Key points are not available for this paper at this time.
BACKGROUND: It is unclear whether publicly reporting hospitals' risk-adjusted mortality affects market share and mortality at outlier hospitals. OBJECTIVES: To examine hospitals' market share and risk-adjusted mortality from 1991 to 1997 at hospitals participating in Cleveland Health Quality Choice (CHQC). RESEARCH DESIGN: Time series. SUBJECTS: Changes in market share were examined for all patients hospitalized with acute myocardial infarction, heart failure, gastrointestinal hemorrhage, obstructive pulmonary disease, pneumonia, or stroke at all 30 nonfederal hospitals in Northeast Ohio. Patients insured by Medicare were used to examine changes in mortality. MEASURES: Trends in market share (proportion of patients with the target conditions discharged from a given hospital) and risk-adjusted 30-day mortality. RESULTS: CHQC identified several hospitals with consistently higher than expected mortality. The five hospitals with the highest mortality tended to lose market share (mean change -0.6%, 95% CI -1.9-0.6), but this was not significant. The only outlier hospital with a large decline in market share had declining volume for 2 years before being declared an outlier. Risk-adjusted mortality declined only slightly at hospitals classified by us as "below average" (-0.8%; 95% CI, 2.9-1.8%) or "worst" (-0.4%; 95% CI -2.3-1.7). However, risk-adjusted mortality at one hospital changed from consistently above expected to consistently below expected shortly after first being declared an outlier. CONCLUSION: Despite CHQC's strengths, identifying hospitals with higher than expected mortality did not adversely affect their market share or, with one exception, lead to improved outcomes. This failure may have resulted from consumer disinterest or difficulty interpreting CHQC reports, unwillingness of businesses to create incentives targeted to hospitals' performance, and hospitals' inability to develop effective quality improvement programs.
Baker et al. (Sun,) studied this question.