Measuring agreement in method comparison studies with heteroscedastic measurements.

Area of Science:

• Biostatistics
• Health Sciences
• Measurement Science

Background:

• Method comparison studies are crucial in health disciplines.
• Assessing agreement between measurement methods is vital for data reliability.
• Variability often changes with the magnitude of continuous variables, complicating analysis.

Purpose of the Study:

• To propose a methodology for evaluating agreement between two measurement methods for continuous variables with magnitude-dependent variability.
• To extend existing agreement evaluation techniques to heteroscedastic data.
• To address computational challenges in modeling such data.

Main Methods:

• Utilized a heteroscedastic mixed-effects model with true measurement as a variance covariate.
• Developed four estimation methods for approximate maximum likelihood estimates.
• Extended homoscedastic agreement evaluation methodology to heteroscedastic models.
• Applied to cholesterol measurement data.

Main Results:

• The proposed methodology effectively evaluates agreement in heteroscedastic conditions.
• Four novel estimation techniques were introduced to handle model fitting challenges.
• Simulations demonstrated the reliability of inference procedures for moderately large samples.
• The method was successfully illustrated using cholesterol measurement data.

Conclusions:

• The developed methodology provides a robust framework for heteroscedastic method comparison studies.
• The proposed estimation methods offer practical solutions for fitting complex models.
• This work enhances the statistical rigor in evaluating measurement agreement in health-related research.