A Novel Microsimulation Model For Evaluating Cvd Risk Assessment With Coronary Artery Calcium In Diabetes

Background: Several studies including the Multi-Ethnic Study of Atherosclerosis (MESA) have demonstrated the value of coronary artery calcium (CAC) for CVD risk prediction. However, whether CAC can refine risk assessment for patients with diabetes to improve long-term outcomes at reasonable cost is uncertain. We therefore developed a microsimulation model and determined its internal validity. Methods: We constructed a state-transition microsimulation model (R and TreeAge software) with Well, Post CVD, and Death states using data on 5,492 MESA participants (N=607 with diabetes at baseline, median follow-up 8.6 years). To generate individualized transition probabilities, we used Cox regression of coronary heart disease, heart failure, stroke, and competing death with age as time scale. Microsimulations were based on baseline risk factors (sex, ethnicity, smoking, diabetes, bp, lipids, BMI, CAC) with incident diabetes (N=610), CAC and CVD as transient states. We compared predicted with observed 9-year risks and calculated C-statistics. Results: Our model showed good calibration and discriminative performance across the various outcomes in the MESA cohort at 9 years (Table). Conclusion: We developed a new microsimulation model that will be used to evaluate long-term cost effectiveness of CAC-based risk assessment in diabetes. Validity will be further evaluated in independent external data. Disclosure B. Ferket: None. M. Hunink: None. U. Masharani: Research Support; Self; Clementia Pharmaceuticals. W. Max: None. K. Fleischmann: None. Funding American Diabetes Association (1-18-ICTS-041 to K.F.)