论文部分内容阅读
目的初步建立心脏手术后急性肾损伤(AKI)患者发病的预测模型,并进行验证。方法连续收集2010年5月—2011年1月于复旦大学附属中山医院心外科行冠状动脉旁路移植术(CABG)、非体外循环冠状动脉旁路移植术(OPCAB)、瓣膜手术或瓣膜手术联合CABG的1 394例患者的临床资料。采用2012年KDIGO(kidney disease:improving global outcomes)指南的AKI诊断标准,其中1 067例患者作为模型建立组,对心脏手术后AKI的发病率和危险因素进行统计分析,建立数学预测模型;余300例患者为模型验证组,通过分辨力(ROC曲线的AUC)和校准度(Hosmer-lemeshow拟合优度检验)验证模型的预测价值。结果模型建立组手术后AKI发病率为20.3%(217/1 067),肾脏替代治疗(RRT)治疗率为3.5%(38/1 067),病死率为3.3%(35/1 067)。AKI患者的RRT治疗率为17.5%(38/217),病死率为13.8%(30/217)。单因素分析结果显示,心脏手术后AKI发病危险因素包括术前因素(男性、年龄、糖尿病、需要治疗的糖尿病、高血压、冠状动脉性心脏病、脑血管病、慢性阻塞性肺疾病、术前7d内行冠状动脉造影检查、纽约心脏病学会心功能分级Ⅲ级及以上、左心室射血分数<0.5、术前血清肌酐水平、术前肾小球滤过率估算值、术前尿液检查异常)、术中因素(CPB时间、瓣膜手术联合CABG、术中输红细胞悬液>400mL、术中输血浆>400mL、术中净入量>1 200mL)和术后因素[术后发生低心排综合征、术后中心静脉压>14cmH2O(1cmH2O=0.098kPa)]。Logistic逐步回归分析显示,男性(OR=1.645,95%CI为1.142~2.369)、有脑血管病史(OR=5.534,95%CI为3.061~10.006)、瓣膜手术联合CABG(OR=3.511,95%CI为0.827~14.906)、术前血清肌酐水平>115μmol/L(OR=2.325,95%CI为1.351~4.001)、术前7d内行冠状动脉造影检查(OR=1.609,95%CI为1.147~2.257)、术前纽约心脏病学会心功能分级Ⅲ级及以上(OR=1.509,95%CI为1.040~2.188)、术前输血浆>400 mL(OR=2.466,95%CI为1.290~4.715)、术后发生低心排综合征(OR=3.602,95%CI为2.242~5.787)、术后中心静脉压>14cmH2O(OR=2.156,95%CI为1.001~4.644)为心脏手术后AKI发病独立危险因素(P值分别<0.01、0.05)。术后AKI预测发生率=ex/(1+ex),x=0.498×男性+1.256×瓣膜手术联合CABG+1.711×既往脑血管病+0.475×术前7d内行冠状动脉造影检查+0.411×纽约心脏病学会心功能分级Ⅲ级及以上+1.281×术后发生低心排综合征+0.844×血清肌酐水平>115μmol/L+0.902×术中输血浆>400mL+0.768×中心静脉压>14cmH2O-2.752。验证组对模型的分辨力(AUC为0.810±0.028,95%CI为0.757~0.868)和校准度(χ2=10.13,P=0.256)均较好。结论本研究建立的数学预测模型在预测心脏手术后AKI方面具有较好的时效性和预测价值,对临床工作具有较好的指导作用。
Objective To establish a predictive model of acute kidney injury (AKI) patients after cardiac surgery. Methods Continuous coronary artery bypass grafting (CABG), non-cardiopulmonary bypass coronary artery bypass grafting (OPCAB), valvular surgery or valvular surgery were performed in the Department of Cardiac Surgery, Zhongshan Hospital Affiliated to Fudan University from May 2010 to January 2011. Clinical data of 1 394 patients with CABG. According to the AKI diagnostic criteria of KDIGO guideline (KDIGO 2012), 1,067 patients were established as models to statistically analyze the incidence and risk factors of AKI after cardiac surgery and establish a mathematical prediction model; more than 300 The patients were model validation group and the predictive value of the model was verified by resolution (ROC curve AUC) and calibration (Hosmer-lemeshow goodness-of-fit test). Results The incidence of AKI after operation was 20.3% (217/1 067) in the model group, 3.5% (38/1 067) in the RRT group and 3.3% (35/1 067) in the model group. The RRT rate for AKI patients was 17.5% (38/217) and the case fatality rate was 13.8% (30/217). Univariate analysis showed that the risk factors for AKI after cardiac surgery included preoperative factors (male, age, diabetes, diabetes requiring treatment, hypertension, coronary heart disease, cerebrovascular disease, chronic obstructive pulmonary disease, preoperative 7d underwent coronary angiography, the New York Heart Association Society of Cardiac Function grade Ⅲ and above, left ventricular ejection fraction <0.5, preoperative serum creatinine, preoperative glomerular filtration rate estimates, preoperative urinalysis abnormalities ), Intraoperative factors (CPB time, valvular surgery combined with CABG, intraoperative transfusion of red blood cell suspension> 400mL, intraoperative blood transfusion> 400mL, intraoperative net amount> 1 200mL) and postoperative factors [postoperative low Cardiovelectic syndrome, postoperative central venous pressure> 14cmH2O (1cmH2O = 0.098kPa) ]. Logistic stepwise regression analysis showed that patients with cerebrovascular disease had a history of cerebrovascular disease (OR = 5.534, 95% CI 3.061-10.006), valvular surgery combined with CABG (OR = 3.511, 95% CI = 0.827-14.906). Preoperative serum creatinine level was> 115 μmol / L (OR = 2.325, 95% CI 1.351-4.001). Coronary angiography was performed 7 days before operation (OR = 1.609, 95% CI 1.147-2.257 (OR = 1.509, 95% CI: 1.040-2.188), preoperative plasma infusion> 400 mL (OR = 2.466, 95% CI: 1.290-4.715), preoperative heart rate Low cardiac output syndrome (OR = 3.602, 95% CI 2.242-5.787) and postoperative central venous pressure> 14 cm H2O (OR = 2.156, 95% CI 1.001-4644) were independent risk factors for AKI after cardiac surgery Factors (P values <0.01, 0.05, respectively). Postoperative AKI predicted incidence = ex / (1 + ex), x = 0.498 × male + 1.256 × valve surgery combined with CABG + 1.711 × previous cerebrovascular disease + 0.475 × preoperative 7d undergoing coronary angiography + 0.411 × New York Heart Cardiac dysfunction grade Ⅲ and above + 1.281 × Postoperative low cardiac output syndrome + 0.844 × serum creatinine level> 115μmol / L + 0.902 × intraoperative blood transfusion> 400mL + 0.768 × central venous pressure> 14cmH2O-2.752. The resolution of the model group (AUC 0.810 ± 0.028, 95% CI 0.757 ~ 0.868) and calibration (χ2 = 10.13, P = 0.256) were all better. Conclusions The mathematical prediction model established in this study has good timeliness and predictive value in predicting AKI after cardiac surgery, which has a good guiding role in clinical work.