Testing equality of relative survival patterns based on aggregated data

Area of Science:

• * Biostatistics and Survival Analysis
• * Cancer Epidemiology and Outcomes Research

Background:

• * Relative survival rate is a key metric for assessing cancer patient outcomes, adjusting for general population mortality.
• * Existing methods for comparing relative survival rates have limitations in scope and applicability.
• * Analyzing survival data requires robust statistical tools to account for competing risks.

Purpose of the Study:

• * To develop and evaluate novel maximum likelihood ratio tests for comparing relative survival rates between patient groups.
• * To assess the performance of these new tests against existing statistical methods.
• * To extend the application of relative survival rates within proportional hazards regression models.

Main Methods:

• * Construction of maximum likelihood ratio tests using aggregated data.
• * Testing equality of relative survival rates against proportional hazards and general alternatives.
• * Application to Finnish nationwide colon cancer patient data.
• * Simulation studies to compare performance with alternative methods.

Main Results:

• * The proposed maximum likelihood ratio tests demonstrate favorable comparison with previously suggested methods.
• * These new tests offer more extensive coverage and applicable alternative hypotheses.
• * Simulation results support the superiority of the maximum likelihood ratio tests.

Conclusions:

• * Maximum likelihood ratio tests provide a powerful and versatile tool for analyzing cancer patient survival data.
• * The developed methods enhance the statistical rigor in comparing relative survival rates.
• * Future work includes extending these methods to proportional hazards regression models for comprehensive survival analysis.