结直肠腺瘤性息肉发生的危险因素分析及预测模型构建

Risk factors of colorectal adenomatous polyps and construction of prediction model

目的分析结直肠腺瘤性息肉(CAP)发生的危险因素,建立风险预测模型并检测该模型的预测价值。方法回顾性分析2018年3月至2019年3月在南阳市第一人民医院诊治的103例行结直肠息肉切除术患者的临床资料,根据息肉的病理诊断结果分为腺瘤组46例和非腺瘤组57例。分析比较两组的病历资料,采用Logistic回归分析影响CAP发生的危险因素,并建立风险预测模型,采用Hosmer-Lemeshow拟合优度检验评价预测模型的校准度,受试者工作特征曲线(ROC曲线)检测该模型的预测效能。结果CAP的发生与患者体质指数(BMI),是否长期吸烟、饮酒,是否合并糖尿病、高血脂,临床是否有腹痛,便血特征、息肉数目、息肉直径,既往胆囊息肉史及胆囊结石史有关(P<0.05);Logistic回归分析显示,BMI≥24 kg/m^(2)、长期吸烟、长期饮酒、合并糖尿病、合并高血脂(OR=3.728、4.315、3.873、4.358、4.208,P<0.05)均为影响CAP发生的独立危险因素,建立Logistic回归风险预测模型:P=1/[1+e(-3.507+1.316×(BMI)+1.462×(吸烟)+1.354×(饮酒)+1.472×(糖尿病)+1.437×(高血脂))],Hosmer-Lemeshowχ^(2)=2.114,P=0.146,ROC分析显示,该模型预测CAP的AUC为0.778。结论BMI≥24 kg/m^(2)、长期吸烟、长期饮酒、合并糖尿病、合并高血脂的患者为CAP发生的重点关注人群,Logistic回归风险预测模型对CAP发生的预测有较好的效能。

Objective To analyze the risk factors of colorectal adenomatous polyps(CAP),construct risk prediction model and detect the predictive value of the model.Methods A retrospective analysis was performed on the clinical data of 103 patients undergoing colorectal polypectomy in Nanyang First People’s Hospital from March 2018 to March 2019.According to pathological diagnosis results of polyps,they were divided into adenoma group(46 cases)and non-adenoma group(57 cases).The data of the two groups were analyzed and compared.The risk factors of CAP were analyzed by Logistic regression analysis.The risk prediction model was constructed.The calibration of the prediction model was evaluated by Hosmer-Lemeshow goodness of fit test,and its predictive efficiency was detected by receiver operating characteristic(ROC)curve.Results The occurrence of CAP was related to BMI,long-term smoking,drinking,diabetes mellitus,hyperlipidemia,clinical abdominal pain,hemafecia,number and diameters of polyps,previous history of gallbladder polyps and history of gallbladder stones(P<0.05).Logistic regression analysis showed that BMI≥24 kg/m^(2),long-term smoking,long-term drinking,diabetes mellitus and hyperlipidemia(OR=3.728,4.315,3.873,4.358,4.208,P<0.05)were independent risk factors of CAP.The Logistic regression risk prediction model was as follow:P=1/[1+e(-3.507+1.316×(BMI)+1.462×(smoking)+1.354×(drinking)+1.472×(diabetes mellitus)+1.437×(hyperlipidemia))],Hosmer-Lemeshow χ^(2)=2.114,P=0.146.ROC analysis showed that AUC of the model for predicting CAP was 0.778.Conclusions CAP is more prone in patients with BMI≥24 kg/m^(2),long-term smoking,long-term drinking,diabetes mellitus and hyperlipidemia.The predictive efficiency of Logistic regression risk prediction model is good for CAP.