摘要
目的评价7种危险评分系统对于单中心接受冠状动脉旁路移植术（CABG）患者术后病死率的预测效能。方法收集2010年1月至2011年1月施行CABG患者围手术期资料。应用7种危险评分系统：SinoSCORE、AdditiveEuroSCORE、LogisticEuroSCORE、OPR、Clevelandmodel、Parsonnetseore、QMMI预测每一位患者术后病死率和全部患者的平均预计病死率。根据预计病死率将全部患者分为6组：I组（0—1．99％），Ⅱ组（2．00％～3．99％），1I组（4．00％～5．99％），IV组（6．00％-7．99％），V组（8．00％～9．99％），VI组（〉10％）。比较不同危险程度的实际病死率和预计病死率评价评分系统的预测能力。运用Hosmer—Lemeshow拟合优度检验评价评分系统的校准度，运用ROC曲线下面积（AUC）评价评分系统的分辨力。结果全组共1103例患者，平均年龄（62．8±8．8）岁。患者术后22例死亡，实际病死率1．99％。SinoSCORE、AdditiveEuroSCORE、LogisticEuroSCORE、OPR、Clevelandmodel、Parsonnetscore、QMMI预测平均全组病死率分别为3．01％、4．38％、3．83％、1．69％、4．42％、6．71％、3．71％，其中最接近实际病死率的是OPR、SinoSCORE、QMMI。分组比较显示：LogisticEuroSCORE在各组中全部高估术后病死率。AdditiveEuroSCORE在Ⅵ组中预测病死率明显低于实际病死率，其他各组均高估了术后病死率。而SinoSCORE、Clevelandmodel、Parsonnetscore、QMMI除了在I组低估了患者术后病死率，其他各组高估了术后病死率。OPR低估了I组和Ⅳ组患者的病死率，高估了其他组患者的病死率。利用Hos—mer—Lemeshow拟合优度检验评价7种评分系统的校准度，结果显示7种评分系统校准度尚可，P值全部〉0．05。通过ROC检验比较7种评分系统的分辨力，其中只有SinoSCORE的AUC=0．751（〉0．70），证明SinoSCORE对于本组患者的死亡分辨力良好。结论通过比较，SinoSCORE对于本中心接受CABG患者术后病死率预测效能好，可以运用于术前危险性评估。
Objective To access the predictive value in Postoperative mortality in patients undergoing coronary artery bypass grafting（CABG） by seven different Risk scoring system. Methods To collect the clinical information of patients under- going CABG in our department. SinoSCORE, Additive EuroSCORE, Logistic EuroSCORE, OPR, Cleveland model, Parsonnet score, QMMI was used to predict postoperative mortality for all patients, and calculate the mean predictive postoperative mortal- ity. To devided the patients to six group ： group I （ 0 - 1.99% ）, group II （2.00% - 3.99 % ）, group III （4.00% - 5.99% ）, group IV （ 6.00% - 7.99% ）, group V （ 8.00% - 9.99% ）, group VI （ 〉 10% ） by predictive postoperative mortality. Access the performance of risk scoring system predict the mortality through compare the predictive mortality and the observed mortality in different Risk stratification. To use Hosmer-Lemeshow goodness-of-fit test access the calibration. Discrimination was tested by determining the area under the receiver operating characteristic （ROC） curve （AUC）. Results The overall mean baseline age was （62.8 ±8.8 ）years. The observed mortality in all our study patients was 1.99% （22/1103 cases）. The predictive mortality calculated by SinoSCORE, Additive EuroSCORE, Logistic EuroSCORE, OPR, Cleveland model, Parsonnet score and QMMI were 3.01%, 4.38%, 3.83%, 1.69%, 4.42%, 6.71%and3.71%. And the most closest is OPR, SinoSCORE, QMMI. Group tests confirmed Logistic EuroSCORE Overestimated the mortality in all the group. Predicted mortality calculated by Additive EuroSCORE was lower than the actual mortality in group VI and higher than the observed mortality in other group. SinoSCORE ,Cleveland model,Parsonnet score,QMMI Overestimated the mortality in all the group expect group I ~ OPR fore- cast a lower mortality Compared with observed mortality in group I and group 1V and a Slightly higher mortality in group I1 , group HI. To use Hosmer - Lemeshow goodness-of-fit test access the calibration. The text proved all the risk scoring system had a good calibration for postoperative mortality （ P 〉 0.05 ）. Discrimination was tested by ROC, only SinoSCORE （ AUC -- 0.751 〉 O, 70） showed high discriminatory ability in predicting mortality. Conclusion SinoSCOBE have a good forecast performance for the postoperative mortality in the patients undergoing CABG in our department in seven different Risk scoring system. And SinoSCORE could be used in preoperative risk assessment.
出处
《中华胸心血管外科杂志》
CSCD
北大核心
2012年第3期152-156,160,共6页
Chinese Journal of Thoracic and Cardiovascular Surgery
