Multiple tests for restricted mean time lost with competing risks data.

Area of Science:

• Biostatistics
• Survival Analysis
• Competing Risks

Background:

• Restricted mean time lost (RMTL) is a valuable estimand in competing risks survival analysis.
• Existing statistical tests for RMTL are limited to simple comparisons and a small number of event types.
• Continuity assumptions in current methods restrict their applicability to real-world data with ties.

Purpose of the Study:

• To develop general statistical tests for comparing RMTL in factorial designs with an arbitrary number of event types.
• To address limitations of existing RMTL tests by accommodating data ties and improving small sample performance.
• To introduce multiple testing procedures for simultaneous RMTL comparisons with enhanced statistical power.

Main Methods:

• Development of Wald-type test statistics for RMTL comparisons.
• Implementation of a permutation approach to enhance reliability and small sample performance.
• Incorporation of the asymptotic dependence structure for powerful multiple testing.

Main Results:

• The proposed methods provide flexible and robust RMTL comparisons for complex designs.
• The permutation-based tests demonstrate improved small sample performance.
• Multiple testing procedures effectively control Type I error rates while increasing power.

Conclusions:

• The developed statistical tests offer a significant advancement for RMTL analysis in competing risks settings.
• These methods are applicable to practical scenarios, including those with data ties.
• The study provides a powerful framework for analyzing complex survival data, as illustrated in a leukemia patient example.