先天性巨结肠根治术后小肠结肠炎风险预测

Enterocolitis after radical resection of congenital megacolon

目的构建并验证先天性巨结肠根治术后并发小肠结肠炎风险预测模型。方法选取2018年1月至2020年9月海南省妇女儿童医学中心200例术后先天性巨结肠患儿,采用数字随机分组,训练集140例与验证集60例。分析训练集术后小肠结肠炎发生风险,并构建预测模型。采用受试者工作特征(ROC)曲线、拟合优度(H-L)检验及校准曲线评估模型性能,绘制临床决策曲线评估模型临床适用性。结果小肠结肠炎组的痉挛肠段长度[10(26.32)、28(73.68)]、术前营养状况[16(42.11)、22(57.89)]、术前上呼吸道感染(URTI)病史[14(36.84)、24(63.16)]、术前肠炎[23(60.53)、15(39.47)]、术前洗肠时间[18(47.37)、20(52.63)]、术后肠梗阻[10(26.32)、28(73.68)]的平均值低于非小肠结肠炎痉挛肠段长度[55(53.92)、47(46.08)]、术前营养状况[15(14.71)、87(85.29)]、术前上呼吸道感染(URTI)病史[11(10.78)、97(89.22)]、术前肠炎[16(15.69)、86(84.31)]、术前洗肠时间[77(75.49)、25(24.51)]、术后肠梗阻[8(7.84)、94(92.16),P<0.05]。训练集ROC曲线下面积(AUC)[0.904,95%可信区间(CI)=0.854~0.954]高于验证集(0.889,95%CI=0.767~0.971,P<0.05),MAE(0.030)高于验证集(0.022)。H-L拟合优度检验结果显示,模型拟合情况良好(χ^(2)=5.210、3.777,P>0.05)。结论先天性巨结肠患儿根治术后并发小肠结肠炎发生率较高,构建的预测模型具有较高的临床应用价值。

Objective To establish and validate the risk prediction model of enterocolitis in children with Hirschsprung’s disease after radical operation.Methods From January 2018 to September 2020,200 children with congenital megacolon after operation in Hainan Women and Children’s Medical Center were selected and randomly divided into training set(140 cases)and validation set(60 cases).The risk of postoperative enterocolitis in the training set was analyzed and a prediction model was constructed.Receiver operating characteristic(ROC)curve,goodness-of-fit(H-L)test and calibration curve were used to evaluate the performance of the model.The clinical decision curve was drawn to evaluate the clinical applicability of the model.Results The average values of the length of spastic intestinal segment[10(26.32),28(73.68)],preoperative nutritional status[16(42.11),22(57.89)],preoperative history of upper respiratory tract infection(URTI)[14(36.84),24(63.16)],preoperative enteritis[23(60.53),15(39.47)],preoperative bowel washing time[18(47.37),20(52.63)],postoperative intestinal obstruction[10(26.32),28(73.68)]in the enterocolitis group were lower than those in the non-enterocolitis spastic intestinal length[55(53.92),47(46.08)],preoperative nutritional status[15(14.71),87(85.29)],preoperative upper respiratory tract infection(URTI)history[11(10.78),97(89.22)],preoperative enteritis[16(15.69),86(84.31)],preoperative bowel washing time[77(75.49),25(24.51)],postoperative intestinal obstruction[8(7.84),94(92.16),P<0.05].The area under the ROC curve(AUC)of the training set[0.904,95%confidence interval(CI)=0.854-0.954]was higher than that of the validation set[0.889,95%CI=0.767-0.971,P<0.05],and the MAE(0.030)was higher than that of the validation set(0.022).The results of H-L goodness of fit test showed thatχ^(2)=5.210,3.777,P>0.05,and the model fitting was good.Conclusion The incidence of enterocolitis after radical resection in children with congenital megacolon is high,and the constructed prediction model has high clinical application value.