Testing treatment effects in the presence of competing risks

Area of Science:

• Clinical Trials
• Biostatistics
• Survival Analysis

Background:

• Competing risks are common in clinical research, necessitating careful endpoint selection.
• The time to the first event is often used as an overall measure, but may not reflect treatment-specific effects.
• Failure-specific endpoints are crucial for evaluating treatments targeting particular disease aspects.

Purpose of the Study:

• To review methodologies for testing failure-specific treatment effects in the presence of competing risks.
• To compare the performance of the cause-specific log-rank test and cumulative incidence-based approaches for this purpose.

Main Methods:

• Review of statistical methodologies for analyzing time-to-event data with competing risks.
• Simulation studies to evaluate the robustness and power of different statistical tests.
• Comparison of cause-specific log-rank tests versus cumulative incidence-based methods.

Main Results:

• The cause-specific log-rank test demonstrated robustness in preserving nominal significance levels.
• The cause-specific log-rank test exhibited good power for detecting differences in marginal latent failure-time distributions.
• Cumulative incidence-based approaches were found to be potentially problematic for testing specific failure-time differences.

Conclusions:

• The cause-specific log-rank test is a reliable and powerful tool for assessing treatment effects on specific failure types in competing risks settings.
• Researchers should carefully consider the choice of statistical methods when analyzing failure-specific outcomes in clinical trials.
• The findings suggest that cause-specific log-rank tests are preferable to cumulative incidence methods for evaluating marginal failure-time distributions.