Analysis of cohort studies with multivariate and partially observed disease classification data

Area of Science:

• Epidemiology
• Biostatistics
• Cancer Research

Background:

• Complex diseases like cancer are often classified into subtypes based on pathological and molecular traits.
• Accurate analysis of disease incidence in cohort studies is crucial for understanding disease progression and risk factors.

Purpose of the Study:

• To develop and validate statistical methods for analyzing disease incidence in cohort studies with multiple disease traits.
• To address challenges posed by missing trait data in competing-risk analyses.

Main Methods:

• A two-stage semiparametric Cox proportional hazards regression model was developed.
• An extended estimating equation approach was proposed for handling missing trait data in competing-risk scenarios.
• Asymptotic unbiasedness and a novel sandwich variance estimator were established for the proposed methods.

Main Results:

• The proposed methods allow for the examination of covariate effect heterogeneity across different disease trait levels.
• The estimating equation approach demonstrated asymptotic unbiasedness under missing-at-random assumptions.
• The methods were validated through simulation studies and a real-world application.

Conclusions:

• The developed statistical framework enhances the analysis of complex diseases by incorporating multiple traits and handling missing data effectively.
• This approach offers a robust tool for epidemiological research, particularly in cancer studies, to better understand disease subtypes and risk factors.