Association within twin pairs for a dichotomous trait

Area of Science:

• * Biostatistics
• * Genetic Epidemiology
• * Statistical Genetics

Background:

• * Traditional measures of twin association for dichotomous traits can be complex to analyze.
• * Existing methods may require arbitrary choices (e.g., index twin) or estimation of nuisance parameters.
• * There is a need for simpler, more robust statistical measures in twin studies.

Purpose of the Study:

• * To propose a novel odds-ratio measure for assessing twin association with dichotomous traits.
• * To develop straightforward methods for estimating association tests and confidence intervals.
• * To provide robust statistical tests for homogeneity of association in different twin types (monozygotic and dizygotic).

Main Methods:

• * Development of an odds-ratio measure for twin association.
• * Estimation of the odds ratio without specifying an index twin or nuisance parameters.
• * Proposal of large-sample and exact tests for homogeneity of association across twin zygosities.
• * Introduction of a log-linear parameterization for complex data modeling.

Main Results:

• * The proposed odds-ratio measure simplifies the estimation of twin association for dichotomous traits.
• * Association tests and confidence intervals are easily computed using this odds-ratio parameterization.
• * Both large-sample and exact tests for homogeneity of association are presented, with the exact test recommended for small cell counts.
• * A log-linear model is introduced for more intricate genetic epidemiology data.

Conclusions:

• * The odds-ratio measure offers a more accessible and computationally simpler approach to twin association analysis.
• * The developed homogeneity tests provide reliable methods for comparing association across monozygotic and dizygotic twins.
• * The log-linear parameterization enhances the flexibility of statistical modeling in twin-based genetic studies.