Purpose: This study aims to create a prediction model which can estimate risk for CVD death among Filipinos based on currently known risk factors using survival analysis.
Methods: Baseline characteristics of age, sex, HDL, LDL, diabetes mellitus (DM), hypertension, and smoking among Filipinos were derived from the Food and Nutrition Research Institute (FNRI) data of 1998. The 2007 data of the Philippine Cardiovascular Outcomes Study (PhilCOS) identified the CVD outcomes after 9 years. Six out of 13 national regions were included in the cohort of 1749 responders. Model selection used was combination of forward selection and backward elimination also known as stepwise procedure. Survival techniques specifically Weibull regression model and frailty model, to account for the other unmeasured risk factors, were fitted to the data. Predicted survival time in years was computed and compared to the observed data.
Results: Age, gender, blood pressure, DM disease, and smoking history significantly predicted CVD death among Filipinos. The effect of HDL and LDL was not statistically significant in the model development. In general, predicted values were higher by an average of 1.1 years.
Conclusions: Among Filipinos the most significant risk factors for CVD death were age, gender, blood pressure, DM disease, and smoking history. Filipinos who had a greater chance of survival were younger females, with normal blood pressure, non-diabetics and never smokers. A sex-specific model using Weibull distribution was chosen to be the distribution of choice. Incorporating frailty was not needed in this data.
See more of: SMDM Asia-Pacific Conference