A Simple Approach for Sample Size Calculation for Comparing Two Concordance Correlation Coefficients Estimated on the...

Area of Science:

• Biostatistics
• Medical Sciences
• Pharmaceutical Research

Background:

• Assessing agreement between raters and instruments is crucial in medical and pharmaceutical research.
• The concordance correlation coefficient (CCC) is a standard measure for continuous variable agreement.
• Comparing agreement across multiple assessment methods requires robust statistical approaches.

Purpose of the Study:

• To develop a method for determining the sample size required to test the equality of two concordance correlation coefficients (CCCs).
• To address the challenge of correlated CCC estimates when using the same sample for multiple assessment methods.
• To provide a computationally efficient and extensible approach for sample size calculation.

Main Methods:

• Simulating a large dataset based on the joint distribution of pairwise ratings for two assessment methods.
• Employing Taylor series linearization to create two new random variables reflecting the variance-covariance matrix of dependent CCC estimates.
• Utilizing the simulated data to calculate the necessary sample size for hypothesis testing (H0: CCC1 = CCC2 vs. HA: CCC1 ≠ CCC2).

Main Results:

• The proposed simulation-based approach provides a method for calculating sample sizes for comparing two CCCs.
• The technique accounts for the dependency between CCC estimates arising from using the same sample.
• The method demonstrates minimal computing time and adaptability for comparing more than two CCCs or Kappa statistics.

Conclusions:

• This study presents an effective and efficient method for sample size calculation in agreement studies involving continuous variables.
• The approach is valuable for researchers in medical and pharmaceutical fields needing to compare the performance of assessment methods.
• The methodology is extendable, offering a flexible tool for various agreement analysis scenarios.