On estimating the time to statistical cure

Area of Science:

• Biostatistics
• Survival Analysis
• Cancer Epidemiology

Background:

• Cancer patients often have elevated mortality risk post-diagnosis compared to the general population.
• For certain cancers, this risk converges to general population levels over time.
• The 'cure point' signifies this convergence and is critical for patient and healthcare planning.

Purpose of the Study:

• To review existing cure point estimation methods.
• To introduce new, clinically relevant measures for quantifying excess mortality in cancer patients.
• To assess the performance of these methods in estimating the cure point.

Main Methods:

• Review of statistical methodologies for cure point estimation.
• Development and application of novel measures for excess mortality.
• Simulation studies to evaluate method performance.
• Illustration using survival data from Danish colon cancer patients.

Main Results:

• Estimated cure point bias is influenced by the chosen mortality measure, margin of clinical relevance, and estimation procedure.
• Interdependencies exist between mortality measure selection, defining clinical relevance, and computation accuracy.
• Confidence intervals for cure points can be wide, potentially limiting clinical applicability.

Conclusions:

• Cure point estimation, while clinically valuable, presents significant challenges.
• Practitioners must be aware of the numerous choices and potential pitfalls in the estimation process.