Regression-based rectangular tolerance regions as reference regions in laboratory medicine.

Area of Science:

• Laboratory Medicine
• Biostatistics
• Clinical Chemistry

Background:

• Reference ranges are crucial for interpreting laboratory test results.
• Multivariate reference regions (MRRs) are needed for analyzing multiple analytes simultaneously.
• Traditional ellipsoidal MRRs cannot detect component-wise outliers.

Purpose of the Study:

• To develop methodologies for computing rectangular MRRs.
• To incorporate covariate information into these rectangular MRRs.
• To ensure the constructed regions possess the multiple use property.

Main Methods:

• Development of methodologies for rectangular MRRs.
• Incorporation of covariate information.
• Construction of reference regions using tolerance-based criteria.

Main Results:

• Proposed rectangular MRRs provide accurate coverage probabilities.
• The methods are robust to variations in sample size.
• Demonstrated application to the insulin-like growth factor system.

Conclusions:

• Rectangular MRRs offer an improvement over traditional ellipsoidal regions.
• The developed methods effectively incorporate covariates for enhanced diagnostic accuracy.
• The approach is applicable to real-world clinical laboratory data analysis.