Assessing the conversion of electronic medical record data into antibiotic stewardship indicators

BACKGROUND: Measuring the appropriateness of antibiotic use is crucial for antibiotic stewardship (ABS) programmes to identify targets for interventions. OBJECTIVES: To assess the technical feasibility of converting electronic medical record (EMR) data into ABS indicators. METHODS: In this observational feasibility study covering a period of 2 years, the EMRs of patients hospitalized at a large non-university hospital network and receiving at least one dose of a systemic antibiotic were included. ABS indicators measuring steps in the process of antibiotic prescription proposed by the literature were collected and rephrased or defined more specifically to be calculable if needed. Algorithms were programmed in R to convert EMR data into ABS indicators. The indicators were visualized in an interactive dashboard and the plausibility of each output value was assessed. RESULTS: In total, data from 25 337 hospitalizations from 20 723 individual patients were analysed and visualized in an interactive dashboard. Algorithms could be programmed to compute 89% (25/28) of all pre-selected indicators assessing treatment decisions automatically out of EMR data, with good data quality for 46% (13/28) of these indicators. According to the data quality observed, the most important issues were (i) missing or meaningless information on indication (e.g. ‘mild infection’) and (ii) data processing issues such as insufficiently categorized metadata. CONCLUSIONS: The calculation of indicators assessing treatment decisions from EMRs was feasible. However, better data structure and processing within EMR systems are crucial for improving the validity of the results.