Key points are not available for this paper at this time.
AIMS: Several models for predicting the prognosis of heart failure (HF) patients have been developed, but all of them focus on a single outcome variable, such as all-cause mortality. The purpose of this study was to develop a multistate model for simultaneously predicting survival and HF-related hospitalization in patients discharged alive from hospital after recovery from acute HF. METHODS AND RESULTS: The model was derived in the COACH (Coordinating Study Evaluating Outcomes of Advising and Counseling in Heart Failure) cohort, a multicentre, randomized controlled trial in which 1023 patients were enrolled after hospitalization because of HF. External validation was attained with the FINN-AKVA (Finish Acute Heart Failure Study) cohort, a prospective, multicentre study with 620 patients hospitalized due to acute HF. The observed vs. predicted 18-month survival was 72.1% vs. 72.3% in the derivation cohort and 71.4% vs. 71.2% in the validation cohort. The corresponding values of the c statistic were 0.733 95% confidence interval (CI) 0.705-0.761 and 0.702 (95% CI 0.663-0.744), respectively. The model's accuracy in predicting HF hospitalization was excellent, with predicted values that closely resembled the values observed in the derivation cohort. CONCLUSION: The COACH risk engine accurately predicted survival and various measures of recurrent hospitalization in (acute) HF patients. It may therefore become a valuable tool in improving and personalizing patient care and optimizing the use of scarce healthcare resources.
Postmus et al. (Wed,) studied this question.