A hybrid parametric and empirical likelihood model for evaluating interactions in case-control Studies

Area of Science:

• Biostatistics
• Epidemiology
• Statistical Modeling

Background:

• Case-control studies are effective for collecting covariate data based on disease status.
• Standard logistic regression may lack efficiency for modeling covariate interactions under specific constraints.
• Existing methods may not be optimal for analyzing interactions between various covariate types.

Purpose of the Study:

• To develop a more efficient hybrid statistical approach for inferring interactions between two covariates in case-control designs.
• To address limitations of prospective logistic regression when covariate distribution constraints exist.
• To provide a flexible method applicable to discrete and continuous covariates.

Main Methods:

• A hybrid approach combining parametric and nonparametric modeling for covariate distributions.
• Utilizing a parametric model for one covariate's conditional distribution given another in controls.
• Employing a maximum hybrid parametric and empirical likelihood method for parameter estimation.

Main Results:

• Developed a semiparametric model for evaluating interactions between diverse covariate types.
• Established asymptotic properties for the proposed estimators and test statistics.
• Demonstrated the method's advantages over existing approaches via simulations and a real data example.

Conclusions:

• The proposed hybrid method offers an efficient and flexible approach for analyzing covariate interactions in case-control studies.
• The method is suitable for various covariate types and provides robust statistical inference.
• This semiparametric model advances statistical analysis in epidemiological research.