920-52 Are Provider Profiles Affected by Risk-adjustment Methodology? Results from the Cooperative Cardiovascular Project

Health care payors and consumers have a growing interest in risk-adjusted provider profiles. Using chart-abstracted clinical data from the Cooperative Cardiovascular Project, we ranked 28 hospitals performing bypass surgery in Alabama and Iowa by their risk-adjusted surgical mortality rates using three published risk-adjustment methodologies: Parsonnet (PI, O’Connor (a) and Hannan (H). In total. 3653 bypass surgery cases performed from 6/92 to 3/93 were reviewed (mean 130 cases/hospital). The discriminatory abilities of each method for predicting surgical mortality were quite similar (area under ROC curves 0.72–0.75). Below, we display the risk-adjusted hospital rankings (comparing observed with expected mortality) by these three riskadjustment techniques: Download : Download high-res image (98KB) Download : Download full-size image In terms of hospital rankings, there was generally close correlation between any two of the methods (Spearman's R = 0.87,0.88, and 0.93, comparing P-O, P-H, and H-O). Rankings for an individual hospital varied, however, an average of ± 3.3 ranks (range 0–12 ranks) depending on which riskadjustment methodology was used. Conclusion In general. published methods of risk-adjustment for bypass surgery accurately identify institutions with low, moderate and high adjusted mortality outcomes. The precise ranking of an individual hospital. however, may vary depending on the risk adjustment method applied.