Related Experiment Video
Updated: Jan 19, 2026
Establishing a Competing Risk Regression Nomogram Model for Survival Data
Published on: October 23, 2020
Risk score development for pancreatic cancer in Chinese men: A population-based cohort study
Jie Cai1, Hongda Chen2, Yuhan Zhang3
1Department of Hospital Infection Control, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing 100730, China.
Background:
Pancreatic cancer (PC) is a malignant tumor without effective screening methods in the general population. This study aimed to develop a PC risk score to identify men at high risk for PC for early intervention.
Methods:
Based on a prospective cohort study conducted in China in 2006, 90,402 men with at least two fasting blood glucose (FBG) measurements were included in this study. The PC risk model was developed using Cox regression, including the risk factors for age, total cholesterol, FBG and body mass index (BMI) change rate. The power of discrimination was examined using Harrell's C-index and time-dependent area under curves (AUCs).
Results:
During the 10-year follow-up period, 82 men developed PC (total person-years: 821,346). Age, total cholesterol level, FBG and BMI change rate were included in the final PC risk model, with a C-index of 0.754 (95 % CI 0.709-0.799). After internal validation, the AUCs of the model in different years (2, 5, 8, and 10 years) were > 0.700 in all men and in men without diabetes. In the derived scoring system (score range:-2-10), there was an approximately 40 % increased risk (Hazard ratio, 1.415; 95 % CI, 1.291-1.550) with every 1-point increase. Participants with the top 20 % scores (6-10 points) had a 30-fold increased risk of PC compared with individuals in the lowest score group (-2 - -1 points).
Conclusions:
A scoring system for identifying high-risk men with PC was developed, which may be helpful for early detection and intervention of PC in the future.
