Abstract 19894: Lifetime Predictions of Coronary Heart Disease, Congestive Heart Failure, and Stroke to Enable Preventive Treatment Decisions in Type 2 Diabetes Patients
Introduction: Lifetime CVD predictions are increasingly recommended to support shared decision making on preventive interventions. We sought to predict lifetime CVD outcomes in T2D patients according to various preventive scenarios using the Action to Control Cardiovascular Outcomes in Diabetes (ACCORD) trial cohort.
Methods: We developed cumulative incidence functions for prediction of coronary heart disease (CHD), congestive heart failure (CHF), and stroke in 9,688 participants. We performed cause-specific Cox regression with age as time scale and adjusted for study arm, race, gender, smoking, systolic blood pressure, cholesterol, history of CVD, body mass index, renal function, and HbA1C. Discrimination (C-statistic) and calibration were assessed by cross-validation using 7-year follow-up data from the 7 clinical center networks, leaving one network out at each modelling step. We subsequently used hazard ratios for statin and aspirin therapy from meta-analyses of preventive trials to make projections of lifetime outcomes according to “status quo” and preventive intervention scenarios.
Results: C-statistics in the original study population (mean age 63±7, age range 44-79, 38% female) were: 0.69 (95% CI 0.67-0.70) for CHD, 0.76 (95% CI 0.74-0.78) for CHF, and 0.73 (95% CI 0.69-0.77) for stroke. In cross-validation, these values were generally reproduced. Prediction models calibrated well, with average predicted risks falling within 95% CIs of observed risks in most of the 7 validation populations (Table). A beta version of a decision tool (CO-DECIDER™) to calculate lifetime risks before and after preventive intervention is available at http://tinyurl.com/hvkrzb7.
Conclusions: We developed and cross-validated lifetime predictions of CHD, CHF and stroke in T2D patients. The lifetime risk calculator can be useful to assess the expected benefit of preventive interventions and could enable shared decision making between T2D patients and providers.
Author Disclosures: B.S. Ferket: None. G.N. Nadkarni: None. S.G. Coca: None.
- © 2016 by American Heart Association, Inc.