Introduction to method validation

1 Introduction tomethod validationWhat is method validation ? method validation provides documented objectiveevidence that a method measures what it is intendedto measure, with acceptable performance parameterssuch as bias and precision. It is a continuation of methoddevelopment and aims to assess and, if necessary,optimise method performance in a way that meets yourcustomer s needs in a cost effective of validation1 validation has three parts and when applied tomethod validation , these translate as:1. The specific intended use is the analyticalrequirement which is set by the problem that theanalysis is intended to The objective evidence is usually in the form ofdata from planned experiments, from which theappropriate method performance parameters The confirmation is taken as a satisfactorycomparison of the method performanceparameters with what is required, evidence that the method is fit for is method validationnecessary?

2 method validation is an essential part of goodmeasurement practice, because valid data can onlybe produced when the strengths and weaknesses of amethod are understood. For the same reasonsmethod validation is an important requirement oflaboratory accreditation to ISO/IEC 17025:20051. When do you need to validatemethods? validation usually begins during the methoddevelopment stage, when some performanceparameters are evaluated approximately to determinewhether the method s capabilities are in line with thelevels required. If a previously validated method hasnot been used for a period of time, its performanceshould be checked (verified) before it is reinstated. Ifthe scope of the method is altered, the method isto be applied to different sample types or analytelevels than it was originally validated for, theperformance must be validated for the new type ofsample. It is important to remember that you mustverify the performance of all test methods before theyare used for the analysis of customer samples,including published and standard methods which havebeen validated by do you validate methods ?

3 method validation should always be a planned first stage is to examine the problem. Look at thereasons behind carrying out the analysis and find outwhat it is that the customer hopes to establish fromhaving the work carried out. It should then be possibleto decide which method performance parameters arerelevant to the work and the target values that arerequired for each parameter. A set of experiments canthen be designed which can be used to evaluate theperformance of the chosen planning a method validation study it is alsoimportant to consider how the experimental data willbe analysed and the statistical tests that will be usedto assess fitness for validation plan should comprise details of thematerial that is going to be analysed to assess eachof the performance parameters, the number ofreplicate measurements required and the statisticalanalysis that will be carried out to evaluate the parameters may be examined in one set ofexperiments in which case the order in which thingsare done can be the plan is formalised, experiments can becarried out to produce data to allow the methodperformance parameters to be evaluated.

4 Theresulting data are compared with the target values todetermine if the method is fit for do you decide fitness forpurpose ?Once the method performance data have beengenerated and collated, an assessment can be madeabout whether the required performance target valueshave been met. If they have been achieved then themethod can be declared fit for purpose andconsidered validated. If the target values are notachieved further development of the method will benecessary, followed by reassessment against thetarget tomethod validation Confirmation, by the examination and theprovision of objective evidence, that theparticular requirements for a specific intendeduse are fulfilled. 11 ISO/IEC 17025:2005. General requirements for the competence of testing and calibration laboratoriesPrecisionSpecificity/selecti vityBiasRuggednessLinearity/working rangeLimit of detectionLimit of quantitationType of analysisQualitativeMajor component Trace analysisPhysical properties The validation processMethod validation builds on the information obtained during method development.

5 The process of validation issummarised in Figure Central to the process is the purposeof the exercise. You need to be clear what the analytical method is intended to achieve. It is useful to consider whether the method evaluation is intended to demonstrate performance against a requirement ( validation ) or performance up to a standard already demonstrated (verification). What performance parameters are critical? The end use should help to define the performance requirements in terms of bias, precision, ruggedness etc. What values should they have? It should be possible to establish some guideline figures for performance. These form the initial performance specification. How should the performance be measured? How much experimentation is required? Once you have the specification it should be possible to plan a set of tests to establish whether the method performance is adequate.

6 The number of tests required can be estimated statistically or by following published best practice or sectoral guidelines. How should the data be interpreted? This could be a simple comparison of performance data with the performance specifications, aided by statistical testing. In some cases the data are reviewed by a third planInterpretationPerformance parametersThe key performance parameters that require attention during validation vary from one analytical requirement to anotherand from method to method , but some commonly important parameters are listed in the table 1: The validation processParameterDifferent method performance parameters will beimportant in different situations. For example, accuracy(precision and bias) will be important for determiningabsolute values of properties or analyte trace work, limits of detection and of quantitation areimportant but these parameters are less important if theanalyte is present at higher concentrations.

7 Whenplanning calibration strategies it is important to know therange over which the response is linear. Ruggednessstudies will indicate which parameters need to becontrolled in order to preserve performance. The amountof validation required will also depend on the source ofthe test method and the extent of any previous validationstudies. You are likely to have to carry out a significantamount of work to validate a method that you havedeveloped yourself, compared to verifying theperformance of a published test method that has alreadybeen validated by an interlaboratory of precision2 Precision is a measure of the spread of results, howclose a group of results are to each other. Precision givesno indication as to how close results are to the true precision of an analytical method is evaluated bymaking repeat independent measurements on identicalsamples. From the observed spread of the results a valuefor the precision of the method can be calculated.

8 It isusually expressed as a standard deviation or as a percentrelative standard deviation. The magnitude of the precisionis influenced by the size of random is not necessary to know the exact concentration of theanalyte in the sample used in a precision , the material must be sufficiently stable andhomogeneous and should be similar to the test sample interms of the analyte concentration and matrix. Theconditions under which the measurements are madedetermine the type of precision estimate obtained. Repeatabilityrepresents the tightest extreme ofindependent replicate measurements. It describes theprecision that you would expect for a set of replicatemeasurements made one after the other, in a singlelaboratory, by a single analyst on a single type of precision study is useful for evaluating thelikely variation in measurements made in the same batchof analyses.

9 Reproducibilityrepresents the widest extreme ofprecision. It describes the variation within a set ofmeasurements made on a sample over an extended timeperiod, in several laboratories, by a number of differentanalysts on different instruments. You would expectreproducibility to reflect variation in the method from allpossible sources. This type of precision represents theexpected variation in results when a method is used toanalyse a sample in several different is a less widely accepted termcompared to repeatability and reproducibility. Itrepresents the variation in results obtained in a singlelaboratory over an extended time period. When a singlelaboratory uses several analysts or sets of equipment fora particular test method , intermediate precision has agreat practical value. Compared to repeatability,intermediate precision is likely to give a better estimate ofthe precision of the method in routine use and istherefore the most appropriate precision value for settingquality control limits.

10 There are various combinations ofconditions that can lead to an estimate of intermediateprecision so the conditions used in the study shouldalways be stated. This type of precision estimate issometimes referred to as, within laboratoryreproducibility .If the method is to be applied to a range of sample types( different analyte concentrations or sample matrices)then the precision will need to be evaluated for arepresentative range of samples. In particular precisioncan vary significantly with analyte of trueness2with a note thatFigure 2: BiasThe bias of a method (as illustrated in Figure 2) is thedifference between the average of a number of measuredvalues and the true value. Bias is evaluated by carryingout replicate analysis of a sample with a known oraccepted reference value, ideally a certified referencematerial (CRM), and comparing the average of the The closeness of agreement between independenttest results obtained under stipulated conditions.