Algorithmic Identification of Patients With Aspirin-Exacerbated Respiratory Disease Using an Electronic Health Record

Abstract

Objective

To determine whether an electronic health record (EHR) system can be used to identify cases of aspirin-exacerbated respiratory disease (AERD) in an area outside of a regional referral center with low rates of aspirin desensitization therapy.

Study Design

Retrospective chart review single academic tertiary care hospital.

Setting

Single-site academic tertiary care hospital.

Methods

Using Epic's SlicerDicer function, an algorithm was created and applied to all patient charts from 2013 to 2021. The algorithm was as follows: “Allergy/Contraindication to NSAIDs OR aspirin” AND “Diagnosis of Nasal polyp AND “Diagnosis of Asthma.” Clinical data including demographics, NSAID reaction, and specialist involvement was collected.

Results

A total of 54 potential cases of AERD were identified. Thirty-two were determined to have AERD after chart review, yet 12 of these patients (37.5%) had no mention of AERD within the chart. The 54 patients were stratified into 2 cohorts based on reaction to NSAIDs: respiratory (n = 29) or unspecified (n = 25). Of the patients in the respiratory reaction group, 26 were found to have clinical AERD, demonstrating a positive predictive values (PPV) of 89.7%. The overall PPV was 59.3%. Those with a respiratory reaction to NSAIDS listed in the EHR were more likely to have clinical AERD (odds ratio 27.44; confidence interval 6.08-123.85; p < 0.0001). Only 2 patients (6.3%) underwent aspirin desensitization.

Conclusion

AERD remains under-diagnosed in the study population. The informatics algorithm presented here has a high positive predictive value for identifying clinical AERD patients in a geographical area with low rates of aspirin desensitization and may aid in identifying candidates for expanded treatment options.