A global odds ratio regression model for bivariate ordered categorical data from ophthalmologic studies

Area of Science:

• Ophthalmology
• Biostatistics
• Epidemiology

Background:

• Bilateral eye diseases often present correlated outcome data in clinical and epidemiological studies.
• Analyzing paired ordered categorical responses requires specialized statistical approaches.
• Identifying risk factors for diseases like diabetic retinopathy is crucial for public health.

Purpose of the Study:

• To propose a novel latent variable regression model for bivariate ordered categorical data.
• To effectively model the dependency between fellow eyes using the global odds ratio.
• To identify risk factors associated with diabetic retinopathy in younger-onset diabetics.

Main Methods:

• Developed a latent variable regression model for bivariate ordered categorical outcomes.
• Utilized Plackett's cross ratio distribution for the joint distribution of latent variables.
• Applied the model to data from the Wisconsin Epidemiologic Study of Diabetic Retinopathy.

Main Results:

• The proposed model successfully analyzes correlated bivariate ordered categorical data.
• The cross ratio distribution effectively models inter-eye dependency via the global odds ratio.
• Risk factors for diabetic retinopathy were identified in the study population.

Conclusions:

• The latent variable regression model offers a robust method for analyzing bilateral eye disease data.
• This approach enhances the understanding of disease associations and risk factors.
• The model has significant implications for clinical trials and epidemiological research in ophthalmology.