Standard Document
Third Edition
Method Evaluation

CLSI EP09

Measurement Procedure Comparison and Bias Estimation Using Patient Samples

CLSI EP09 provides essential guidance for measurement procedure comparison and bias estimation using patient samples. Designed for both laboratorians and manufacturers, this guideline outlines procedures for determining bias between two measurement procedures and offers key considerations for experiment design and data analysis. It includes methods for visualizing data with scatter and difference plots, techniques for quantifying relationships between measurement procedures, and recommendations for bias evaluation and claims reporting.

 

June 20, 2018
Jeffrey R. Budd, PhD

{{FormatPrice(currentPrice)}}

Free

{{FormatPrice(nonMemberPrice)}} List Price
This is your member pricing.
Notify Me About New Editions
Abstract

Clinical and Laboratory Standards Institute guideline EP09—Measurement Procedure Comparison and Bias Estimation Using Patient Samples is written for laboratorians and manufacturers. It describes procedures for determining the bias between two measurement procedures, and it identifies factors for consideration when designing and analyzing a measurement procedure comparison experiment using patient samples. An overview of the measurement procedure comparison experiment includes considerations for both manufacturers and laboratorians. Details on how to create difference and scatter plots for visual inspection of the data are provided. Once the data are characterized, various methods are introduced for quantifying the relationship between two measurement procedures, including bias estimates and regression techniques. The final chapter contains recommendations for manufacturers’ evaluation of bias and statement format for bias claims.

Overview of Changes

This guideline replaces the previous edition of the approved guideline, EP09-A2-IR, published in 2010. Several changes were made in this edition, including: 

• Broader coverage of measurement procedure comparison applications 

• More reasons for comparisons based on patient samples (factor comparisons [eg, sample tube types]) 

• Visualization/exploration of data using difference plots 

• Regression descriptions including weighted options, Deming, and Passing-Bablok techniques • Measurement of bias using difference plots 

• Measurement of bias at clinical decision points 

• Computation of confidence intervals for all parameters 

• Outlier detection using extreme studentized deviate 

• Relocation of most of the detailed mathematical descriptions to the appendixes 

This guideline was corrected in 2018 and replaces the original third edition of the approved guideline, EP09-A3, published in 2013. Corrections were made as follows: 

• Reorganizing the content to emphasize the process of performing a measurement procedure comparison 

• Clearly specifying that manufacturers should use regression analysis to characterize bias 

• Adding information on using precision profile information in performing Deming regressions 

• Adding more information on determining confidence intervals for bias estimates at specified concentrations using regression fits 

• Making corrections to the description of the Passing-Bablok regression technique 

• Adding a detailed description of the bootstrap iterative technique for bias estimation 

• Correcting minor miscellaneous errors in equations

Scope

This guideline provides recommendations for designing an experiment and selecting methods to quantify systematic measurement error (bias or difference) between measurement procedures based on comparing patient samples. It provides both difference plot and regression procedures to determine the relationship between two measurement procedures either across their measuring intervals or at selected concentrations. Intended users of this guideline are manufacturers of in vitro diagnostic (IVD) reagents—which includes those who create laboratory-developed tests—as well as regulatory authorities and medical laboratory personnel. 

This guideline is for use with measurement procedures that provide quantitative numerical results. This guideline is not intended for use with ordinal IVD examinations, commonly referred to as qualitative procedures (see CLSI document EP12). This guideline is not intended to provide information on evaluation of random error (see CLSI documents EP05 and EP15) or to determine the total error inherent in a comparison of measurement procedures (see CLSI document EP21). It is not intended to measure the variability of multiple replicates collected during the measurement of a sample, nor is it intended to measure the bias of individual measurements such as those resulting from sample interference (as covered in CLSI document EP07).

Product Details
EP09Ed3cE
978-1-68440-007-2
116
Additional Details

This document is available in electronic format only.

The U.S. Food and Drug Administration (FDA) has evaluated and recognized this approved-level consensus standard for use in satisfying a regulatory requirement.

