Monitoring outpatient antibiotic utilization using reimbursement and retail sales data: a population-based comparison in France, 2012-17.

ABSTRACT

OBJECTIVES: To assess whether a retail sales database could be used to monitor antibiotic utilization in the outpatient setting at the national level. METHODS: We extracted 2012-17 outpatient antibiotic extrapolated retail sales (IQVIA's Xponent) and reimbursement data from the National Health Insurance (SNDS) in metropolitan France. We compared estimates of antibiotic use and consumption [number of antibiotic drug deliveries (DrID) and defined daily doses (DID) per 1000 inhabitants per day]. We relied on relative differences, Pearson's r statistics and time series using autoregressive integrated moving average (ARIMA) modelling to study (i) differences in point estimates, (ii) correlation, and (iii) consistency in time trends between Xponent and SNDS. The analysis was conducted overall and in subgroups (age groups, therapeutic classes, major antimicrobial agents and regions). RESULTS: We analysed approximately 377 million antibiotic drug deliveries, comprising nearly 3.4 billion DDDs. Overall, Xponent slightly overestimated SNDS point estimates with yearly relative differences of +3.5% for DrID and +3.3% for DID. Peaks in relative differences were observed for July and August months. Relative differences were <5% in most subgroups, except for fosfomycin and three French regions. Overall and across most subgroups, the correlation between Xponent and SNDS monthly aggregated estimates was almost perfect (r ≥ 0.992 for all subgroups, except for one region). ARIMA modelling showed high consistency between Xponent's and SDNS's DrID time series, but detected timepoints where the series significantly diverged. CONCLUSIONS: IQVIA's Xponent and SNDS data were highly consistent. Xponent database seems suitable for monitoring outpatient antibiotic utilization in France.