Abstract: Objective The study aimed at developing a 5-year risk prediction model of cardiovascular disease (CVD) based on competing risk model, and developing a user-friendly risk score tool for risk assessment of cardiovascular disease.Methods We used competing risks model to evaluate the risks of developing a first CVD event. Sub-distribution hazard ratios (SHR) and 95% confidence intervals (95%CI) were computed for the associations between each risk factor and CVD.Time-dependent receiver operating characteristic (ROC) curve and time-dependent areas under the ROC curves (AUC) were used to evaluate the discrimination ability of the models, calibration plot was applied to assess the calibration of the models. Internal validation of predictive accuracy was performed via 1 000 times of bootstrap resampling.Results Of the 1 775 subjects without CVD at baseline, 473 incident cases of CVD were documented for a median 8-year follow-up. The participants were 886 men and 889 women between the ages of 55 and 96 years old without cardiovascular disease at baseline with 20 years of follow-up.Multivariable risk functions were derived that incorporated age, total cholesterol(TC), high-density lipoprotein cholesterol(HDL-C), blood pressure, smoking, diabetes, marital status and overweight or obesity body mass index (BMI). The model showed good discrimination and calibration.Male and female cardiovascular disease risk scores were 65 points and 62 points, the best point were 34 points and 30 points. Conclusion A sex-specific multivariable risk factor algorithm based competing risk model was made. The BLSA CVD risk prediction model can be used to predicted an individual's risk of CVD and provides a useful guide to identify the groups at high risk for CVD among over 55 years old man. A user-friendly risk score tool predicting 5-year probability of cardiovascular disease was developed.

Key words: cardiovascular disease, competing risks model, risk assessment, elderly population