Factors Associated With Inter-Hospital Variability of Inpatient Antibiotic Use in a Cohort of US Hospitals

Abstract

Background. The National Action Plan for Combating Antibiotic Resistance Bacteria calls for standardized methods to monitor and benchmark inpatient antibiotic usage. Inter-facility comparisons will require risk adjustment to account for differences in patient mix and hospital characteristics. We examined facility level factors associated with inter-hospital variability in a large cohort of US hospitals. Methods. We used data from the Truven Health MarketScan Hospital Drug Database, which contains detailed administrative records, including inpatient drug utilization data based on billing records, for all patients discharged from a convenience sample of over 500 US hospitals. We retrospectively estimated days of therapy (DOT)/1000 patient days (PDs) by year, for hospitals reporting data between 2006 and 2012. We also created a multivariable model that adjusts for hospital-specific location of antibiotic use (ICU versus other) and hospital-specific annual summary measures for the following: average patient age, average patient co-morbidity score, number of hospital beds, teaching status, urban or rural location, proportion of discharges with a surgical diagnosis related code, case mix index, and proportion of patient days with an infectious disease primary ICD-9-CM discharge code. Results. The mean DOT for all hospitals across all years was 777/1000 PDs (SD = 189). DOT varied significantly between hospitals; the 10th to 90th percentile values for hospital DOT ranged from 546 to 998/1000 PDs. The variables included in our model accounted for 47%-53% of the inter-facility variability, depending on year, almost all of which (46%–51%) was explained by 2 predictors: proportion of PDs with an infectious disease diagnosis code and hospital location (ICU versus other). Conclusion. Inter-facility comparisons of antimicrobial use may require adjusting for the proportion of PDs with an infectious disease primary diagnosis code and the hospital location of antibiotic use. Because a large proportion of inter-hospital variability in antimicrobial use remains unexplained, we hypothesize that the residual variability may be the result of variations in prescribing behavior that could potentially be addressed through antibiotic stewardship programs. Disclosures. All authors: No reported disclosures.