allergy labels in primary care are not standardised. By accessing individual EMRs, data on number of admissions since testing, name of present GP and current drug allergy labels were collected and entered into a REDCap database. In the hospital EMR, penicillin allergy is registered by the code at third anatomical therapeutic chemical (ATC) classification level (J01C). A letter informing about the study protocol was sent to all identified GP clinics and followed up with a phone call 1–2 weeks later. Data were either collected by phone or secure email. If the GP clinic was unavailable during first phone call attempt, at least one further attempt was made. The drug information was collected at the ATC classification level it was stated as in the EMR. If only free text was entered, it was translated to an ATC-code. Drug allergy registrations describing non-allergic side effects only were not registered as an allergy. If patients had more than one penicillin DPT, the positive DPT, if any, was used as allergy test result. Allergy labels were considered a match if the third ATC classification level matched, that is, if any type of penicillin allergy was registered in the EMR. For patients with disproved allergy (uneventful DPT), labels were considered a match if there was no penicillin allergy registration. A multiple logistic regression was performed with the mismatch between the penicillin allergy testing result and the penicillin allergy label at the GP clinic as the outcome. The following factors: sex, age, type of reaction at index, DPT results, number of admissions, new GP since DPT and number of GPs at clinic were investigated. p-values <.05 were considered statistically significant. Data were analysed using R version 4.2.0. A total of 849 patients from 390 different GP clinics were identified. Forty-one patients (14 GP clinics) were excluded due to for example, emigration, death etc. Out of the 376 clinics contacted, it was possible to receive data on 511 patients (60%) from 255 clinics (68%) (Table 1). The median age at DPT was 50 years (IQR: 37–63). Penicillin allergy was confirmed in 77 patients (15.1%) and disproved in 434 patients (84.9%). Penicillin allergy was confirmed in 9% of males (13/134) and 17% of females (64/377). A match in penicillin allergy labels between hospital EMR and allergy test results was seen in 94.3% (n = 482) of patients (Table 2). All 77 patients (100%) with a confirmed penicillin allergy were correctly labelled in the hospital EMR. Twenty-nine patients with a disproved allergy still had an incorrect penicillin allergy label in the hospital EMR, and of those, 16 (55.1%) also had an incorrect penicillin allergy label in primary care. Of the 405 patients with a disproved allergy and a correctly removed label in the hospital, 77 (19.0%) still had an incorrect penicillin allergy label in primary care. A penicillin allergy label was missing in primary care for 20 patients (26.0%) with confirmed allergy belonging to 20 GP clinics. Among patients with a disproved allergy, 93 patients (21.4%) still had an incorrect penicillin allergy label in primary care (79 GP clinics). Multiple regression showed that the odds were lower for having a mismatch in the GP clinic for males (odds ratio (OR): 0.52, 95% CI: 0.29–0.88, p-value: .018) and for those with three or more admissions to the hospital since allergy testing (OR: 0.31, 95% CI: 0.09–0.82, p-value: .033) (Table 3). Age, sex, index reaction type, DPT result, change in GP and number of GPs in the GP clinic were not associated with mismatch. This study showed that there was a large mismatch between the penicillin allergy test result and the allergy label in primary care, while the mismatch was smaller between allergy test result and the hospital allergy label, also reported in another study.5 Mismatches in the hospital EMR only concerned patients where allergy labels were not removed despite negative allergy testing (Table 2). One explanation for this could be that allergy labels are typically updated when writing the discharge letter, which might be delayed in patients still under follow-up in the department. It is also possible that some patients were relabelled, either intentionally due to a new allergic reaction after DPT or accidently for example, by incorrect information from a non-updated label in a subsequent referral from the GP. As hospital labels were correctly updated in most cases, but primary care labels were more often incorrect, sector shifts seem to have implications for the allergy label updates. Overall, about one in five labels were incorrect in primary care, which although high, is lower than reported in other studies.4, 5 In our region, there is no automatic exchange of allergy label information between primary care and hospital EMR systems. The information about a disproved or confirmed penicillin allergy is instead written in free text in an electronic discharge letter to inform the GP. Discharge letters' layout may affect the likelihood of the information being noticed and registered in primary care.6 Also, discharge letters may be read during the patients' next appointment causing a delay of months/year in the update of allergy labels, especially if the drug allergy referral was not sent from the GP and if the patient rarely seek the GP. Another reason for mismatch could be that the patient feels unsafe about future penicillin use, despite negative testing.5 In such cases, to maintain a trusting doctor-patient relationship and improve antibiotic compliance the GP may choose not to remove the allergy label. An explanation for patients with three or more admissions to hospital after allergy testing having higher odds of a correct penicillin label, could be that the GP receives several discharge letters, including allergy status, or that such patients have more frequent contact to their GP. The reason for a significant difference in OR for sex is uncertain and similar results cannot be found in other studies. Limitations of this study include the relatively high number of not included patients and clinics (45%), potentially causing selection bias. Clinics with little interest in allergy might not be as likely to update allergy labels and be less likely to participate in the study. This study focused on whether allergy labels matched the allergy test result and labels were not systematically updated. In theory, part of the observed mismatch could be explained by the occurrence of new allergic reactions after allergy testing. However, re-referrals to the allergy clinic due to new allergic reactions to penicillins are rare. Improvements in communication between health care sectors are necessary. Follow-up phone calls, involvement of community pharmacist or drug allergy passports have been suggested to improve communication.5, 7, 8 Changes between sectors, non- standardised discharge letters as well as non-compatible electronic systems may all contribute to the relatively high number of incorrect penicillin allergy labels found in this study. A national drug allergy warning system accessible from all sectors would facilitate communication regarding correct drug allergy status and thereby improve patient safety and management. L.H.G., H.F.M., S.R., A.H, J.H. and S.F. took initiative to the study with regard to the conception and design. S.F. and N.K. acquired data. S.F. and J.B.B. analysed the data. S.F. drafted the first draft. All authors helped with interpretation of data and provided critical comments on the intellectual content of the article. All authors have approved the final version. No conflict of interest. The study was approved by the Capital Region by the Center for Regional Development (Journal-nr.: R-20068776), by the Regional Centre for Research Approvals (P-2020-1090) and registered at Clinicaltrials.gov (GEH-R-20068776). Under the Danish law, ethical approval was not required for this study. The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.