Revisiting the Extrapolated Reference Values (E-Ref) Estimation

Area of Science:

• Clinical Chemistry
• Laboratory Medicine
• Biostatistics

Background:

• Extrapolated reference values (E-Ref) offer a method for deriving reference values (RVs) from datasets containing both normal and abnormal results.
• Laboratories can utilize E-Ref to establish their own specific RVs.
• This methodology has undergone enhancements since its initial description.

Purpose of the Study:

• To present the current version of the E-Ref algorithm.
• To describe the interpretation of results generated by the E-Ref method.
• To address the growing interest in E-Ref methodology.

Main Methods:

• The cumulative distribution function (CDF) is central to the E-Ref methodology.
• Computer simulations generated normal and mixed datasets for conduction velocity.
• The E-Ref algorithm was refined based on CDF analysis from simulations.
• Bootstrap analysis assessed estimate variability for RV finalization.
• Median motor nerve conduction parameters from patient studies were analyzed.

Main Results:

• The CDF exhibits a plateau containing most normal data, even in mixed datasets.
• The mean of the plateau data approximates the mean of normal measurements.
• The plateau's range approximated two standard deviations (SDs).

Conclusions:

• The revised algorithm categorizes findings into definitely normal, definitely abnormal, and borderline ranges.
• This aligns with clinical practice, acknowledging that abnormalities are not always absolute.
• Bootstrap analysis is crucial for reliable estimates and assessing sample size adequacy.