EuroSCORE Ⅱ模型对瓣膜手术患者在院死亡风险预测的评价

Validation of EuroSCORE Ⅱ in predicating in-hospital mortality among Chinese patients undergoing heart valve surgery

目的评价第2版欧洲心脏手术危险评估系统(EuroSCOREⅡ)模型对本中心瓣膜手术患者在院死亡率预测的准确性。方法回顾性收集2006年1月至2011年12月因瓣膜疾病在本中心行外科治疗的3 479例患者的临床资料,按第1版的EuroSCORE(additive EuroSCORE、logistic EuroSCORE)模型和EuroSCOREⅡ模型模给予评分,并对患者的实际死亡率与预测死亡率进行对比。模型预测的符合程度应用H-L X^2检验,而预测的鉴别效度则通过R()C曲线下面积反映。结果 3479例患者在院死亡112例,全组实际在院死亡率为3.2%。Additive EuroSCORE、logistic EuroSCORE及EuroSCOREⅡ模型预测的在院死亡率分别为3.8%、3.3%和2.5%,其中logistic EuroSCORE对全组患者在院死亡的预测符合程度较高(P=0.08),而additive EuroSCORE高估了实际在院死亡率(P=0.013),EuroSCOREⅡ则低估了实际在院死亡率(P<0.000 1)。EuroSCOREⅡ模型对单瓣膜手术患者在院死亡预测具有较好的准确性(P=0.103,ROC曲线下面积为0.792),而对多瓣膜手术组患者的预测准确性则较差(P<0.000 1,ROC曲线下面积为0.605)。Additive EuroSCORE、logistic EuroSCORE及EuroSCOREⅡ模型对全组患者在院死亡预测的鉴别效度均较差(ROC曲线下面积分别为0.684、0.673和0.685)。结论 EuroSCOREⅡ模型对本中心单瓣膜手术患者死亡风险预测的准确性较好,但对多瓣膜手术患者死亡风险预测的准确性较差,不适合多瓣膜手术患者的在院死亡风险预测,在临床实践中应慎重考虑。

Objective To assess the performance of the European System for Cardiac Operative Risk Evaluation Ⅱ （EuroSCORE Ⅱ） in predicating in-hospital mortality among Chinese patients undergoing heart valve surgery at our center. Methods From January 2006 to December 2011, 3 479 consecutive patients who underwent heart valve surgery at our center were enrolled in this study and they were scored by the original EuroSCORE（addtive EuroSCORE and logistic EuroSCORE） and EuroSCORE Ⅱ model. The actual mortality rate of patients was compared with those of the predicted ones. The performances of the original EuroSCORE and EuroSCORE Ⅱ model were assessed by the Hosmer-Lemeshow （H-L） test. The discrimination validity of prediction was tested by calculating the area under the receiver operating characteristic （ROC） curve. Results There were 112 in-hospital deaths among the 3 479 patients, with an in-hospital mortality rate of 3.2% , compared to the predicted mortality rates of 3.84 % by the additive EuroSCORE （H-L： P= 0. 013, suggesting a higher prediction） , 3.33 % by the logistic EuroSCORE （H-L： P=0.08, suggesting good consistency）, and 2. 520%by the EuroSCORE Ⅱ （H-L： P〈0. 0001, suggesting a lower prediction）. EuroSCORE Ⅱshowed a good calibration in predicting in-hospital mortality for patients undergoing single valve surgery （H-L： P=0. 103, area under the ROC curve of 0. 792） and a poor calibration for patients undergoing multiple valve surgery （H-L： P〈0. 0001, area under the ROC curve of 0. 605）. The discriminative powers of the predictions by additive EuroSCORE, logistic EuroSCORE, and EuroSCORE Ⅱ were poor for the entire cohort, with the areas under the ROC curve being 0. 684, 0. 673, and 0. 685, respectively. Conclusion EuroSCORE Ⅱ has a better accuracy for predicting mortality of patients undergoing single valve surgery, but not for those undergoing multiple valve surgery, which should be considered in clinical practice.