Abstract 15516: Predicting Heart Failure Hospitalization in Type 2 Diabetes Mellitus: Validation of the TRS-HFDM Score in the ACCORD Trial
Background: Patients with type 2 diabetes mellitus (T2DM) are at an increased risk for developing heart failure (HF), and clinical models are needed to identify patients with a greater risk for HF hospitalization (HHF). The Thrombolysis in Myocardial Infarction Risk Score for Heart Failure in Diabetes (TRS-HFDM) was recently developed and validated to predict HHF in patients with T2DM in two large clinical trials (SAVOR-TIMI 53 and DECLARE-TIMI 58). We aimed to validate the TRS-HFDM in another cohort of patients with T2DM. Methods: We validated the TRS-HFDM score in 5,123 patients with T2DM for fatal or non-fatal HHF in the placebo arm of the ACCORD (Action to Control Cardiovascular Risk in Diabetes Study Group). The TRS-HFDM included: prior HF, history of atrial fibrillation, coronary artery disease, estimated glomerular filtration rate, and urine albumin-to-creatinine ratio. We evaluated discrimination with the Harrell C index, and calibration by comparing observed event rates for HHF with predicted risk, and the Nam-D’Agostino statistic. Results: During a mean follow up of 4.8 years, 212 patients (4.14%) experienced at least one HHF. The mean age was 63 ±6.7 years, and 38% were female. The baseline hemoglobin A1C was 8.3%, and 36% were on insulin. The ACCORD patients had less comorbidities with a lower proportion of patients with renal dysfunction (8% vs. 16% vs. 17%), established cardiovascular disease (35% vs.79% vs. 41%) and pre-existing HF (5% vs. 13% vs. 10%) compared to patients enrolled in the SAVOR-TIMI 53 and DECLARE TIMI-58 trials respectively. In our cohort, the TRS-HFDM score predicted well HHF events with a Harrel C index of 0.78. The integer-based score was well calibrated, with the observed Kaplan-Meier HHF rates closely predicting the Kaplan-Meier event rates at the end follow up (Nam D`Agostino test: 65.39 (p<0.0001). Discussion: Our findings confirm the applicability of the TRS-HFDM in a large cohort of patients with T2DM with less high-risk features than the SAVOR-TIMI 53 and DECLARE-TIMI 58 populations. Future validation of this risk score in observational cohort studies is needed to evaluate its external validity in a general clinical setting.