HSR&D Citation Abstract
Search | Search by Center | Search by Source | Keywords in Title
How Well Can We Identify the High-Performing Hospital?
Shwartz M, Cohen AB, Restuccia JB, Ren Z, Labonte A, Theokary C, Kang R, Horwitt J. How Well Can We Identify the High-Performing Hospital? Poster session presented at: AcademyHealth Annual Research Meeting; 2010 Jun 28; Boston, MA.
Research Objectives: To analyze the extent to which hospitals perform similarly across multiple performance measures.
Study Design: We examined five performance measures calculated from publicly-available data and two survey-based measures: risk-adjusted inpatient mortality for the AHRQ inpatient quality indicators; risk-adjusted efficiency; adherence to process of care measures for three conditions (acute myocardial infarction, heart failure, and pneumonia) reported on the CMS Hospital Compare web site; risk-adjusted readmission rates for the same conditions, patient satisfaction (as measured by the Hospital Consumer Assessment of Healthcare Providers and Systems, also reported on the CMS website), survey-assessed patient care quality by chief quality officers (CQOs), and survey-assessed patient care quality by a small sample of front-line clinicians. We examined the correlation among the measures and, as a composite measure, the sum of the quintiles when hospitals were ranked on the individual measures.
Population Studied: The initial sample consisted of 556 hospitals that responded to at least one of two complementary surveys (one for CQOs and the other for front-line clinicians) administered in 2006 to assess patient care quality and quality improvement activities. This convenience sample was augmented with a stratified sample (by region and bedsize) of 556 hospitals that did not respond to the survey. The final sample of 1006 hospitals contained 462 survey hospitals for which the CQO and at least three front-line clinicians responded and for which MEDPAR data were available, and 544 non-responding hospitals for which MEDPAR data were available.
Principal Findings: Among the five publicly-available measures, the highest correlation was between adherence to processes of care and patient satisfaction (0.32, p < .05). All other correlations were 0.10 or less. Across all seven measures, the highest correlation was between front-line clinician assessment of quality and patient satisfaction (0.48, p < .05). Other notable correlations were between patient satisfaction and CQO assessment of quality (0.23, p < .05); CQO and front-line clinician assessments of quality (0.25, p < .05); and adherence to processes of care and CQO and front-line clinician assessments of quality (0.18 and 0.22, respectively, both p < .05).
There were 577 hospitals for which data were available on all five publicly-available measures. The highest ranked hospital on the composite score was in the top quintile on four measures and in the second quintile on one. Only 11 other hospitals achieved a sum of quintiles of eight or less, suggesting that very few hospitals are high performers on all measures. For about 10% of hospitals, the sum of quintiles was 10 or less (meaning they were in the second quintile, on average, across all five measures).
Conclusions: We found little correlation among performance measures that can be calculated from publicly-available data. Nevertheless, since formative composite scales do not require correlation among scale components, it is reasonable to calculate a composite measure of hospital performance.
Implications for Policy, Delivery or Practice: Since hospitals that score well on a composite measure are not likely to be top performers on many components that comprise the composite, pay-for-performance programs and benchmarking efforts to identify high performers should consider performance based both on composite measures and on the individual measures included in the composite.