Prediction of outcomes in patients with heart failure (HF) may inform prognosis, clinical decisions regarding treatment selection, and new trial planning. The VICTORIA trial included high-risk patients with HF and reduced ejection fraction and a recent worsening HF event. The study participants had an unusually high event rate despite usage of contemporary guideline-based therapies. To provide generalizable predictive data for a broad population with a recent worsening HF event, we focused on risk prognostication in the placebo group.
Data from 2524 participants randomized to placebo with chronic HF (New York Heart Association class [NYHA] II-IV) and ejection fraction <45% were studied and backward variable selection was used to create Cox proportional hazards models for clinical endpoints, selecting from 66 candidate predictors. Final model results were produced, accounting for missing data, non-linearities, and interactions with treatment. Optimism-corrected c-indices were calculated using 200 bootstrap samples.
During a median follow-up of 10.4 months, the primary outcome of HF hospitalization or cardiovascular death occurred in 972 (38.5%) patients. Independent predictors of increased risk for the primary endpoint included HF characteristics (longer HF duration and worse NYHA class), medical history (prior myocardial infarction), and laboratory values (higher N-terminal pro-hormone B-type natriuretic peptide, bilirubin, urate; lower chloride and albumin). Optimism-corrected c-indices were 0.68 for the HF hospitalization/cardiovascular death model, 0.68 for HF hospitalization/all-cause death, 0.72 for cardiovascular death, and 0.73 for all-cause death.
Predictive models developed in a large diverse clinical trial with comprehensive clinical and laboratory baseline data-including novel measures-performed well in high-risk HF patients who were receiving excellent guideline-based clinical care.
Clinicaltrials.gov identifier, NCT02861534.
Patients with heart failure may benefit from tools that help clinicians better understand patient’s risk for future events like hospitalization. Relatively few risk models have been created after worsening of heart failure in a contemporary cohort. We provide insights on risk factors for clinical events from a recent large, global trial of patients with worsening heart failure to help clinicians better understand and communicate prognosis and select treatment options.

Copyright © 2021. Published by Elsevier Inc.