Outcome prediction in cardiac surgery: the first logistic scoring model for cardiac surgical intensive care patients.

BACKGROUND: In the process of risk stratification, a logistic calculation of mortality risk in percentage is easier to interpret. Unfortunately, there is no reliable logistic model available for postoperative intensive care patients. The aim of this study was to present the first logistic model for postoperative mortality risk stratification in cardiac surgical intensive care units. This logistic version is based on our previously presented and established additive model (CASUS) that proved a very high reliability.
METHODS: In this prospective study, data from all adult patients admitted to our ICU after cardiac surgery over a period of three years (2007-2009) were collected. The Log-CASUS was developed by weighting the 10 variables of the additive CASUS and adding the number of postoperative day to the model. Risk of mortality is predicted with a logistic regression equation. Statistical performance of the two scores was assessed using calibration (observed/expected mortality ratio), discrimination (area under the receiver operating characteristic curve), and overall correct classification analyses. The outcome measure was ICU mortality.
RESULTS: A total of 4054 adult cardiac surgical patients was admitted to the ICU after cardiac surgery during the study period. The ICU mortality rate was 5.8%. The discriminatory power was very high for both additive (0.865-0.966) and logistic (0.874-0.963) models. The logistic model calibrated well from the first until the 13th postoperative day (0.997-1.002), but the additive model over- or underestimated mortality risk (0.626-1.193).
CONCLUSION: The logistic model shows statistical superiority. Because of the precise weighing the individual risk factors, it offers a reliable risk prediction. It is easier to interpret and to facilitate the integration of mortality risk stratification into the daily management more than the additive one.