Expression of Measurement Uncertainty in Laboratory Medicine (C51-P)
Background, methods, and examples explored in upcoming release of proposed-level CLSI guideline

Measurement uncertainty is defined in the International Vocabulary of Basic and General Terms in Metrology (VIM93) as “the parameter, associated with the result of a measurement, that characterizes the dispersion of the values that could reasonably be attributed to the measurand.” As further stated in International Organization for Standardization (ISO) document, Laboratory medicine – Requirement for reference measurement laboratories (ISO 15195), the parameter may be, for example, a standard deviation (or a given multiple of it), or the half-width of an interval having a stated level of confidence.

Overall, measurement uncertainty estimates are used to determine the quality of test results, contribute to improved interpretation of patient data, and allow for optimal comparison of results across time and place.

Best Practices Outlined in CLSI Guideline

Laboratories are being encouraged to evaluate uncertainty, in order to assess the quality of their results and to determine whether the result is close to any specified limit. Currently, “There is no standard written in the language of the clinical laboratorian providing instructions as to how to estimate, report, or utilize uncertainty. There is an ISO document called the Guide to Uncertainty in Measurement (GUM); however, this document is primarily written by people who were doing physical measurements, such as wave-length or a steel gauge block, and they were able to collect the right kinds of information relatively easily,” says Richard Miller, Dade Behring, Inc., and Chairholder of the CLSI subcommittee that is developing the C51 guideline. There is an expressed need for specific “how to” guidance for measuring uncertainty that translates to laboratory medicine.

This year, CLSI is scheduled to publish Expression of Measurement Uncertainty in Laboratory Medicine; Proposed Guideline (C51-P). The focus of the guideline is to describe a practical approach to developing relevant and useful estimates of measurement uncertainty and for using the information to maintain and improve the quality and application of clinical laboratory measurements.

The guideline gives recommendations for defining the measurand, listing sources of uncertainty, generating statistical estimates of combined uncertainty, and general guidance for reporting uncertainty to users of the results. In addition, commonly used practical strategies for producing estimates in the clinical laboratory are included, such as use of quality control data, information from manufacturers, and method validation results. 

David L. Duewer, PhD, National Institute of Standards and Technology, and C51 subcommittee member, describes the document as providing “1) a broad overview "why" estimating measurement uncertainty is useful and 2) sufficient "how" detail to identify the necessary information and make the estimates. This is done by both providing instructions and a number of fully worked examples.  The document is intended to help put the why and how of measurement uncertainty estimation into a practical, assessable clinical laboratory context.”

Essential Resource for Various Users

As stated in the C51-P guideline, the recommendations are intended for use by the following parties:

• those involved in generating and applying estimates of the uncertainty of measurement in the clinical laboratory; 
• those concerned with oversight of laboratory quality (such as government regulatory
  and accreditation bodies); and
• providers of laboratory products that influence the uncertainty of measurement, such
   as:
  o manufacturers of in vitro diagnostic medical devices,
  o producers of reagents and reference materials, and
  o providers of external quality assessment / proficiency testing schemes.

Miller explains, “For auditors, the CLSI document provides a mechanism to understand how the laboratory that wishes to be compliant with ISO 15189 should be calculating uncertainty.

For in vitro diagnostic (IVD) manufacturers, Miller says, “This document delineates the responsibilities for providing information to the clinical laboratory that is calculating the uncertainty.

 “For clinical laboratories, this document outlines practical techniques for how to perform the new and unusual task of measuring uncertainty,” says Miller. Dr. Duewer agrees, (C51 is) “the only 'measurement uncertainty' guide that explicitly addresses problems that clinical scientists consider relevant.”

“Many external quality assessment organizations at the present time are moving towards traceable systems. Part of a traceable system is the need to understand the uncertainty of values assigned to external quality assessment materials. Secondly, in theory, the medical laboratory that provides the result should be providing an uncertainty. Pass/fail requirements may be based on that uncertainty. Providers of external quality assessments have an interest as well because they need to know the uncertainty that they are introducing into the system, as well as whether the laboratory is capable of providing acceptable results,” comments Miller.  

