Related Experiment Videos
Regression models for twin studies: a critical review.
John B Carlin1, Lyle C Gurrin, Jonathan Ac Sterne
1Clinical Epidemiology and Biostatistics Unit, Murdoch Children's Research Institute, Royal Children's Hospital, Melbourne, Australia. jbcarlin@unimelb.edu.au
International Journal of Epidemiology
|August 10, 2005
Summary
Twin studies help separate genetic and environmental influences on health. Specialized regression models, accounting for twin data structure, are crucial for accurate analysis of these paired observations.
Area of Science:
- Epidemiology
- Biostatistics
- Genetics
Background:
- Twin studies are valuable for disentangling genetic and environmental factors in disease etiology.
- Life-course epidemiology highlights the importance of developmental influences on long-term health.
- Twins offer naturally matched pairs, controlling for numerous confounding factors.
Purpose of the Study:
- To review specialized regression methods for analyzing twin data.
- To explain the relationship between advanced methods and traditional difference analyses.
- To provide guidance on interpreting results from twin studies.
Main Methods:
- Review of generalized least squares and linear mixed models for twin data.
- Comparison with within-twin-pair difference analysis.
- Illustration using birth weight and cord blood erythropoietin data.
Main Results:
- Standard regression models are insufficient for correlated twin data.
- Specialized models are necessary to account for the paired structure.
- A general model with separate within- and between-pair effects is recommended.
Conclusions:
- Specialized regression techniques are essential for valid twin data analysis.
- A unified model enhances the interpretation of genetic and environmental effects.
- Guidelines are provided for robust analysis and interpretation of twin study findings.