Re-Evaluating Recurrent Events in Heart Failure Trials: Patterns, Prognostic Implications, and Analytical...

Area of Science:

• Cardiology
• Clinical Trials
• Biostatistics

Background:

• Repeat hospitalizations for heart failure (HFHs) are frequently analyzed in clinical trials.
• Current analyses often assume events occur randomly over time.
• The statistical power of using repeat events is often considered beneficial.

Purpose of the Study:

• To challenge the assumption of random event occurrence in heart failure trials.
• To evaluate the impact of within-patient event clustering on statistical analysis.
• To reassess the optimal methods for analyzing repeat events in heart failure.

Main Methods:

• Utilized data from four heart failure trials involving sodium-glucose cotransporter 2 inhibitors.
• Investigated within-patient time clustering of repeat hospitalizations and cardiovascular death.
• Compared traditional models (Lin-Wei-Yang-Ying, negative binomial) with alternative methods (area under the curve, win ratio) and time-to-first event analyses.

Main Results:

• Demonstrated significant within-patient time clustering of repeat events, with elevated risks post-hospitalization.
• Found that standard models do not adequately capture this clustering.
• Alternative methods showed some benefit, but time-to-first event analyses provided the strongest evidence.

Conclusions:

• Commonly used repeat event analyses in heart failure trials may not be optimal.
• The non-random, clustered nature of repeat events requires re-evaluation of analytical approaches.
• Rethinking the utilization of repeat events data is crucial for accurate interpretation in heart failure research.