Linda Thienpont, PhD, University of Ghent, and C51 document development subcommittee member adds, “The demand to know the uncertainty of measurement extends to all parties concerned with physicochemical laboratory analyses (environmental analysis, toxicology, food chemistry, etc.).”

Global and Accreditation Aspects

Accreditation bodies are responsible for ensuring that accredited laboratories meet the requirements of Medical laboratories – Particular requirements for quality and competence (ISO 15189) and/or General requirements for the competence of testing and calibration laboratories (ISO/IEC 17025). Both ISO/IEC 17025 and ISO 15189 cite the GUM as an acceptable model for estimating uncertainty. GUM, jointly published by BIPM, IEC, ISO, IUPAC, IUPAP, and IFCC, is a broad-based, widely used, and accepted method of calculating uncertainty.

The recommendations provided in C51-P are consistent with ISO as expressed in the GUM, and with documents concerned with laboratory accreditation, namely, ISO 15189 and ISO/IEC 17025.

Miller explains, “C51-P will provide laboratories with a mechanism to meet one of the essential requirements of ISO 15189. It also provides each individual laboratory a mechanism for estimating and understanding how reproducible and comparable their results are with laboratories in other parts of the world.” 

At the present time, Miller says, “If a patient has the same test done in more than one laboratory, perhaps in different countries, estimates for comparability of results would be doubtful. C51 will help users understand the level of variability in the results.”   
 
When asked how C51-P differs from the ISO documents and GUM, Miller says, “In general, the ISO standards are a list of minimum requirements without much description about how to implement. CLSI documents include more in-depth explanation of 'how to' and 'why.' As a result, the ISO and CLSI documents are complementary resources.”

Dr. Thienpont elaborates, “Laboratories all over the world are faced with the demand of estimating the uncertainty of their measurements. The rationale behind that demand is that the uncertainty of measurement affects clinical decision making and a means of judging the fitness for purpose of a measurement procedure. National regulations or regulations with regard to laboratory accreditation may be at the basis of this demand.   Namely, to comply with ISO 15189 or ISO/IEC 17025, a laboratory must provide the uncertainty of a measurement. Although everyone recognizes that the estimation should be done consistent with the GUM, it is extremely complicated to know which input elements to include. Consequently, today, there is no coherence with regard to estimating uncertainty; hence, the numbers generated may not be realistic or reliable. Since the forthcoming CLSI document C51 will formulate clear recommendations in this respect, enforced by real life examples, it is a document that the laboratory director should have on his/her desk.”

Proper Uncertainty Measurement Improves Quality and Reliability of Results

Making estimates of the measurement uncertainty of routine methods is an essential step for clinical laboratories when evaluating methods to ensure they produce patient results that are fit for clinical use. Estimating measurement uncertainties also provides laboratories with a better understanding of the performance characteristics and limitations of their methods, and may identify technical steps where uncertainty can be reduced.

Dr. Theinpont says, “I think that a correct estimation of the uncertainty will indirectly improve the quality of the work/products in laboratory medicine. The interesting aspect of correctly estimating the input elements to uncertainty is that one can infer which part of the analysis or a product contributes most. In this way, I suppose that, when necessary, a respectable laboratory director or manufacturer of in vitro diagnostic medical devices will try to improve the analysis or the product in order to decrease the uncertainty to an acceptable level.”

Miller concludes, “The uncertainty concept is one that provides the good, clear understanding of the interchangeability of laboratory results by different laboratories, in different geographic areas, as well as by using different methods. In order to have results that are consistent, which is becoming increasingly important, we have to understand when they are comparable and when they are not. This new CLSI guideline provides the comparison criteria so that somebody can understand and recognize whether results from a Beckman system, or an Abbott system, or a Dade Behring system, for example, would be comparable. It also provides an understanding of whether they would be comparable in Europe vs. Asia or North America.”

A proper measurement of uncertainty is good professional practice and can provide laboratories and patients around the world with valuable information about the quality, reliability, and comparability of results.

For more information about related documents and products, visit www.clsi.org.
  [ return to top ]