Abstract TP109: Impact of a Covert Cerebrovascular Disease-Aware Vascular Risk Model on Preventive Treatment Escalation and Major Adverse Cardiovascular Events: A MICROSIM Study

Introduction: Covert cerebrovascular disease (CCD) predicts subsequent cardiovascular events. However, conventional cardiovascular risk prediction tools, such as ASCVD, do not account for the added risk associated with CCD. A CCD-aware risk model may enable optimized prevention strategies. Objective: To compare changes in vascular risk factor treatment and clinical outcomes (stroke, myocardial infarction) between a conventional risk-based model and a CCD-aware model. Methods: MIchigan ChROnic Disease SIMulation (MICROSIM) is a population health simulation framework which estimates effects of vascular risk factor treatment on longitudinal cardiovascular outcomes. We adapted MICROSIM to capture patient characteristics of over 18,000 patients from Kaiser Permanente with available CCD status (CCD identification modality (CT or MRI), presence of silent brain infarction, and presence/severity of white matter hyperintensities). Two baseline cardiovascular risk models were implemented and compared: a conventional vascular risk-based model and a CCD-aware model that added CCD covariates. Risk-aware treatment (high intensity statin and SPRINT-level blood pressure control) was implemented for individuals with model predicted 10-year cardiovascular risk > 7.5%. Changes in treatment and outcomes were tabulated comparing a usual care population to risk-aware treatment using the conventional cardiovascular risk model and the CCD risk model. Results: Baseline risk in the CCD population is high ~ 20% 10-year risk of stroke/MI. The CCD-aware risk model substantially reclassified conventional baseline risk: 18% of patients were classified in a higher risk quintile when accounting for CCD and 27% of patients were classified in a lower risk quintile. Compared to usual care, a conventional model led to substantial changes in treatment (mean of 0.90 blood pressure meds added, 60% with statin added) and resulted in 38% relative risk reduction for stroke and 28% for MI. The CCD risk model led to very small treatment changes compared to the conventional model, and little change in outcomes. Conclusion: CCD populations are very high risk for cardiovascular events and may be good targets for interventions to reduce cardiovascular risk. Accounting for CCD information substantially altered 10-year cardiovascular risk estimates, but only modestly impacted treatment assignment and outcomes likely because baseline risk was too high for risk reassignment to alter treatment.