895-P: External Validation of the BRAVO Diabetes Model Using Data from the EXSCEL Study

We developed an individual-level discrete-time microsimulation model, namely the Building, Relating, Assessing, and Validating Outcomes (BRAVO) diabetes model. The BRAVO model was externally validated using cohort-level data in people with type 2 diabetes (T2D) . This study further externally validated the BRAVO diabetes model using patient-level data from the Exenatide Study of Cardiovascular Event Lowering Trial (EXSCEL) . Baseline characteristics and trajectories of key biomarkers, including A1c, LDL, SBP, and BMI, were retrospectively extracted from the EXSCEL study database. To protect patients’ private information, variables of age and BMI were masked in ranges from the EXSCEL study (e.g., BMI 35 coded as BMI>30) . To convert the ranges back into specific values to support the simulation, we used the National Health and Nutrition Examination Survey (NHANES) data for the data imputations. We used the median value for each population subgroup from NHANES to replace the ranges. The BRAVO diabetes model was applied to the EXSCEL baseline data to predict the 7-year EXSCEL study outcomes. We assessed the BRAVO model’s discrimination power using C-statistics and calibration using the Brier score. The model demonstrated good discrimination and calibration in predicting the first occurrence of non-fatal myocardial infarction (C-index 0.620 [95% CI 0.590-0.650] and Brier Score (BS) 0.080 [0.074-0.085]) , non-fatal stroke (C-index 0.696 [0.634-0.758] and BS 0.031 [0.028-0.034]) , heart failure (C-index 0.700 [0.666-0.733] and BS 0.037 [0.034-0.041]) , revascularization (C-index 0.664 [0.590-0.738] and BS 0.036 [0.033-0.039]) , and all-cause mortality (C-index 0.746 [0.711-0.781] and BS 0.102 [0.098-0.106]) , respectively. Although the majority of the data obtained from EXSCEL trial was obtained in range instead of specific values, the BRAVO model remained robust in prediction performance in diabetes complications and mortality.