The NHSN 2015 Rebaseline—How Updating CAUTI Risk Adjustment Affects the SIRs

BACKGROUND: In January 2017, the CDC’s National Healthcare Safety Network (NHSN) updated both the aggregate healthcare associated infection (HAI) data that serve as baselines for standardized infection ratios (SIRs) and the risk adjustment (RA) methods used to calculate SIRs. The new baseline data are 2015 HAI incidence data; the new risk adjustment (RA2) models are predictive models developed using new baseline data. The previous RA1 method for catheter-associated urinary tract infections (CAUTI) used location- and facility-stratified rates to predict CAUTI incidence. The 2015 rebaseline used location- and facility-level factors in log-linear models to determine the expected number of infections. The objective of this study was to compare RA2 vs. RA1 SIRs from acute care hospitals (ACH). METHODS: We analyzed 1 year of CAUTI-ACH data reported to NHSN (2015). We compared differences between SIRs under RA2 vs RA1 at the national- and facility- levels. ACHs with <1 predicted infections were excluded. We used paired T and Empirical Distribution Function (EDF) tests to compare differences between means and medians of the SIR distributions respectively. Using quintiles, we compared shifts in facility-level SIRs between RAs (RA1 - referent group). RESULTS: 4318 ACHs reported CAUTI data to NHSN in 2015. 2917 (67%) vs 2468 (57%) ACHs had ≥1 expected infections and SIRs calculated using RA1 and RA2 respectively; 2466 (57%) ACHs had SIRs calculated using either RA method. The 2015 national pooled mean RA1 SIR was 0.580 (95% CI: 0.573, 0.588), and 0.983 (95% CI: 0.970, 0.996) under RA2. The means and medians of the two SIR distributions were significantly different (P < 0.0001). At the facility level, 1992 (81%) ACHs had SIRs that did not change between RA1 and RA2, 238 (10%) had SIRS that improved (lower in RA2 vs RA1), while 235 (10%) had SIRs increased (higher in RA2 vs RA1). CONCLUSION: There was a marked increase in the SIR with the introduction of the new baseline towards 1.00 (the national benchmark). This was attributed to the new RA2 models which control for many factors that are known to influence the number of HAI infections, unlike RA1. Comparable proportions of facilities had SIRS improve or worsen between RA1 and RA2 indicating that shifts in facility level SIRs were not large. DISCLOSURES: All authors: No reported disclosures.