The dynamic relationship between antibiotic use and the incidence of vancomycin-resistant Enterococcus: time-series modelling of 7-year surveillance data in a tertiary-care hospital.

The role of antibiotics in the epidemiology of vancomycin-resistant Enterococcus (VRE) has been studied extensively, but controversies remain as to which, and to what extent, antibiotics facilitate the emergence and dissemination of VRE in hospitals. Aggregate data on the use of several antibiotic classes in terms of defined daily doses (DDD) per 100 patient-days (PD), and VRE incidence rates in terms of clinical isolates per 1000 PD, were evaluated during a 7-year period at a tertiary-care hospital. Time-series analysis (autoregressive integrated moving average (ARIMA) and transfer function models) was used to quantify the temporal effect of antibiotic use on VRE incidence and estimate effect-delays. The incidence rate of VRE observed in a specific bimester was found to be a function of its value during the preceding bimester and of prior changes in the volume of use of four antibiotic classes. In particular, an increase of one DDD/100 PD in the use of glycopeptides, fluoroquinolones, extended-spectrum cephalosporins and beta-lactam-beta-lactamase inhibitor combinations resulted, independently, in average changes of +0.024, +0.015, + 0.020 and -0.010 isolates per 1000 PD in the incidence of VRE, with average delays of 2, 4, 2 and 6 months, respectively, which explained 56% of the observed variation in VRE rates over time. Efforts to reduce VRE cross-transmission should be supplemented by targeted antibiotic control policies. The use of glycopeptides, broad-spectrum cephalosporins and fluoroquinolones in high amounts should be the targets of such policies. Penicillin-beta-lactamase inhibitor combinations might be suitable substitutes for extended-spectrum cephalosporins.