Assessing the Validity of Electronic Medical Records for Identifying High Antibiotic Prescribers in Primary Care

• Main Authors: Warren J. McIsaac (Author), Sahana Kukan (Author)
• Format: Book
• Published: SAGE Publishing, 2023-11-01T00:00:00Z.
• Subjects: article
• Online Access: Connect to this object online.
• Notes: 2150-1327; 10.1177/21501319231210616
• Keywords: Computer applications to medicine. Medical informatics; R858-859.7; Public aspects of medicine; RA1-1270
• Source: Journal of Primary Care & Community Health, Vol 14 (2023)

Objectives: Electronic medical record (EMR) prescription data may identify high antibiotic prescribers in primary care. However, practitioners doubt that population differences between providers and delayed antibiotic prescriptions are adequately accounted for in EMR-derived prescription rates. This study assessed the validity of using EMR prescription data to produce antibiotic prescription rates, accounting for these factors. Methods: The study was a secondary analysis of antimicrobial prescriptions collected from 4 primary care clinics from 2015 to 2017. For adults with selected respiratory and urinary infections, EMR diagnostic codes, prescription data, clinical diagnoses and demographics were abstracted. Overall and delayed prescription rates were produced for EMR diagnostic codes, clinical diagnoses, by clinic, and types of infection. Direct standardization was used to adjust for case mix differences by clinic. High antibiotic prescribers, above the 75th percentile for prescriptions, were compared with low antibiotic prescribers. Results: Of 3108 EMR visits, there were 2577 (85.4%) eligible visits with a clinical diagnosis and prescription information. Overall antibiotic prescription rates were similar utilizing EMR records (31.6%) or clinical diagnoses (32.6%, P = .40). When delayed prescriptions were removed, prescribing rates were lower (22.4%, P < .01). EMR data overestimated prescribing rates for conditions where antibiotics are usually not indicated (17.7% EMR vs 7.6% clinical diagnoses, P < .001). High antibiotic prescribers saw more cases where antibiotics are usually indicated (23.4%) compared to low prescribers (16.8%; P = .001). Conclusions: Electronic medical record prescribing rates are similar to those using clinical diagnoses overall, but overestimate prescribing by clinicians for conditions usually not needing antibiotics. EMR prescription rates do not account for delayed antibiotic prescriptions or differences in infection case-mix.