Moving beyond the hazard ratio in quantifying the between-group difference in survival analysis

Area of Science:

• Biostatistics
• Clinical Trials
• Survival Analysis

Background:

• Time-to-event endpoints are common in clinical studies, with hazard ratios (HR) frequently used for group comparisons.
• The HR assumes proportional hazards, meaning the relative risk between groups is constant over time.
• Violations of the proportional hazards assumption complicate HR interpretation and clinical relevance.

Purpose of the Study:

• To highlight the limitations of the hazard ratio when its underlying assumption is violated.
• To discuss alternative statistical methods for analyzing time-to-event data in clinical trials.
• To guide researchers in selecting appropriate measures for quantifying between-group differences in survival analysis.

Main Methods:

• Review of statistical literature on survival analysis and time-to-event data.
• Illustrative examples using data from three recent cancer clinical trials.
• Discussion of model-free measures and robust estimation procedures.

Main Results:

• The proportional hazards assumption is frequently violated in practice, challenging standard hazard ratio interpretation.
• Various alternative statistical approaches exist for quantifying differences in time-to-event outcomes.
• Cancer trial data demonstrate diverse scenarios where hazard ratios may be misleading.

Conclusions:

• Researchers should critically evaluate the proportional hazards assumption before relying solely on hazard ratios.
• Consideration of alternative, clinically meaningful, and model-free measures is recommended when the assumption is questionable.
• Robust estimation procedures enhance the reliability of inferences in time-to-event analyses.