Interaction analysis of the new pooled cohort equations for 10-year atherosclerotic cardiovascular disease risk estimation: a simulation analysis

OBJECTIVES: To evaluate the individual and interacting impacts of the continuous variables (age, total cholesterol (total-C), high-density lipoprotein cholesterol (HDL-C) and systolic blood pressure(BP)) on 10-year atherosclerotic cardiovascular disease (ASCVD) risk and better understand the pattern of predicted 10-year risk with change of each variable using recently published new pooled cohort equations. DESIGN: Simulation analysis was performed across the whole range of the boundary limits suggested for the continuous variables for groupings based on race and gender in the pooled cohort 10-year risk equations. SETTING: Computer-based simulation analysis. PARTICIPANTS: Data were generated by simulation using prespecified variable ranges. INTERVENTION: Data simulation and visual display of the hazard analysis. MAIN OUTCOME MEASURES: Interactions of age with other variables were analysed using multidimensional visualisation and hazard analysis. RESULTS: In African–American females, due to the interaction of age with HDL-C, treated BP and untreated BP, increasing age may not always increase 10-year risk. Furthermore, in the same cohort, increasing HDL-C level may result in higher 10-year risk for older individuals. For Caucasian females, due to square of Ln (age) term in the equation, the age-risk curve does not monotonically increase with age. The vertex is within the given age range of 40–79 years for a certain range of total-C and HDL-C, indicating that age may not always result in increased predicted 10-year risk. CONCLUSIONS: The new pooled cohort equations are sophisticated as they take into account the interactions of the continuous variables in predicting 10-year risk. We find situations where the estimated 10-year risk does not follow the general secular trends. The impact of such interesting patterns may be substantial and therefore further exploration is needed as it has direct implications in clinical management for primary prevention.