Use of win time for ordered composite endpoints in clinical trials.

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Survival Analysis

Background:

• Clinical trials often assess treatment benefit by analyzing time-to-first-event data.
• Existing methods like the ordinary win ratio may not fully capture the patient's overall treatment experience.
• There is a need for statistical approaches that provide a more comprehensive assessment of treatment benefit and harm.

Purpose of the Study:

• To introduce and evaluate novel variants of the win ratio for clinical trials measuring multiple clinical events.
• To develop methods that better reflect the patient's perspective on overall treatment benefit and harm.
• To compare the performance of these new methods against existing approaches like the composite event and ordinary win ratio.

Main Methods:

• Development of several new win ratio variants incorporating time spent in clinical states.
• One variant prioritizes death while accounting for time in other states.
• Other variants include average pairwise win time, expected win times (EWTs) against a reference distribution, and methods using estimated clinical state distributions.
• A combination testing approach is proposed for robust power across various scenarios.
• Methods were compared using simulations and re-analysis of a heart failure trial.

Main Results:

• Simulations indicate that variants based on expected win times (EWTs) or estimated clinical state distributions offer substantially higher power when treatment benefit on death is significant.
• These novel methods demonstrated superior performance compared to pairwise comparison or composite event methods in specific scenarios.
• The methods provide a more nuanced evaluation of treatment effects, aligning better with patient-centered outcomes.

Conclusions:

• The proposed win ratio variants and combination testing approach offer improved statistical power and a more patient-relevant assessment of treatment benefit in clinical trials.
• Methods incorporating time spent in clinical states, particularly EWTs and estimated state distributions, are effective for detecting treatment benefits, especially concerning mortality.
• These advanced statistical tools enhance the interpretation of clinical trial results for guiding therapeutic decisions.