Clinical and Laboratory Standards Institute document EP05-A3—Evaluation of Precision of Quantitative Measurement Procedures; Approved Guideline—Third Edition provides guidance for evaluating the precision of in vitro diagnostic quantitative measurement experimental designs and includes recommendations for establishing precision performance. Included are guidelines for duration, experimental designs, materials, data analysis, summarization, and interpretation—techniques adaptable for a wide spectrum of measurands and system complexity. These guidelines are intended for manufacturers or developers of clinical laboratory measurement procedures, and for users who wish to determine their own performance characteristics. A balance is created in the document between complexity of design and analysis, and simplicity of operation.

The third edition narrows the scope of EP05 by limiting its discussion of single-site experimental designs to procedures suitable for establishing or validating precision performance characteristics. Accordingly, EP05 is now addressed primarily to manufacturers and developers. Recommendations for end-user laboratories for verifying repeatability and within-laboratory precision claims can be found in CLSI document EP151. The precision verification protocol in that guideline has been tailored for compatibility with EP05’s single-site study designs. 

The single-site protocol familiar from previous editions of EP05—calling for measurements on 20 days, with two runs per day and two replicates per run for a given sample, reagent lot, etc.—is retained in this third edition as a standardized experiment for use by manufacturers and developers in evaluating the repeatability and within-laboratory (within-device) precision of a measurement procedure (or “assay”). 

No matter how these performance characteristics are established, it is important that the assessments be verifiable, and that they characterize precision over a substantial period of time and across most of the assay’s stated measuring interval. The single-site experimental designs described in EP05 meet these requirements (see Chapter 3). It is expected that the original “20 × 2 × 2” design will continue to serve well for the great majority of quantitative assays used in clinical laboratories. However, extensive guidance was added on how to optimize this design for a given assay in light of its sources of variation and their relative magnitudes and interrelationships (see Chapter 2). 

Moreover, in recognition of the wide diversity of quantitative devices in use today, which differ in character and complexity, variants of the 20 × 2 × 2 design are also discussed. Appendix C is devoted to advanced models—multifactor designs—for use when a two-factor design lacks the ability to do justice to the major sources of variation affecting an assay’s within-laboratory precision. Depending on the assay, some of these models should also prove useful to manufacturers for the insights they can yield both during assay development and optimization and after the assay enters routine production. 

New to EP05 is a second standardized experiment: a multisite protocol calling, minimally, for repeated measurements at three sites on five days. Both 3 (sites) × 5 (days) × 5 (replicates per day) and 3 (sites) × 5 (days) × 2 (runs per day) × 3 (replicates per run) implementations are described (see Chapter 4). This ancillary protocol addresses site-to-site variability and estimates of reproducibility. It has been tailored for suitability in the context of validating a new assay, when such a study may be required due to the assay’s character and/or to regulatory demands.

To help foster understanding of basic concepts, the new edition includes an extensive tutorial for the nonstatistician (see Section 1.5). Numerical examples illustrating a single-site 20 × 2 × 2 study and a complete multisite 3 × 5 × 5 study are presented in the appendixes. 

Due to the complex nature of the calculations in this guideline, it is recommended that the user have access to a computer and statistical software, such as StatisPro™ method evaluation software from CLSI.

This document provides guidance for studies intended to establish the within-site precision performance characteristics of quantitative measurement procedures used in clinical laboratories, and also for studies addressing site-to-site variability. Multiple experimental protocols are described, along with considerations on how to select and optimize the protocol(s) best suited for a specific measurement procedure (or “assay”) and its intended use.