Development And Validation Of A Clinical Prediction Model For Predicting Risk Of Colorectal Adenoma

Objective:Aim to identify the risk factors of colorectal adenoma in average-risk population,and develop a predictive model.Methods:Data were collected from 1050 patients who were hospitalized in the Department of Gastroenterology,Xinjiang Autonomous Region People’s Hospital and underwent colonoscopy between October 2021 and April 2022.Among them,365 patients with colorectal adenoma confirmed by pathology were recorded as case group;685 patients without adenoma on colonoscopy were recorded as controls.The clinical data and laboratory data were collected and compared between the two groups.Predictors were selected using a least absolute shrinkage and selection operator(LASSO)logistic regression model and a backward stepwise strategy in a multivariate Logistic regression model.Combining these characteristics,a prediction model was constructed and a nomogram was drawn.The performance of the model was assessed by receiver operating characteristic(ROC)curves,calibration curves,and decision curve analysis(DCA),and internally validated by bootstrap resampling(1000 times).Results:Multivariate regression analysis showed that age≥60 years,smoking,hypertension,and diabetes were associated with adenoma in the general risk population of colorectal cancer.Based on this,a predictive model for adenoma in the general risk population of colorectal cancer was constructed.The area under the curve(AUC)from the nomogram prediction model was 0.787(95%CI[0.758,0.816]).Internal validation yielded an AUC of 0.786.The calibration curve shows good agreement between the predicted probability and the actual probability.The decision curves show that the predictive model is clinically useful.Conclusion:Age≥60 years,smoking,hypertension,and diabetes are risk factors for adenoma development in a population at general risk for colorectal cancer,and predictive models based on these factors may be useful for individualized counseling and treatment.