Development and validation of a risk score predicting risk of colorectal cancer

摘要: Background: Quantifying the risk of colorectal cancer (CRC) for individuals is likely to be useful health service provision. Our aim was develop and externally validate a prediction model predict 5-year CRC risk. Methods: We used proportional hazards regression based on established personal lifestyle factors using data from 197,874 45 Up Study, Australia. subsequently validated 24,233 participants Melbourne Collaborative Cohort Study (MCCS). Results: A total 1,103 224 cases were diagnosed in development validation sample, respectively. that includes age, sex, BMI, prevalent diabetes, ever having undergone screening, smoking, alcohol intake, exhibited discriminatory accuracy 0.73 (95% CI=0.72-0.75) 0.70 CI=0.66-0.73) Calibration good both study samples. Stratified models according screening history, additionally included family showed accuracies 0.75 (0.73-0.76) (0.67-0.72) unscreened screened In discrimination 0.68 (0.64-0.73) 0.72 (0.67-0.76), Conclusion: adequate predictive performance which maintained external population. Impact: The may design more powerful prevention trials group individuals, as pre-selection tool population-based programs.