Methods for Identifying and Reconciling Allergy Information in the Electronic Health Record

Abstract

Patient allergies are documented in multiple areas of the electronic health records (EHRs) including areas beyond the allergy list (e.g., notes, laboratory tests, and flowsheets). Systematically reconciling allergy information is critical to improve the accuracy of the EHR allergy list and facilitate clinical decision support to maintain patient safety. Our retrospective study included all patients who visited Mass General Brigham (MGB) in 2018. We examined patient allergy information documented in five EHR areas (i.e., allergy list, clinical notes, laboratory tests, flowsheets, and drug-allergy alerts) and proposed approaches to reconciling discrepancies between these EHR areas. We used natural language processing to identify free-text reactions in allergy lists. We consolidated duplicate allergens in allergy lists by mapping previous entries with the same name/ingredients into a preferred allergen. We manually reviewed amoxicillin challenge tests performed where the allergy was deleted, but then later re-added. We reviewed drug-allergy interaction alerts and identified active allergies with ≥3 allergy alert overrides by a clinician to identify active allergies unlikely to be clinically relevant. Of 1,588,979 patients studied, 266,275 (16.8%) had allergy lists indicating need for review and reconciliation. Allergy issues detected included free-text reactions (n=428,770), ≥3 allergy alert overrides (n=10,896), duplicate allergies listed (n=21,051), penicillin allergy active despite negative challenge test (n=36). Incomplete and inaccurate EHR allergy lists are common. Reconciliation methods may allow for allergy information identification and reconciliation. Further work is needed to develop automated reconciliation methods to ensure the accuracy of patient allergy lists and provide safe patient care.