Survival Analysis with Multiple Causes of Death: Extending the Competing Risks Model

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Traditional mortality statistics rely on a single underlying cause of death, which is insufficient for complex diseases in aging populations.
• The limitations of single-cause mortality analysis are increasingly apparent with the rise of chronic and degenerative diseases.
• Existing research acknowledges multiple causes of death but lacks a formal statistical modeling framework.

Purpose of the Study:

• To propose and validate a formal statistical model for analyzing multiple cause of death data.
• To extend the competing risks model for conceptualizing single-cause mortality to a multiple-cause framework.
• To provide a method for studying the burden and etiology of mortality related to each disease using death certificate data.

Main Methods:

• Development of an empirical approach that assigns weights to each cause listed on a death certificate.
• Extension of the competing risks model to accommodate multiple causes of death.
• Application of Cox regression methodology for analyzing weighted multiple-cause mortality data.
• Comparison of multiple-cause, single-cause, and "any-mention" approaches.

Main Results:

• The proposed weighted model provides a formal framework for multiple-cause mortality analysis.
• Simulation studies and an application to socioeconomic inequalities in mortality demonstrate the model's value.
• The method offers new insights into disease burden and etiology, particularly for specific conditions.

Conclusions:

• The proposed multiple-cause mortality model offers a significant advancement over traditional single-cause methods.
• This framework effectively utilizes comprehensive death certificate data to reveal complex mortality patterns.
• The approach is valuable for epidemiological research, particularly in understanding disease burden and socioeconomic disparities in mortality.