A scoring system for the assessment of post-CABG death risk / 中华胸心血管外科杂志

ABSTRACT

Objective Patients in the intensive care unit (ICU) are at high risk for multi-organ failure and death. Few well-established scoring systems have been used in the cardiac surgery. We try to identify a feasible score system for the risk assessment after CABG. Methods 1028 consecutive patients who had received CABG and staying in a single cardiac postoperative intensive care unit of Anzhen hospital were enrolled in the prospective study between October of 2007 and May of 2008 and assessed daily with three score systems, the multiple organ dysfunction score (MODS), the sequential organ failure assessment (SOFA) and our postoperative score for cardiac surgery (PSCS). Four new variances were added to PSCS system, which were Euroscore, IABP, ECMO and dialysis. Euroscore was considered as a preoperative factor which provided the preoperative information of the patient. IABP and ECMO were considered as assistant factors to support circulatory system. The dialysis was also considered as an assistant factor to support renal system. The differences with the respects of sensitivity and specificity among the three systems were compared with Hosmer-Lemeshow goodness-of-fit and receiver operating characteristic (ROC)curve. Results The new variances added to PSCS system were shown to be associated with mortality in a multivariate logistic regression analysis. The EXP(B) value for Euroscore was 3. 803, the EXP(B) value for IABP was 1. 645, the EXP(B) value for ECMO was 3.803, and the EXP(B) value for dialysis was 1.605. In discrimination analysis, ROC values of the operative day score were 0.602 for MODS, 0.571 for SOFA, and 0. 821 for PSCS; ROC values of the maximum score were 0.847 for MODS, 0.830 for SOFA, and 0.929 for PSCS; ROC values of the maximum score during the first 3 days were 0.838 for MODS, 0.814 for SOFA, and 0.919 for PSCS; score changes of ROC value between the third day and the first day were 0.767 for MODS, 0.779 for SOFA, and 0. 780 for PSCS. In calibration analysis, we compared the x2 values, P values and overall corrected percentage of the three different systems. x2 values of the three systems were 6. 763 for MODS, 4. 101 for SOFA, and 1.687 for PSCS; P values of the three systems were 0.454 for MODS, 0. 848 for SOFA, and 0. 975 for PSCS; overall corrected percentages of the three systems were 98.1%, 98%, and 98.3% respectively. Conclusion The sensitivity and specificity of PSCS were superior to MODS and SOFA in predicting death risk after CABG.