Variation of Inpatient Antibiotic Use in Acute Care Facilities of the Veterans Health Administration between 2005 and 2009

Background: Surveillance of antibiotic use is a key component of antibiotic stewardship but comprehensive data about inpatient antibiotic use in US hospitals are still limited.

Objective: Assess the quantity and variation of antibiotic use between facilities and over time in acute-care hospitals of the Veterans Health Administration (VA).

Methods: Barcode medication administration (BCMA) data for antibiotics administered between January 2005 and December 2009 were analyzed from all VA sites with an average of greater than 100 occupied patient days on acute-care wards per month. Only acute care and rehabilitation wards were included and the analysis was restricted to facilities with BCMA data available for the entire study period. Days of Therapy (DOT) were calculated for antibiotics belonging to class J01 of the WHO Anatomical Therapeutic Chemical (ATC) classification as well as oral metronidazole, oral vancomycin and rifampin. Rates were normalized per 1000 patient days (PD) and per one admission. To increase comparability, hospitals were stratified according to the 2009 VHA Facility Quality and Safety Report, which categorizes VA facilities into 3 levels of complexity according to patient population, clinical services, academic affiliation and administrative complexity.

Results: 110 of the 124 VA acute-care sites fulfilled the inclusion criteria. Median total antibiotic use over the study period was 721 DOT/1000PD (Interquartile range 623-811 DOT/1000PD) and 4.3 DOT/admission (IQR 3.9 – 5.2). There was a gradual increase in median overall antibiotic use over the study period (649 DOT/1000PD in 2005, 775 DOT/1000PD in 2009) with the biggest relative increases being observed for carbapenems (+102%), intravenous vancomycin (+79%) and penicillin/beta-lactamase-inhibitors (+41%). Overall 52% of patients received at least one dose of an antibiotic during their stay. Fluoroquinolones were the most commonly used antibiotic class and represented 19.3% of total use with a wide variation between hospitals (range 6.6-42.9%). Median antibiotic use was comparable across complexity levels and all four VA regions independent of the denominator used (PD or admissions).

Conclusions: The quantity of antibiotic use in VA hospitals expressed as DOT per 1000 PD is similar to reported data from non-VA hospitals in the US and seems to be increasing. There is considerable variation in antibiotic use across VA hospitals, which does not appear to be explained by case-mix.  This variation merits further investigation, as does the observed heavy use of fluoroquinolones due to their negative ecologic impact and association with a number of nosocomial pathogens. Tailored interventions targeting decisions to start and stop antibiotics, as well as antibiotic choice are likely needed to improve antibiotic use across the entire VA system.