Multivariate methods in ophthalmology with application to other paired-data situations.

Area of Science:

• Ophthalmology
• Biostatistics
• Genetics

Background:

• Ophthalmologic data often exhibits intraclass correlation due to paired eyes.
• Standard regression methods may not adequately account for this correlation.
• Nested and matched data structures are common in genetic and clinical studies.

Purpose of the Study:

• To present statistical methods for analyzing ophthalmologic data with correlated outcomes.
• To extend these methods to nested and matched data structures.
• To apply these methods to understand factors influencing vision in retinitis pigmentosa.

Main Methods:

• Multiple regression and multiple logistic regression analyses were employed.
• Methods specifically addressed intraclass correlation between eyes.
• Techniques were adapted for nested data and matched studies with variable ratios.

Main Results:

• The methods successfully analyzed ophthalmologic data with correlated outcomes.
• Application to retinitis pigmentosa patients revealed relationships between genetic type and visual parameters.
• Age, sex, and cataract presence were controlled for in the analyses.

Conclusions:

• The presented statistical methods are effective for analyzing correlated ophthalmologic data.
• These methods enhance the understanding of genetic influences on visual outcomes.
• The approach is applicable to various nested and matched study designs.