医学
逻辑回归
结肠镜检查
风险评估
内科学
危险分层
弗雷明翰风险评分
无症状的
人口
风险因素
粪便潜血
结直肠癌
胃肠病学
置信区间
粪便
接收机工作特性
优势比
癌症
环境卫生
疾病
计算机科学
计算机安全
作者
Yi Liao,Senmao Li,Hao Chen,Chun-Yu Chen,Jintuan Huang,Feng Lin,Jian-Ping Wang,Zuli Yang
标识
DOI:10.1097/cej.0000000000000622
摘要
Fecal immunochemical test (FIT) is often used for preselection for colonoscopy, but FIT has nonoptimal sensitivity. Selection of study participant for colonoscopy based on the result of combining FIT with risk factors could improve the sensitivity of a screening program. We aimed to develop a risk prediction system of colorectal neoplasia among asymptomatic Chinese subjects. A total of 6265 asymptomatic participants with age between 50 and 70 years were invited to undergo colonoscopy screening. They were also asked to take FIT and complete a questionnaire for collecting information on risk factors. Independent risk factors were identified by binary logistic regression for colorectal neoplasia. A risk score model was developed by using the odds ratios of significant risk factors. The scoring system was divided into two groups of risk: negative risk and positive risk. The performance of the risk score model in terms of predicting colorectal neoplasia was evaluated. Of the 1786 colonoscopy screening participants, 1546 completed FIT and questionnaires. A total of 462 cases of neoplasia were detected. Based on the scoring stratification, 966 (62.5%) participants were in negative risk tier and 580 (37.5%) were in positive risk tier. The incidence of colorectal neoplasia in negative risk and positive risk groups was 18.4 and 49.0%, respectively. Risk stratification model has better ability to discriminate those with or without colorectal neoplasia than FIT-only model. Classification improved significantly with risk stratification-based screening (net reclassification improvement = 0.064, P = 0.032). Risk stratification system increases the predictive value of FIT-based screening and is useful for preselection for colonoscopy in colorectal cancer screening program.
科研通智能强力驱动
Strongly Powered by AbleSci AI