Result Interpretation in Skewed Populations-Are Common Reference Intervals Inadequate?

Area of Science:

• Clinical Chemistry
• Biomarker Analysis
• Laboratory Medicine

Background:

• Reference intervals (RIs) define the central 95% of healthy population values.
• Many common biomarkers exhibit skewed, not normal, result distributions.
• This necessitates careful interpretation of laboratory results.

Purpose of the Study:

• To demonstrate interpretive variations arising from Gaussian versus skewed reference interval distributions.
• To evaluate the impact of distribution type on biomarker interpretation.
• To explore the utility of interpretive aids for skewed data.

Main Methods:

• Analyzed de-identified patient data for sodium (Gaussian) and total testosterone (skewed).
• Utilized the refineR package to obtain RIs and Box-Cox parameters.
• Evaluated results using RI midpoint, median, IQR, z scores, and cumulative percentiles.

Main Results:

• Sodium RI midpoint and median were similar, reflecting a Gaussian distribution.
• Total testosterone showed a skewed distribution, with the RI midpoint closer to the 75th percentile.
• A skewed testosterone result was initially misinterpreted as "borderline low" without z score/percentile context.

Conclusions:

• Reference intervals aid interpretation against a "disease-free" population.
• Skewed biomarker distributions can complicate interpretation if clinicians are unaware of distribution characteristics.
• Interpretive aids, like z scores and cumulative percentiles, can enhance comprehension and prevent misdiagnosis.