Evaluating the Fitness for Purpose of Primary Care Data from EHRs for Automated Antimicrobial Prescribing Audits

OBJECTIVE To determine whether primary care EHR data are sufficiently complete and plausible to support automated audits of antimicrobial prescribing quality. METHODS Cross-sectional descriptive assessment of antimicrobial auditing-related fields in Patron, a large Australian primary care EHR dataset with 3.5 million patients from 129 consenting general practices. Data from 2018 to 2022 were evaluated using the Harmonized Data Quality Assessment Terminology and Framework, covering conformance, completeness, and plausibility. RESULTS Thirty-one fields (137,776,804 rows; 1,406,364 patients across 116 practices) were assessed. Value conformance and plausibility were high for most core audit variables, including demographics, antimicrobial name, dose, allergy status, and visit date. Prescribing indication was incompletely captured (13-27% completeness), and allergy severity was recorded in 26% of allergy entries. Vendor-level heterogeneity contributed substantially to variation in field completeness. CONCLUSION Australian primary care EHR data capture the core structured elements required for automated antimicrobial prescribing audits, enabling assessments of spectrum suitability, microbiology mismatch, and prescribing prevalence. Incomplete and inconsistent documentation of indication and allergy severity necessitates the use of proxy fields or inference for more complex evaluations. Greater standardization across EHR systems is required to enhance the scalability and clinical utility of automated audits in primary care.