Authors
Jeffrey R. Budd, PhD
A. Paul Durham, MA
Thomas E. Gwise, PhD
Douglas M. Hawkins, PhD
Mark Holland, PhD
Beimar Iriarte, MS
Anders Kallner, MD, PhD
Kristian Linnet, MD, PhD
Robert Magari, PhD
Jeffrey E. Vaks, PhD
Abstract

Clinical and Laboratory Standards Institute guideline EP09—Measurement Procedure Comparison and Bias Estimation Using Patient Samples is written for laboratorians and manufacturers. It describes procedures for determining the bias between two measurement procedures, and it identifies factors for consideration when designing and analyzing a measurement procedure comparison experiment using patient samples. An overview of the measurement procedure comparison experiment includes considerations for both manufacturers and laboratorians. Details on how to create difference and scatter plots for visual inspection of the data are provided. Once the data are characterized, various methods are introduced for quantifying the relationship between two measurement procedures, including bias estimates and regression techniques. The final chapter contains recommendations for manufacturers’ evaluation of bias and statement format for bias claims.

Overview of Changes

This guideline replaces the previous edition of the approved guideline, EP09-A2-IR, published in 2010. Several changes were made in this edition, including: 

• Broader coverage of measurement procedure comparison applications 

• More reasons for comparisons based on patient samples (factor comparisons [eg, sample tube types]) 

• Visualization/exploration of data using difference plots 

• Regression descriptions including weighted options, Deming, and Passing-Bablok techniques • Measurement of bias using difference plots 

• Measurement of bias at clinical decision points 

• Computation of confidence intervals for all parameters 

• Outlier detection using extreme studentized deviate 

• Relocation of most of the detailed mathematical descriptions to the appendixes 

This guideline was corrected in 2018 and replaces the original third edition of the approved guideline, EP09-A3, published in 2013. Corrections were made as follows: 

• Reorganizing the content to emphasize the process of performing a measurement procedure comparison 

• Clearly specifying that manufacturers should use regression analysis to characterize bias 

• Adding information on using precision profile information in performing Deming regressions 

• Adding more information on determining confidence intervals for bias estimates at specified concentrations using regression fits 

• Making corrections to the description of the Passing-Bablok regression technique 

• Adding a detailed description of the bootstrap iterative technique for bias estimation 

• Correcting minor miscellaneous errors in equations

Scope

This guideline provides recommendations for designing an experiment and selecting methods to quantify systematic measurement error (bias or difference) between measurement procedures based on comparing patient samples. It provides both difference plot and regression procedures to determine the relationship between two measurement procedures either across their measuring intervals or at selected concentrations. Intended users of this guideline are manufacturers of in vitro diagnostic (IVD) reagents—which includes those who create laboratory-developed tests—as well as regulatory authorities and medical laboratory personnel. 

This guideline is for use with measurement procedures that provide quantitative numerical results. This guideline is not intended for use with ordinal IVD examinations, commonly referred to as qualitative procedures (see CLSI document EP12). This guideline is not intended to provide information on evaluation of random error (see CLSI documents EP05 and EP15) or to determine the total error inherent in a comparison of measurement procedures (see CLSI document EP21). It is not intended to measure the variability of multiple replicates collected during the measurement of a sample, nor is it intended to measure the bias of individual measurements such as those resulting from sample interference (as covered in CLSI document EP07).

Additional Details

This document is available in electronic format only.

The U.S. Food and Drug Administration (FDA) has evaluated and recognized this approved-level consensus standard for use in satisfying a regulatory requirement.

Authors
Jeffrey R. Budd, PhD
A. Paul Durham, MA
Thomas E. Gwise, PhD
Douglas M. Hawkins, PhD
Mark Holland, PhD
Beimar Iriarte, MS
Anders Kallner, MD, PhD
Kristian Linnet, MD, PhD
Robert Magari, PhD
Jeffrey E. Vaks, PhD