Background: The Seattle Heart Failure Model (SHFM) predicts mortality in patients with heart failure, but no study has investigated model performance in subjects with cardiac resynchronization therapy (CRT).
Methods and results: Prospectively collected registry data of 427 patients with CRT were used for the evaluation of the discrimination and calibration performance of the SHFM. The primary end point was all-cause mortality. Calibration profile was assessed with the use of Hosmer-Lemeshow tests, and discrimination was assessed by calculating areas under receiver operating characteristic curves (AUCs) derived from the model. Mean (SD) age of the study population was 61.2 (11.1) years (72.6% male). During a median follow-up of 24.7 months, 16.9% (72/427) of the patients reached the primary end point, with observed outcomes of 34/303, 54/234, and 72/97 at years 1, 2, and 5, respectively. No evidence of insufficient fit was found in any calibration tests. AUC estimates in all subjects were 0.7377 (95% CI 0.6575-0.8179) at year 1, 0.7936 (95% CI 0.7317-0.8556) at year 2, and 0.7572 (95% CI 0.6455-0.8689) at year 5.
Conclusions: SHFM offers an accurate prediction of survival in patients with CRT, with good observed calibration and AUC estimates indicating discrimination performance similar to those found by earlier validation studies.
Copyright © 2012 Elsevier Inc. All rights reserved.