Guest Essay

A new edition of EP46 has been published. Thie guideline focuses on Determining allowable total error goals and aligns with the CLSI EP21 guideline on calculating total error. Dr. Paulo Pereira, heavily involved in the creation of these documents, provides us a quick introduction.

Introduction to the new edition of CLSI EP46: Determining allowable total error goals and limits for quantitative medical laboratory measurement procedures

Dr. Paulo Pereira
June 2025

Fitness for purpose

This new edition of CLSI EP46 provides essential guidance for defining, understanding, and applying allowable total error (ATE) goals and limits for quantitative medical laboratory measurement procedures (1). With laboratory testing playing a critical role in clinical decision-making worldwide, it is essential to set realistic yet clinically meaningful performance expectations that protect patient safety while supporting innovation in testing technologies.

Total analytical error (TAE) represents the combined impact of both random errors (imprecision) and systematic errors (bias) that can occur during laboratory testing. CLSI EP46 is fully aligned with CLSI EP21- Estimation of Total Analytical Error for Quantitative Medical Laboratory Measurement Procedures (2), which provides the practical framework for estimating TAE in quantitative measurement procedures. TAE is not just a theoretical concept - it is the key evaluation standard used by medical laboratory professionals, manufacturers of in vitro diagnostic medical devices (IVD-MDs), and regulatory agencies alike. By comparing the estimated TAE of a measurement procedure against established ATE goals or limits, stakeholders can objectively assess whether a test is suitable for its intended use and meets the clinical performance required for safe, effective patient care.

This guideline was developed through the CLSI consensus process by a diverse group of global experts - medical laboratory professionals, IVD manufacturers, regulators, and researchers - who serve on the Document Development Committees (DDC) responsible for both EP21 and EP46. Their collective expertise has shaped this comprehensive update to ensure the guidance reflects current scientific knowledge, practical realities of laboratory medicine, and the evolving needs of healthcare systems.

Parametric and non-parametric approaches to TAE

EP46 includes two well-established approaches for calculating TAE: the Westgard Parametric Approach and the CLSI EP21 Non-Parametric Approach. They are primarily used by clinical laboratories and IVD-MD manufacturers, respectively. The Westgard approach, introduced in 1974, represents a parametric model for estimating Total Analytical Error by combining systematic and random components of error into a single quantitative expression (3). The mathematical model for TAE is: TAE = |Bias| + z × SD_WL

Where bias is the systematic error (deviation from the true value), SD_WL is the within-laboratory imprecision (random error), z is the z-score corresponding to a desired confidence interval (typically 1.96 for 95% two-sided or 1.65 for 95% one-sided).

This parametric approach assumes a normal (Gaussian) distribution of analytical errors, uses independently estimated bias and imprecision values to calculate TAE, expresses TAE as a defined interval of expected analytical performance, typically capturing 95% of results, and provides a practical, mathematically simple method, widely used in laboratory quality assessment and Six Sigma applications (4).

Otherwise, the Non-Parametric Approach to TAE, as described in the guideline EP21, represents a non-parametric, empirical method for estimating TAE. Unlike parametric models that rely on normality assumptions, EP21 uses empirical data from patient specimens to directly estimate TAE, incorporating both random (imprecision) and systematic (bias) errors, as well as other sources such as matrix effects or non-linearity. The approach involves comparing the results of a candidate measurement procedure to those of a comparator measurement procedure, ideally a reference method or a method traceable to a higher-order reference. Differences between paired results are calculated across a representative sample of patients, and the TAE is expressed as the interval that contains a specified proportion of those differences, typically the central 95%.

This method does not separately quantify bias and imprecision but instead captures the combined effect of all relevant analytical error sources under real-world conditions. It is particularly useful for assessing whether a test meets pre-established ATE limits, which are set based on CLSI EP46 guidance, clinical requirements, and medical decision levels.

ATE "goals" and "limits" significance and Milan alignment

Several key updates and improvements distinguish this edition. The guideline clearly differentiates between "goals" and "limits" for ATE:

• Goals represent ideal, aspirational levels of analytical performance, guiding innovation and improvement.
• Limits define the minimum acceptable performance levels required to ensure tests can be safely and reliably used in practice.

