[Derivation of latent reference intervals from routine laboratory database: its principle, validity and...

Area of Science:

• Clinical Chemistry
• Biostatistics
• Laboratory Medicine

Context:

• Establishing accurate reference intervals (RI) is crucial for interpreting laboratory test results.
• Traditional RI determination often relies on costly and time-consuming studies of healthy individuals.
• Routine laboratory databases contain vast amounts of data that could potentially be leveraged.

Purpose:

• To develop and validate a new method for deriving reference intervals (RI) directly from routine laboratory database intervals.
• To utilize 'latent reference values' by excluding data with multiple abnormalities in related tests.
• To compare the validity of these newly derived latent reference intervals (LRI) against conventionally determined RIs and other validation datasets.

Summary:

• A novel parametric method using a modified Box-Cox transformation and maximum likelihood estimation was developed to derive RI from routine laboratory data.
• The method iteratively refines RI estimates by excluding data with multiple abnormalities, creating 'latent reference values'.
• The derived latent reference intervals (LRI) were validated against RIs from healthy individuals, blood donors, and a simulated healthy population, demonstrating comparable validity.

Impact:

• Provides a cost-effective and efficient method for establishing reference intervals using existing laboratory data.
• Enhances the utility of routine laboratory databases for clinical decision-making.
• Offers a robust alternative for RI determination, particularly when traditional methods are challenging to implement.