Abstract 18551: Predictors of Heart Failure and Death in a High Risk Diabetic Population: The DAVID-Berg Study

Abstract

Introduction: Diabetic subjects are at high risk of heart failure (HF) and death. Few data are available on the predictive value of NTproBNP, ECG, and echocardiogram in this population Hypothesis: A screening strategy with NTproBNP, ECG, and echo may improve prediction of HF/death events in a diabetic population in HF-stages A & B, beyond clinical data Methods: In 2008, among 4047 subjects aged >55 and <80 yrs treated by three general medicine group practices, the DAVID-Berg study enrolled all subjects with diabetes &/or hypertension &/or cardiovascular disease, without history/symptoms/signs of HF (n=623). Subjects underwent clinical visit, 12-lead ECG, echocardiogram, NTproBNP point of care testing. Recently data on HF/death events have been prospectively collected for all diabetic subjects included in the study (n=198) Results: Median age of the study population was 69 yrs, 62% were men, mostly hypertensive (86%) with known cardiovascular disease (44%). During a median follow-up of 5.2 [4.9,5.5] yrs there were 45 HF/death. At Cox regression analyses univariate predictors of events were known cardiovascular disease, insulin therapy, chronic kidney dysfunction (CKD), NTproBNP≥300 pg/mL, abnormal ECG (atrial fibrillation &/or LV hypertrophy &/or bundle branch block &/or Q wave), left atrial volume index (LAVI)≥40 mL/m 2 , LV ejection fraction (LVEF)≤55%, and LV mass index. At multivariate analyses (stepwise forward selection applied to univariate predictors), clinical variables (CKD and known cardiovascular disease), ECG (HR 2.66 [95% CI 1.08-6.55]), LAVI≥40 mL/m 2 (HR 2.50 [95% CI 1.28-4.89]), and LVEF≤55% (HR 2.70 [95% CI 1.45-5.04]) remained significantly associated with the outcome. Adding sequentially NTproBNP, ECG, and echo (LAVI, LVEF) to clinical data improved event prediction (Figure) Conclusions: High risk diabetic subjects may be correctly risk stratified integrating sequentially clinical data with NTproBNP, ECG, and echocardiogram