结直肠息肉术后6个月复发风险诺莫图预测模型的建立及验证

Establishment and validation of nomograph prediction model for recurrence risk of colorectal polyp 6 months after operation

目的分析结直肠息肉术后复发的危险因素,构建预测模型,并对模型进行验证以评价其预测效能。方法收集2021年7月至2021年12月于新疆维吾尔自治区人民医院消化内科因结直肠息肉治疗后6个月复查结肠镜的患者共312例,将其列为训练组,另收集2022年1月至4月因结直肠息肉治疗后6个月复查结肠镜的患者共134例作为验证组,通过单因素分析及多因素Logistic回归分析建立诺莫图预测结直肠息肉复发的模型,采用ROC曲线评估预测模型的区分度,校准曲线评估数据拟合情况,决策曲线分析图评估使用该模型的获益情况。结果年龄(OR=1.096)、性别(OR=5.752)、腺瘤性息肉(OR=7.604)、息肉个数≥3(OR=7.818)、吸烟(OR=7.846)是术后6个月结直肠息肉复发的独立危险因素(P<0.05),上述因素构建诺莫图预测模型,在训练组中AUC为0.874(95%CI:0.830~0.919),验证组AUC为0.861(95%CI:0.792~0.930),Hosmer-Lemeshow拟合优度检验显示平均绝对误差小,拟合度良好。决策性曲线图提示获益概率范围广,有实用价值。结论联合年龄、性别、腺瘤性息肉、息肉个数、吸烟建立的结直肠息肉术后6个月复发风险预测模型具有较好的预测能力,有利于医务人员对结直肠息肉术后的患者复发风险进行预测并进行干预提供有效工具。

Objective To explore the risk factors of colorectal polyp recurrence after operation,construct the prediction model,and verify the prediction efficiency of the model.Methods A total of 312 patients who underwent colonoscopy 6 months after treatment of colorectal polyp in the Department of Gastroenterology,People′s Hospital of Xinjiang Uygur Autonomous Region from July 2021 to December 2021 were included in the training group,and another 134 patients who underwent colonoscopy 6 months after treatment of colorectal polyp from January to April 2022 were included in the verification group.Univariate analysis and multivariate logistic regression analysis were used to establish nomograph model to predict colorectal polyp recurrence.ROC curve was used to evaluate the distinguishing ability of the prediction model,calibration curve was used to evaluate the data fit,and decision curve analysis diagram was used to evaluate the benefit of using the model.Results Age(OR=1.096),gender(OR=5.752),adenomatous polyp(OR=7.604),number of polyps≥3(OR=7.818)and smoking(OR=7.846)were independent risk factors for colorectal polyp recurrence 6 months after surgery(P<0.05).Nomograph prediction model was constructed for these factors.The AUC of the training group was 0.874(95%CI:0.830~0.919),and that of the validation group was 0.861(95%CI:0.792~0.930).Hosmer-lemeshow goodness of fit test showed small mean absolute error and good fit.The decision curve indicates a wide range of benefit probability and has practical value.Conclusion Combined with age,gender,adenomatous polyps,number of polyps and smoking,the prediction model of colorectal polyp recurrence risk has good prediction ability,which is beneficial for medical staff to predict the recurrence risk of patients with colorectal polyps after treatment and provide effective tools for intervention.