Comparing hospitals that perform coronary artery bypass surgery: the effect of outcome measures and data sources.

OBJECTIVES: The relative quality of hospital care often is judged by comparing risk-adjusted rates of adverse outcomes. This study evaluated whether hospital quality comparisons are affected by the choice of outcome and the use of administrative data instead of clinical data.
METHODS: The data were collected from 2687 coronary artery bypass surgery patients from 17 hospitals. All patients were on Medicare. For 10 hospitals with 94 to 713 patients, risk-adjusted outcomes for death, major complications, and any complications were derived from a clinically rich database and an administrative database.
RESULTS: The correlations between adjusted hospital rankings derived from the clinical and administrative databases were not significant: .48 for mortality, .21 for major complications, and -.14 for any complication. When only the clinical database was used, the correlation between risk-adjusted hospital rankings for mortality and major complications was .77 (P < .01) and the correlation between major complications and any complication was -.45.
CONCLUSIONS: These results suggest assessing quality of care by the use of administrative data may not be adequate and that quality assessment by the use of clinical data may depend greatly on the outcome chosen.