Methods: EPHESUS was a multi-center trial conducted to investigate the effect of Eplerenone treatment in acute myocardial infarction survivors with subsequent congestive heart failure. In a sub-sample of 1,628 patients, utilities were assessed using the EQ-5D. Data from the first three months of the trial were excluded in order to prevent bias due to the effect of acute myocardial infarction. We then used mixed linear modeling to account for repeated measurements. The dependent variable of interest was aggregate EQ-5D score of health status based on US preferences. As independent variables we used the New York Heart Association (NYHA) classes of disease severity and number of hospitalizations as both are accepted proxies for disease progression and have been used in previous decision-analytic models. In multivariate analyses, we adjusted the results by age and gender.

Results: We based our analyses on 4,575 measurements from 1,628 patients. In the univariate analysis, utilities associated with NYHA classes I-IV were 0.88, 0.80, 0.70, and 0.53 (p for trend < 0.01), respectively. For number of rehospitalizations (0, 1, 2, 3+) the associated utilities were 0.80, 0.77, 0.76, 0.72 (p for trend < 0.01). While the association between NYHA class and utility remained constant over time, utilities based on the number of rehospitalizations improved with increasing time since previous hospitalization. Adjustment for age and gender did not affect the trend in either analysis. While utility estimates based on NYHA classes varied significantly by gender only, utilities based on number of rehospitalizations also changed significantly with age.

Conclusions: NYHA class and number of rehospitalizations are both established proxies for CHF progression and can be linked to utilities in a meaningful way when used as health states in a Markov model. NYHA class may be more closely linked to utilities as these estimates were not affected by age and did not change over time.