1821. Understanding the Components and Calculation of the SAAR, Illustrative Data

Background The standardized antimicrobial administration ratio (SAAR) compares each hospital’s observed to predicted days of antimicrobial therapy. However, confusion exists about how hospital-level, seasonal, and hospital-peer-based variations in antibiotic use might impact an institution’s SAAR. We characterized the impact of each of these three types of variation on predicted SAARs utilizing local NHSN data. Methods Analysis of antibiotic consumption data from an academic medical center in Chicago, IL was conducted. SAAR and antimicrobial days per 1,000 days present (AD/1,000DP) were compiled in monthly increments from 2014 to 2016.Antimicrobial consumption was aggregated and classified into agent categories according to NHSN criteria. Month-to-month changes in both the SAAR and AD/1,000DP were evaluated. Azithromycin AD/1,000DP from 2012 through 2017 were explored for seasonal variation as defined as >20% increase in AD/1,000DP from each quarter to the overall mean AD/1,000DP for all months. A simulation was performed to explore the potential effect of seasonality on the SAAR. Demographic covariates within the SAAR model were altered while holding constant observed antibiotic use; thus we were able to observe the potential impact of demographics. Finally, a simulation explored the effect of altered consumption at other hospitals on a local institution’s SAAR. Results Across all antibiotic agent categories for both ICU (n = 4) and general wards (n = 4), the average matched-month percent change in AD/1,000DP was highly predicted and correlated with the corresponding change in SAAR (Figure 1, Pearson’s r = 0.99). The monthly mean ± SD AD/1,000DP was 235.0 (range 47.2–661.5), and the mean ± SD SAAR was 1.09 ± 0.26 (range 0.79–1.09) across the NHSN antibiotic agent categories. Five quarters were found to have seasonal variation in AD/1000DP for azithromycin (Figure 2). Simulations demonstrated that changing antimicrobial usage at comparator hospitals does not impact the local SAAR, and seasonal variation may cause fluctuating SAARs. Conclusion Month-to-month changes in the SAAR mirror monthly changes in an institution’s AD/1,000DP. Seasonal variation can impact the SAAR, and the effect changing peer hospital antibiotic consumption is not currently captured by the SAAR methodology. Disclosures J. Liu, Merck: Grant fund from Merck, Research grant. D. Martin, Syneos Health: Employee, Salary. GlaxoSmithKline: Independent Contractor, Salary. M. H. Scheetz, Merck & Co., Inc.: Grant Investigator, Grant recipient. Bayer: Consultant, Consulting fee.