合并代谢综合征的结直肠癌患者术后腹腔感染风险的预测模型构建与验证

Construction and validation of a predictive model for the risk of postoperative abdominal infection in colorectal cancer patients with combined metabolic syndrome

目的:研究代谢综合征(metabolic syndrome,MetS)对结直肠癌(colorectal cancer,CRC)术后腹腔感染的影响,建立模型预测CRC术后腹腔感染风险,并验证模型的有效性。方法:收集我院胃肠外科354例接受结直肠癌根治术患者的临床资料,采用单因素及多因素分析探索CRC患者术后腹腔感染的影响因素,找寻独立危险因素构建预测模型,绘制森林图模型预测风险,并通过内部验证检验其预测性能。结果:纳入的354例CRC患者中,共有68例合并MetS(19.21%),相较非MetS组,MetS组术后腹腔感染发生率更高(P<0.01),单因素Logistic分析表明,年龄≥65岁、合并MetS、术前合并肠梗阻、肿瘤分化程度、高血糖、高血压、高血脂对CRC术后腹腔感染的发生有影响;多因素Logistic分析表明,术前合并肠梗阻[OR:5.729(2.55~12.871),P<0.01]、MetS[OR:5.324(2.825~10.034),P<0.01]、高血糖[OR:2.187(1.204~3.975),P=0.01]是术后腹腔感染的独立危险因素,并构建预测模型预测及量化CRC术后腹腔感染的发生风险(C指数为0.771);利用ROC曲线计算出其95%CI为(0.699~0.842),该预测模型最佳阈值切点为0.172,特异度为0.796,灵敏度为0.636,显示该模型具有较好的区分度,再通过绘制Calibrate校准曲线,得到的模拟曲线及实际曲线的绝对误差为0.03,说明两曲线走势轨迹基本一致。DCA决策曲线显示该模型具有较好的临床有效性和一定的临床价值。结论:本研究构建的预测模型对CRC术后腹腔感染的发生具有良好的预测效果,对于临床工作中的高危患者有指导及警示作用。

Objective:To study the effect of metabolic syndrome(MetS)on postoperative abdominal infections in colorectal cancer(CRC),to establish a model to predict the risk of postoperative abdominal infections in CRC,and to validate the effectiveness of the model.Methods:The clinical data of 354 patients who underwent radical surgery for colorectal cancer were collected from the Department of Gastrointestinal Surgery of our hospital,and the influencing factors of postoperative abdominal infections in CRC patients were explored by using univariate and multivariate analyses to find the independent risk factors to construct the prediction model,to predict the risk by drawing the forest plot model,and to test the prediction performance by internal validation.Results:Among the 354 CRC patients included,a total of 68 cases had combined MetS(19.21%),and the incidence of postoperative abdominal infection was higher in the MetS group compared with the non-MetS group(P<0.01).One-way Logistic analysis showed that age≥65 years,combined MetS,preoperative combined intestinal obstruction,tumor differentiation,hyperglycemia,hypertension,and hyperlipidemia had an effect on the occurrence of abdominal infection after colon cancer surgery.Multifactorial Logistic analysis showed that preoperative combined intestinal obstruction[OR:5.729(2.55~12.871),P<0.01],MetS[OR:5.324(2.825~10.034),P<0.01],and hyperglycemia[OR:2.187(1.204~3.975),P=0.01]were the independent risk factors for postoperative abdominal infection,and a prediction model was constructed to predict and quantify the risk of postoperative abdominal infection after CRC(C-index was 0.771).Using the ROC curve,the 95%CI was calculated as(0.699~0.842),and the optimal threshold cut-off point of the prediction model was 0.172,the specificity was 0.796,and the sensitivity was 0.636,which showed that the model had a better differentiation,and then by drawing the Calibrate calibration curve,the absolute error of the simulated curve and the actual curve was 0.03,which indicated that the trajectories of the two curves were basically the same.The DCA decision curve shows that the model has good clinical effectiveness and certain clinical value.Conclusion:The prediction model constructed in this study has a good predictive effect on the occurrence of postoperative abdominal infections after CRC,which is a guide and warning for high-risk patients in clinical work.