A note on issues in meta-analysis for behavioral genetic studies using categorical phenotypes.

Area of Science:

• Behavioral genetics
• Quantitative genetics
• Statistical genetics

Background:

• Meta-analysis in behavioral genetics aims to refine parameter estimates and resolve conflicting findings.
• Categorical phenotypes present unique challenges for meta-analytic approaches.

Purpose of the Study:

• To investigate issues in meta-analyzing categorical phenotypes using simulated data.
• To evaluate the performance of various summary statistics under different population prevalence.
• To assess the impact of ascertainment correction misspecification on parameter estimates.

Main Methods:

• Simulated data from a multifactorial threshold model with a normal liability distribution.
• Comparison of summary statistics (probandwise concordance rate, recurrence risk ratio, odds ratio, kappa).
• Examination of biases in genetic and environmental parameter estimates due to misspecified ascertainment models.

Main Results:

• Odds ratio and kappa statistics performed well at moderate to high prevalence (15-50%), but all statistics were sensitive to low prevalence.
• Misspecification of ascertainment models led to substantial biases in genetic and environmental parameter estimates.
• Biases varied depending on research design (twin vs. adoption data) and simulated etiological factors.

Conclusions:

• Direct estimation of parameters from summary statistics is preferable when the multifactorial threshold model assumption holds.
• Sensitivity to population prevalence is a critical consideration for categorical phenotypes.
• Careful attention to ascertainment correction is crucial to avoid biased estimates in meta-analyses of behavioral genetic data.