A method for analyzing disease-specific mortality with missing cause of death information

Area of Science:

• Biostatistics
• Survival Analysis
• Epidemiology

Background:

• Analyzing competing risks data is complex when the cause of failure is unknown.
• Existing methods like log-rank subtraction lack systematic study for incomplete failure types.
• Intermediate events are crucial as they precede the cause of interest.

Purpose of the Study:

• To systematically investigate the statistical properties of log-rank subtraction for competing risks data.
• To propose a modified, unbiased method for comparing disease-specific mortality.
• To address biases in analyzing incomplete failure type data.

Main Methods:

• Investigated statistical properties of log-rank subtraction.
• Developed a modified test using a weighted log-rank score statistic.
• Derived asymptotic properties of the proposed test procedure.
• Conducted simulation studies to evaluate performance.

Main Results:

• The proposed weighted log-rank test is unbiased.
• The method demonstrates reasonable statistical power.
• Simulation results validate the effectiveness of the modified approach.
• The method was illustrated using breast cancer study data.

Conclusions:

• The modified weighted log-rank test provides a robust approach for analyzing competing risks data with intermediate events and missing failure types.
• This method enhances the accuracy of disease-specific mortality comparisons.
• The findings have implications for clinical trial analysis and epidemiological studies.