Conditional likelihood inference in a case- cohort design: an application to haplotype analysis.

Area of Science:

• Biostatistics
• Genetic Epidemiology
• Statistical Genetics

Background:

• Case-cohort designs efficiently study rare outcomes by selecting a subcohort and cases from the full cohort.
• Analyzing case-cohort data requires specialized statistical methods to account for the complex sampling design.
• Accurate estimation of population parameters is crucial for understanding disease etiology and genetic associations.

Purpose of the Study:

• To develop and validate a conditional likelihood approach for statistical analysis of case-cohort studies.
• To enable joint estimation of genetic association parameters and population haplotype frequencies.
• To apply the method to single nucleotide polymorphism (SNP) genotype data.

Main Methods:

• Utilized conditional likelihood inference, conditioning on the ascertainment into the second-stage study group.
• Developed likelihood expressions for categorical and time-to-event response models.
• Incorporated the probability of ascertainment within the likelihood framework.

Main Results:

• Demonstrated that conditional likelihood inference yields valid estimation of population parameters.
• Successfully applied the method for joint estimation of haplotype-event association parameters and population haplotype frequencies.
• Validated the approach using SNP genotype data from a case-cohort design.

Conclusions:

• The proposed conditional likelihood method provides a robust framework for analyzing case-cohort studies.
• This approach facilitates accurate estimation of genetic associations and population frequencies, crucial for genetic epidemiology.
• The method is applicable to various response types and SNP genotype data, enhancing its utility.