Risk stratification models fail to predict hospital costs of cardiac surgery patients

Background The aim of this prospective study was to determine if commonly used risk stratification models can predict total hospital costs in cardiac surgical patients. Methods Between October 1st and December 31st 2003, all consecutive adult patients undergoing cardiac surgery on CPB at our institution were classified using seven risk stratification scoring systems: EuroSCORE, Cleveland, Parsonnet, Ontario, French, Pons, and CABDEAL. Total hospital costs for each patient were calculated on a daily basis including preoperative diagnostic tests, operating room costs, disposable materials, drugs, blood components, costs for personnel, and hospital fixed-costs. Linear regression analysis was used to determine the correlation between costs and the seven risk stratifications models as well as length of stay (LOS) on ICU. The Spearman correlation coefficient was calculated from the regression line, and an analysis of residuals was performed to determine the quality of the regression. Results A total of 252 patients were operated for CABG (n = 175), valve (n = 39), CABG plus valve (n = 2 1), thoracic aorta (n = 13) and miscellaneous (2 myxoma, 1 ASD, 1 pulmonary embolism). Mean age of the patients was 66.0 +/- 11.4 years, 29.4% were female. LOS on ICU was 3.3 +/- 6.3 days and the 30-day mortality rate was 6.7%. Spearman correlation between the seven risk stratification models and hospital costs was below r = 0.32 (p = 0.0001), but was r = 0.94 (p = 0.0001) between ICU LOS and costs. Conclusions Total hospital costs can be identified by length of ICU stay. None of the common risk stratification models accurately predicted total hospital costs in cardiac surgical patients.