VRE acquisition in hospital and its association with hospital antimicrobial usage -a non-linear analysis of an extended time series.

Abstract

Vancomycin resistant enterococci (VRE) have become endemic pathogens in many Australian hospitals causing significant morbidity. There are few observational studies that have evaluated the effect of antibiotic usage on VRE acquisition. This study examined VRE acquisition and its association with antimicrobial use. The setting was a NSW tertiary hospital with 800 beds over a 63 month period up to March 2020, straddling piperacillin-tazobactam (PT) shortages that occurred from in September 2017. The primary outcome was monthly inpatient hospital onset Vancomycin-resistant Enterococci (VRE) acquisitions. Multivariate adaptive regression splines (MARS) were used to estimate hypothetical thresholds, where antimicrobial use above threshold is associated with increased incidence of hospital onset VRE acquisition. Specific antimicrobials and categorised usage (broad, less broad and narrow spectrum) were modelled. There were 846 hospital onset VRE detections over the study period. Hospital onset vanB and vanA VRE acquisitions fell significantly by 64% and 36% respectively after the PT shortage. MARS modelling indicated that PT usage was the only antibiotic found to exhibit a meaningful threshold. PT usage greater than 17.4 defined daily doses/1000 occupied bed-days (95%C I: 13.4, 20.5) was associated with higher onset of hospital VRE. This paper highlights the large, sustained impact that reduced broad spectrum antimicrobial use had on VRE acquisition and showed that PT use in particular was a major driver with a relatively low threshold. It raises the question as to whether hospitals should be determining local antimicrobial usage targets based on direct evidence from local data analysed with non-linear methods.