Concordance correlation coefficients estimated by generalized estimating equations and variance components for...

Area of Science:

• Biostatistics
• Longitudinal Data Analysis
• Agreement Measurement

Background:

• The concordance correlation coefficient (CCC) is standard for assessing agreement between two observers.
• Existing methods may not fully account for dependencies in repeated longitudinal measurements.
• Simultaneous assessment of intra-, inter-, and total observer agreement is needed.

Purpose of the Study:

• To propose and evaluate novel statistical approaches for measuring observer agreement in longitudinal studies.
• To extend agreement assessment to include intra-observer, inter-observer, and total agreement simultaneously.
• To compare the performance of proposed methods using simulation studies and a real-world application.

Main Methods:

• Development of a generalized estimating equations (GEE) approach to handle dependent repeated measurements.
• Formulation of variance components (VC) indices using an extended three-way linear mixed model (LMM).
• Comparison of GEE and VC approaches via simulation studies on longitudinal data.

Main Results:

• The GEE approach demonstrated satisfactory performance with small mean square errors and nominal coverage rates for large datasets.
• The VC approach performed outstandingly well across all sample sizes when its underlying assumptions were met.
• Both methods provide valuable tools for assessing different facets of observer agreement in longitudinal studies.

Conclusions:

• The GEE approach offers flexibility in model assumptions and correlation structures for repeated measurements.
• The VC approach, under specific assumptions, provides excellent agreement assessment for all sample sizes.
• These methods enhance the analysis of observer agreement in complex longitudinal study designs.