Concordance rate of a four-quadrant plot for repeated measurements

Area of Science:

• Biostatistics
• Clinical Measurement
• Medical Device Evaluation

Background:

• Four-quadrant plots and concordance rates are used to assess clinical measurement method equivalence.
• Conventional methods may not adequately account for within-subject data correlation.
• Bland-Altman analysis is also a common tool for method comparison.

Purpose of the Study:

• To develop a novel concordance rate for the four-quadrant plot.
• To address the limitations of conventional concordance rates by incorporating within-subject covariance.
• To enhance the evaluation of equivalence between new and standard clinical measurement methods.

Main Methods:

• Proposed a new concordance rate based on multivariate normal distribution.
• Formulated the rate as a conditional probability of agreement.
• Included a parameter for defining the minimum concordant number, allowing flexible interpretation.

Main Results:

• Numerical simulations showed the proposed method achieved an AUC of 0.967 versus 0.938 for the conventional rate.
• Application to a real case yielded an AUC of 0.999 for the proposed method compared to 0.964 for the conventional rate.
• The proposed concordance rate demonstrated superior accuracy and diagnostic capability.

Conclusions:

• The proposed concordance rate offers improved accuracy and diagnosability for clinical measurement method comparison.
• Accounting for within-subject covariance enhances the reliability of equivalence assessments.
• This novel approach provides a more robust tool for validating new clinical measurement techniques.