Validation of the Seattle Heart Failure Model in a community-based heart failure population and enhancement by adding B-type natriuretic peptide.

Management of heart failure (HF) remains complex with low 5-year survival. The Seattle Heart Failure Model (SHFM) is a recently described risk score derived predominantly from clinical trial populations that may enable the prediction of survival in patients with HF. This study sought to validate the SHFM in an independent, nonclinical trial-based HF population. Patients (n = 4,077) from the hospital-based Intermountain Heart Collaborative Study registry with a diagnosis of HF were evaluated using prospectively collected data (mean +/- SD follow-up 4.4 +/- 3.1 years). The SHFM was used to calculate a risk score for each patient. Receiver-operating characteristic area under the curve provided SHFM predictive ability for a composite end point of survival free from death, transplantation, or left ventricular assist device implantation. Addition of creatinine, serum urea nitrogen, diabetes status, and B-type natriuretic peptide (BNP) to the SHFM was also evaluated. Patient age averaged 67 +/- 13 years and 61% were men. Area under the curves were 0.70 (95% confidence interval 0.66 to 0.70), 0.67 (95% confidence interval 0.66 to 0.69), 0.67 (95% confidence interval 0.065 to 0.68), and 0.66 (95% confidence interval 0.63 to 0.67) for 1-, 2-, 3-, and 5-year survivals, respectively. Area under the curves were slightly attenuated in patients >75 years of age (n = 1,339), implantable cardioverter-defibrillator recipients (n = 693), and patients with an ejection fraction >40% (n = 1,634). BNP added significantly to the model (area under the curve +0.06). BNP was found to add additional predictive ability at 1 year (area under the curve change +0.05) and nominally at 5 years (area under the curve change +0.02). In conclusion, the SHFM predicts survival in patients with HF in a hospital-based population, with areas under the curve similar to those from data on which models were initially fit.