Indirect methods for the verification of reference intervals in laboratory medicine

Area of Science:

• Clinical Chemistry
• Laboratory Medicine
• Biostatistics

Background:

• Accreditation standards mandate periodic verification of laboratory reference intervals against local conditions.
• Current CLSI/IFCC C28-A3 guidelines recommend a 20-sample binomial procedure, which has demonstrated inadequate statistical power and sensitivity.
• Existing verification methods offer a false sense of compliance, necessitating a reassessment of current practices.

Purpose of the Study:

• To critically evaluate the limitations of the current 20-sample binomial reference interval verification approach.
• To systematically review methodological advancements in reference interval verification since 2018.
• To advocate for the adoption of indirect statistical methods for more robust and reliable verification.

Main Methods:

• A structured literature search identified 22 relevant publications, supplemented by additional references for comprehensive coverage.
• Evaluation of indirect statistical methods, including open-source R packages (reflimR, refineR) and machine learning techniques.
• Assessment of quantitative acceptance criteria based on equivalence limits (EL) and uncertainty margins (UM).

Main Results:

• Indirect estimation methods, applied to routine laboratory data, demonstrate superior performance compared to the binomial procedure.
• Independent multi-center and multi-method studies confirm the robust performance and comparability of indirect methods across various analytes.
• Machine learning techniques effectively extend verification capabilities to complex datasets, including those with high pathological prevalence.

Conclusions:

• Indirect verification methods have achieved sufficient maturity for guideline endorsement, offering a statistically defensible standard.
• A proposed stepwise workflow integrates reflimR, refineR, and machine learning for efficient and accurate routine laboratory verification.
• New web-based tools and platforms enhance accessibility, enabling practical implementation of these advanced verification strategies.