Importantly, this distinction between "goals" and "limits" is also fully aligned with the recommendations of the European Federation of Clinical Chemistry and Laboratory Medicine (EFLM). The EFLM Strategic Conference in Milan emphasized the need to establish ambitious, clinically-driven goals for analytical performance, while also recognizing the practical role of defining acceptable performance limits achievable with current technologies (5).

Multiple, complementary approaches for determining ATE goals and limits are presented aligned with the Milan overview, including:

• Clinical outcome-based models
• Biological variation data (6,7)
• Peer-group and state-of-the-art performance

Frameworks for applying Total Error concepts

The framework for understanding error has been expanded, with detailed discussions on:

• Classifying errors as random or systematic
• Identifying error sources across the preanalytical, analytical, and postanalytical phases
• Recognizing how error contributes to TAE and impacts patient care

Integrating Total Error, measurement uncertainty, and Sigma Metrics for comprehensive performance assessment

The guideline introduces comparisons between TAE, measurement uncertainty (MU) (8), and the Sigma metric (7), enabling users to integrate these complementary concepts in performance assessment. New content on "error budgeting" helps users identify, quantify, and manage sources of error systematically. The concept of "error grid analysis" is introduced as a practical, visual tool to evaluate the acceptability of test performance in a clinical context.

Conclusion

The publication of this edition of CLSI EP46 enforces the critical importance of having a dedicated, in-depth resource focused specifically on TAE and ATE. While TAE has been a central performance concept for decades, the growing complexity of laboratory-developed tests (LDTs), point-of-care technologies (POCT), and IVD-MDs requires more sophisticated tools for setting performance standards that are both clinically meaningful and achievable with current technologies (9).
By providing this expanded, consensus-based guidance, EP46 supports all stakeholders - from laboratory professionals to IVD manufacturers to regulatory authorities - in applying total error concepts rigorously and consistently. This, in turn, helps ensure that laboratory tests not only meet technical specifications but also provide trustworthy results that clinicians can rely on for diagnosis, monitoring, and treatment decisions.

The global laboratory community's collective input into this document highlights its relevance, credibility, and practical utility. Ultimately, EP46 is a critical tool to advance quality in laboratory testing, promote patient safety, and support global harmonization in medical laboratory standards.

Declaration of Interest: Dr. Paulo Pereira is a member of the CLSI DDC on Total Analytical Error (EP21/EP46) and served as a member of the EP46 and EP21 writing teams.

References

1. Clinical and Laboratory Standards Institute (CLSI). EP46 - Determining Allowable Total Error Goals and Limits for Quantitative Medical Laboratory Measurement Procedures. 1st ed. CLSI; 2025.
2. Clinical and Laboratory Standards Institute (CLSI). EP21 - Estimation of Total Analytical Error for Quantitative Medical Laboratory Measurement Procedures. 2nd ed. CLSI; 2016.
3. Westgard JO, Carey RN, Wold S. Criteria for judging precision and accuracy in method development and evaluation. Clin Chem. 1974;20(7):825–833. doi: 10.1093/clinchem/20.7.825
4. Westgard JO, Groth T. Design and evaluation of statistical control procedures: Applications of a computer "Quality Control Simulator" program. Clinical Chemistry. 1981;27:1536-1545
5. Sandberg S, Fraser CG, Horvath AR, et al. Defining analytical performance specifications: Consensus Statement from the 1st Strategic Conference of the European Federation of Clinical Chemistry and Laboratory Medicine. Clin Chem Lab Med. 2015;53(6):833-835. doi:10.1515/cclm-2015-0067
6. Fraser CG. Biological Variation: From Principles to Practice. Washington, DC: AACC Press; 2001.
7. EFLM Working Group on Biological Variation, EFLM Task Group on Biological Variation Database. EFLM Biological Variation Database [Internet]. European Federation of Clinical Chemistry and Laboratory Medicine; 2019 [cited 2025 Jun 23]. Available from: https://biologicalvariation.eu
8. JCGM 100:2008. Evaluation of Measurement Data - Guide to the Expression of Uncertainty in Measurement (GUM). Joint Committee for Guides in Metrology; 2008. Available from: https://www.bipm.org/en/publications/guides/gum.html
9. International Organization for Standardization. ISO 15189:2022 Medical Laboratories - Requirements for Quality and Competence. 4th ed. Geneva, Switzerland: ISO; 2022.