UC Irvine

Background

States have established public reporting of hospital-associated (HA) infections-including those of methicillin-resistant Staphylococcus aureus (MRSA)-but do not account for hospital case mix or postdischarge events.

Objective

Identify facility-level characteristics associated with HA-MRSA infection admissions and create adjusted hospital rankings.

Methods

A retrospective cohort study of 2009-2010 California acute care hospitals. We defined HA-MRSA admissions as involving MRSA pneumonia or septicemia events arising during hospitalization or within 30 days after discharge. We used mandatory hospitalization and US Census data sets to generate hospital population characteristics by summarizing across admissions. Facility-level factors associated with hospitals' proportions of HA-MRSA infection admissions were identified using generalized linear models. Using state methodology, hospitals were categorized into 3 tiers of HA-MRSA infection prevention performance, using raw and adjusted values.

Results

Among 323 hospitals, a median of 16 HA-MRSA infections (range, 0-102) per 10,000 admissions was found. Hospitals serving a greater proportion of patients who had serious comorbidities, were from low-education zip codes, and were discharged to locations other than home were associated with higher HA-MRSA infection risk. Total concordance between all raw and adjusted hospital rankings was 0.45 (95% confidence interval, 0.40-0.51). Among 53 community hospitals in the poor-performance category, more than 20% moved into the average-performance category after adjustment. Similarly, among 71 hospitals in the superior-performance category, half moved into the average-performance category after adjustment.

Conclusions

When adjusting for nonmodifiable facility characteristics and case mix, hospital rankings based on HA-MRSA infections substantially changed. Quality indicators for hospitals require adequate adjustment for patient population characteristics for valid interhospital performance comparisons.