1231. Reliability-Adjusted Standardized Infection Ratios for Public Reporting
Session: Oral Abstract Session: Impact of National Policy on HAI Reporting and Prevention
Friday, October 19, 2012: 3:00 PM
Room: SDCC 32 AB

Background: CDC's National Healthcare Safety Network (NHSN) provides standardized infection ratios (SIRs) to participating hospitals to help promote healthcare quality. In 2011, the NHSN began reporting central line-associated bloodstream infection (CLABSI) SIRs from intensive care units (ICUs) to the Centers for Medicare and Medicaid Services (CMS) to evaluate hospital reporting and performance.  Efforts to account for differences in reliability of CLABSI SIRs due to exposure volume should improve performance measurement.

Methods: Hospitals report CLABSI incidence data to NHSN that are summarized monthly by ICU location.  Aggregated data allow CDC to determine the hospital and location factors that best account for differences in estimated national CLABSI incidence.  These data are used to estimate the number of predicted CLABSIs for a given location and reporting period.  An SIR summarizes CLABSI experience for a hospital and is defined to be the ratio of the observed to predicted CLABSI incidence where the later was estimated using a log-linear marginal model based on hospital and patient care location factors.  Hospital SIRs were reliability-adjusted (SIR') using a mixed log-linear model fitted through empirical Bayesian analysis that shrinks SIRs based on low exposure volume toward the mean.  Variability in SIR and SIR' distributions were assessed by common measures of dispersion.  Data were analyzed using SAS 9.2.

Results: In 2011, there were 62,637 months of data summarizing CLABSI incidence reported to NHSN from 3,326 hospitals. The distribution of hospital CLABSI SIRs ranged from 0 to 5.9 (excluding 22 extreme outliers) with 46% of facilities reporting an SIR of 0; Adjusted SIRs (SIR') ranged from 0.2 to 3.8 (Figure).  Dispersion measured by the coefficient of variation was reduced from 142.1 (SIR) to 51.0 (SIR') and the coefficient of quartile variation from 1.0 (95% confidence interval [CI]: 0.999-1.001) to 0.22 (95% CI: 0.231-0.234), respectively.   Among facilities reporting a 0 SIR the SIR' ranged from 0.2 to 0.9.

Conclusion: Reliability-adjusted SIRs obtained by shrinking the observed CLABSI incidence toward the expected incidence given exposure volume and relevant factors are more comparable and stable to better measure quality performance.

Jonathan Edwards, MStat, Dan Pollock, MD and Scott Fridkin, MD, FSHEA, Division of Healthcare Quality Promotion, Centers for Disease Control and Prevention, Atlanta, GA


J. Edwards, None

D. Pollock, None

S. Fridkin, None

Findings in the abstracts are embargoed until 12:01 a.m. PST, Oct. 17th with the exception of research findings presented at the IDWeek press conferences.