Coefficients for all models.

Aims

Cardiovascular diseases (CVDs) are the most common cause of death, but they can be effectively managed through appropriate prevention and treatment. An important aspect in preventing CVDs is assessing each individual’s comprehensive risk profile, for which various risk engines have been developed. The important keys to CVD risk engines are high reliability and accuracy, which show differences in predictability depending on disease status or race. Framingham risk score (FRS) and the atherosclerotic cardiovascular disease risk equations (ASCVD) were applied to the Korean population to assess their suitability.

Methods

A retrospective cohort study was conducted using National Health Insurance Corporation sample cohort from 2003 to 2015. The enrolled participants over 30 years of age and without CVD followed-up for 10 years. We compared the prediction performance of FRS and ASCVD and calculated the relative importance of each covariate.

Results

The AUCs of FRS (men: 0.750; women: 0.748) were higher than those of ASCVD (men: 0.718; women: 0.727) for both sexes (Delong test P <0.01). Goodness of fits (GOF) were poor for all models (Chi-square P < 0.001), especially, underestimation of the risk was pronounced in women. When the men’s coefficients were applied to women’s data, AUC (0.748; Delong test P<0.01) and the GOF (chi-square P = 0.746) were notably improved in FRS. Hypertension was found to be the most influential variable for CVD, and this is one of the reasons why FRS, having the highest relative weight to blood pressure, showed better performance.

Conclusion

When applying existing tools to Korean women, there was a noticeable underestimation. To accurately predict the risk of CVD, it was more appropriate to use FRS with men’s coefficient in women. Moreover, hypertension was found to be a main risk factor for CVD.