Abstract 16114: Incremental Prognostic Utility by Cost and Complexity in Chronic Heart Failure
Background: Current risk prediction models for chronic heart failure (HF) rely on a variety of assessments. Given the cost and limited availability of some of these assessments, we examined the incremental prognostic information provided by clinical, laboratory, quality of life (QOL), echocardiography, and cardiopulmonary exercise testing (CPX) data in a well-treated cohort of ambulatory patients with HF.
Methods: We included patients with complete data on clinical, laboratory, QOL, and exercise testing variables from HF-ACTION, a randomized controlled trial of exercise training vs. usual care in patients with chronic systolic HF, with an additional analysis performed in a sub-group with available NTproBNP. Cox proportional hazards regression models for the composite endpoint of all-cause death or all-cause hospitalization were fit by step-wise inclusion of the following information: (a) clinical data (b) lab tests (c) QOL surveys (d) echo parameters (e) CPX. The incremental value of each additional group of assessments was compared with the complete model using net reclassification improvement (NRI) across predefined risk categories (0-25%,25-50%,50-75%,75-100%), with and without availability of NTproBNP.
Results: Models with incremental information, not including NTproBNP, generally showed improvements in C-indices, with the clinical model alone yielding the lowest value (0.62) and the most complete model the highest (0.65) (Table ). Compared with the full model, the % NRI ranged from -27.2 (clinical model) to -14.8 (complete minus CPX). These differences were attenuated for models where NTproBNP data was included; C-indices improved from 0.66 to 0.67 and %NRI changes were marginal: -4.41 (-9.76, 0.93).
Conclusions: Knowledge of clinical data and readily available lab tests including NTproBNP provides similar prognostic information to more complex models that include assessments of echocardiographic and exercise variables.
- © 2013 by American Heart Association, Inc.