Derivation and Validation of a Prediction Rule for Estimating Advanced Colorectal Neoplasm Risk in Average-Risk Chinese

Abstract

No prediction rule is currently available for advanced colorectal neoplasms, defined as invasive cancer, an adenoma of 10 mm or more, a villous adenoma, or an adenoma with high-grade dysplasia, in average-risk Chinese. In this study between 2006 and 2008, a total of 7,541 average-risk Chinese persons aged 40 years or older who had complete colonoscopy were included. The derivation and validation cohorts consisted of 5,229 and 2,312 persons, respectively. A prediction rule was developed from a logistic regression model and then internally and externally validated. The prediction rule comprised 8 variables (age, sex, smoking, diabetes mellitus, green vegetables, pickled food, fried food, and white meat), with scores ranging from 0 to 14. Among the participants with low-risk (≤3) or high-risk (>3) scores in the validation cohort, the risks of advanced neoplasms were 2.6% and 10.0% (P < 0.001), respectively. If colonoscopy was used only for persons with high risk, 80.3% of persons with advanced neoplasms would be detected while the number of colonoscopies would be reduced by 49.2%. The prediction rule had good discrimination (area under the receiver operating characteristic curve = 0.74, 95% confidence interval: 0.70, 0.78) and calibration (P = 0.77) and, thus, provides accurate risk stratification for advanced neoplasms in average-risk Chinese.

Risk Prediction Models for Colorectal Cancer: A Systematic Review

“…It includes only age, gender, and BMI and has AUROCs of 0.65 (95% CI, 0.61-0.69) and 0.71 (95% CI, 0.64-0.78) in external validation studies in China (22,24). The only risk scores using routine data for colon cancer and colorectal cancer were developed by Driver and colleagues from a cohort of 21,581 men in the United States Mutagen index considers cooking temperature, frequency of consumption of meats cooked at high temperature and how well-done the meats are.…”

Risk prediction models for colorectal cancer: A Scoping review

“…Three models have been developed in China [16,23,36] and Japan [40] while, Netherlands [25,26], Germany [21], Taiwan [32,46] and United Kingdom [31,34] have developed two models each. Many other countries, such as Australia [19], Poland [24], Hong Kong [28], Italy [29], Qatar [41], Saudi Arabia [43], Spain [52] and Sri Lanka [15] have also developed country-specific risk prediction models for colorectal cancer.…”

“…However, 11 models are specific for females [20,22,35,42,45,50] while 10 models are specific for males [20,22,35,45,48]. Although many models have not restricted application to a specific age limit, 11 models have been developed from populations aged 30 years and above [15,20,42,43,50], 10 from populations aged 50 years and above [18,25,26,45,51] and seven from populations aged 40 years and above [14,19,23,24,36,38,52]. In addition, one model has used a population above 20 years [35] while another, populations lower than 80 years [34].…”

“…A majority (n=35) of models include variables obtained through a questionnaire [18,19,21,22,23,24,25,26,27,28,30,33,36,37,38,39,40,41,42,43,44,45,46,49,50,51,53,54]. The questionnaires range from those with only one or two simple questions to those with detailed questions on factors such as diet, physical activity, alcohol and smoking habits and past medical facts.…”

“…A majority of the studies have used multiple logistic regression to identify the risk predictors [16,27,28,30,33,34,35,41,43,47,50,52] followed by the allocation of the risk points, based on the values of beta-coefficients [14,15,18,21,23,24,36,37,39,45,46,48]. Fourteen models have incorporated Cox proportional hazards regression to develop the risk model and the score [17,19,20,22,38,40].…”