Cleveland评分对我国成人心脏瓣膜术后急性肾损伤的预测价值

The predictive value of cleveland clinical score for acute renal injury after cardiac valve surgery in Chinese adult patients

目的验证Cleveland评分对预测我国成人心脏瓣膜术后急性肾功能不全需行肾脏替代治疗（RRT—AKI）的准确性。方法收集2010年1月至2014年12月期间长海医院胸心外科收治的所有成人心脏瓣膜手术3350例患者的临床资料，排除不符合病例120例，纳入3230例。术前根据评分的风险分层标准进行分组：0-2分、3-5分、6-8分分别为1-3级，比较总体及各级RRT—AKI实际发生率和预测发生率。模型的预测价值评价，以分辨力（受试者工作特征曲线下面积，AUC—ROC）以及校准度（Hosmer—Lemeshow拟合优度检验）表示。同时根据是否发生RRT—AKI分为RRT组和非RRT组，比较两组住院期间死亡率。结果患者术后RRT—AKI实际发生率对比Cleveland评分预测值为1．67％对1．7％，差异无统计学意义（χ2=0．018，P=0．892）。其中风险等级1级、2级、3级发生RRT—AKI的实际发生率和预测发生率分别为1．23％对0．40％（χ2=4．96，P〈0．01）、2．66％对1．80％（χ2=3．83，P=0．049）、16．70％对9．50％（χ2=0．358，P=0．549）。AUC—ROC结果为0．64[95％CI（0．57，0．71）P〈0．01]，评分的分辨力和校准度均较低。非RRT组和RRT组住院期间死亡比例1．50％对87．00％（χ2=1330，P〈0．01），差异有统计学意义。结论Cleveland评分模型在预测术后RRT—AKI无明显分辨力，可以估计总体及部分3级风险患者术后RRT—AKI的发生率。同时提示随着评分等级的上升，患者术后RRT—AKI发生率显著上升，且一旦发生RRT—AKI患者住院期间死亡等不良预后的风险明显升高。

Objective To validate the value of Cleveland Clinical Score in predicting acute renal injury requiring renal replacement therapy（RRT-AKI） after cardiac valve surgery in Chinese adult patients. Methods An analysis was conducted for all the adult patients who underwent cardiac valve surgery from January 2010 to December 2014 in Changhai Hospital, Shanghai. A total of 3 230 adult patients were included. Based on Cleveland Clinical Score, the patients were divided into 3 risk stages： 0 to 2 point, 3 to 5 point, and 6 to 8 point. The incidence of RRT-AKI were compared between different stages. And the predictive value of the Cleveland Clinical Score model was assessed by area under the receiver operating characteristic curve （AUC-ROC） and the model calibration was assessed using the Hosmer-Lemeshow test. The patients were also divided into two groups： Non-RRT group and RRT-AKI group. The mortality were compared between these two groups. Results The inci- dence of RRT-AKI was 1.67% vs the predicted ratio of RRT-AKI 1.70% （χ2 = 0. 018, P = 0. 892）. Among the stage 1, 2, and 3, the actual incidence of RRT-AKI, was 1.23%, 2.66%, and 16.7% vs the predicted incidence 0.40%, 1.80%, and 9.50%, respectively. The AUC-ROC for Cleveland Clinical Score predicting RRT-AKI was 0.64[95% CI（0. 57,0.71 ） , P 〈 0.01）. Compared with Non-RRT group, the RRT-AKI group got a higher mortality（87.00% vs 1.50%, χ2 = 1 330, P 〈 0.01 ）. Conclusion The Cleveland Clinical score had no real predictive value for RRT-AKI in Chinese adult patients after car- diac valve surgery. The incidence of RRT-AKI of the whole population and the stage 3 patients could be predicted by the model. And the patients with a high Cleveland score got a higher mortality than that of patients with a low Cleveland score.