A New Concept for Reference Change Values-Regression to the Population Mean.

Area of Science:

• Clinical Chemistry
• Biostatistics
• Medical Diagnostics

Background:

• Reference change values (RCVs) traditionally assess analyte concentration changes relative to a single prior measurement (X1).
• Existing RCV theory does not fully account for the influence of X1's position within the reference interval.
• Investigating an alternative RCV model based on X1's location within the reference interval.

Purpose of the Study:

• To determine if the position of the initial measurement (X1) within the reference interval affects the subsequent reference change value (RCV).
• To develop a predictive model for RCV that incorporates the location of X1 within the population's reference interval.
• To refine the understanding of biological variation and measurement error in clinical laboratory testing.

Main Methods:

• Analysis of serum sodium, calcium, and total protein data from European Biological Variation study and routine collections.
• Statistical modeling to describe the effect of X1's position within the reference interval on subsequent results.
• Development of an equation to predict population-based RCVs considering X1's location.

Main Results:

• RCV midpoints were significantly dependent on the position of X1 within the reference interval across all datasets.
• Initial results (X1) below the population mean were more likely followed by higher subsequent results.
• Initial results (X1) above the population mean were more likely followed by lower subsequent results.
• A model incorporating population mean, reference interval dispersion, and diagnostic variation accurately predicted observed changes.

Conclusions:

• The position of the initial measurement (X1) within the reference interval leads to asymmetrical RCVs.
• This asymmetry can be explained by a regression towards the population mean.
• Incorporating this regression concept into RCV theory is crucial for accurate clinical interpretation.