Skip to main content
Open Access Publications from the University of California


UCLA Previously Published Works bannerUCLA

Initiation and termination of antibiotic regimens in Veterans Affairs hospitals



To assess rates of starting or stopping antibiotics across different hospitals.


We used barcode medication administration data to measure antibiotic use on acute-care wards in 128 Veterans Affairs medical centres (VAMCs) in 2010. A treatment day (TD) was defined as the administration of any antibiotic on a given day. A treatment period (TP) was defined as an interval of inpatient antimicrobial therapy with gaps of ≤1 day in TDs. The rate of starting antibiotics was calculated for inpatients who had not yet started antibiotics, as the number of start events divided by the 'person-time at risk'. The rate of stopping antibiotics was calculated analogously for inpatients that were on antibiotics. Once individuals had stopped antibiotics they were removed from further analysis. Per-day start and stop rates were also calculated for each day of hospitalization.


The hospital mean rate of starting the first TP was 18.1 start events/100 days at risk (range 8.4-25.6/100 days at risk). The mean hospital stopping rate was 21.1 stop events/100 days at risk (range 13.3-29.5/100 days at risk). The ratio of a facility's starting and stopping rates was highly correlated with overall antibiotic use in TDs/1000 patient-days (rs=0.92, P<0.001), while starting and stopping rates individually were only moderately correlated (rs=0.39, P<0.001).


VAMCs with similar antibiotic use showed marked differences in their starting and stopping rates of antibiotics. It may be useful to target empirical therapy when starting rates are high and definitive therapy when stopping rates are low.

Many UC-authored scholarly publications are freely available on this site because of the UC's open access policies. Let us know how this access is important for you.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View