中国人群结直肠癌个体发病风险的Rothman-Keller模型构建

Development of a Individual Rothman-Keller-type Risk Prediction Model for Colorectal Cancer in Chinese People

背景结直肠癌是严重威胁我国居民健康的恶性肿瘤之一。有效的风险评估模型能通过评估发病风险、筛选高危人群来引导居民建立健康的生活方式,是预防结直肠癌、降低结直肠癌死亡率的重要手段之一。目的构建适用于中国人群结直肠癌个体发病风险的Rothman-Keller模型,评估个体发病风险并划分风险等级。方法全面检索中国知网、万方数据知识服务平台、维普网、PubMed、EMBase、OVID数据库中关于结直肠癌危险因素的文献。检索年限为建库至2020-08-31。通过文献综述明确结直肠癌主要危险因素及其与结直肠癌发病风险的定量关系,构建中国人群结直肠癌个体发病风险的Rothman-Keller模型,采用二项分布函数法划分风险等级。结果构建的中国人群结直肠癌个体发病风险的Rothman-Keller模型中纳入了结直肠癌家族史、吸烟、饮酒、肥胖、红肉摄入、腌制食品摄入、阑尾炎病史、糖尿病病史、慢性结直肠炎病史、慢性腹泻病史、慢性便秘病史、蔬菜摄入、水果摄入、阿司匹林、体育锻炼等15项危险因素。二项分布函数法模拟10000例样本点的发病风险后,采用分段斜率比较的方式将女性患结直肠癌的低、高风险界值划分为0.0337%,男性患结直肠癌的低、高风险界值划分为0.0463%。结论构建的中国人群结直肠癌个体发病风险的Rothman-Keller模型简单易行,有利于促进高危人群参加结肠镜筛查,改变不良生活方式,使直肠癌预防关口前移。

Background Colorectal cancer is a malignancy seriously threatening the health of Chinese people.Effective risk prediction model for colorectal cancer is an important tool to guide people to develop a healthy lifestyle by screening morbidity risk and identifying high-risk people,to prevent or reduce the mortality of colorectal cancer.Objective To develop a Rothman-Keller-type model for predicting and classifying individual risk of colorectal cancer suitable for Chinese population.Methods We did a review of studies about risk factors for colorectal cancer that were collected from databases of CNKI,Wanfang Data,VIP,PubMed,EMBase,and OVID from inception to 2020-08-31 to quantitatively examine the relationship of colorectal cancer with its major risk factors,then used the results to develop a Rothman-Keller-type model for predicting individual risk of colorectal cancer in Chinese population.We applied binomial distribution function to further rate the risks.Results In total,there were 15 risk factors included in the Rothman-Keller-type model,including family history of colorectal cancer,smoking,drinking,obesity,excessive intake of red meat,high intake of preserved food,history of appendicitis,history of diabetes,history of chronic colorectitis,history of chronic diarrhea,history of chronic constipation,and low vegetable intake,low fruit intake,use of aspirin and take of physical exercise.We generated 10 000 random data sets using binomial distribution function,and using segmented gradient comparison method,we found the risk threshold for women was 0.033 7% and for men was 0.046 3%.Conclusion This risk prediction model is easy-to-use,which facilitates high-risk people to undergo colonoscopy,and help them to develop healthy lifestyles,promoting the prevention of colorectal cancer.