Estimating the Calibrating Coefficient of the Framingham Score to Predict Risk of Coronary Heart Disease in the Taiwanese Population

Background: The Framingham score, which was developed in the United States, is often calibrated and used in various countries to predict 10-year risk of coronary events, based on the measurements of age, sex, total cholesterol, high-density lipoprotein cholesterol (HDL-C), smoking status, and systolic blood pressure. However, no calibration coefficient is currently available for Taiwan. Methods: Data from the Taiwanese Survey on Hypertension, Hyperglycemia, and Hyperlipidemia (TwSHHH) were used to calibrate the Framingham equation for Taiwanese usage and compared with coefficients of the Chinese Multi-provincial Cohort Study (CMCS). Coronary events were identified through the link to National Health Insurance claim data and the national death registry for 2011. The risk factors were total cholesterol (mg/dL), systolic blood pressure (mmHg), cigarette smoking (yes/no), and diabetes (yes/no). The mean of these risk factors and the baseline survival probability were derived from TwSHHH. They were applied to the Framingham score function. Finally, the ratio of observed/ predicted was applied to calibrate the predicted probabilities. Results: When applying the Framingham function, agreement between the predicted and observed risk matched reasonably well in Taiwanese males, but not in females. The CMCS coefficients did not fit the Taiwanese population well. We recommend using 0.7958 and 1 as calibration coefficients for males and females, respectively. Conclusion: We generated Framingham calibration coefficient for the Taiwanese population. We recommend that the mean of predictors and the baseline survivorship derived from TwSHHH should be used in the model. Nonetheless, it is crucial to develop a risk function specific for this population.