Laboratory Method Verification, March 2017

83
Laboratory Method Verification Ola H. Elgaddar MD, PhD, MBA, CPHQ, LSSGB, Lecturer of Chemical Pathology Alexandria University [email protected]

Transcript of Laboratory Method Verification, March 2017

Page 1: Laboratory Method Verification, March 2017

Laboratory Method Verification

Ola H. ElgaddarMD, PhD, MBA, CPHQ, LSSGB,Lecturer of Chemical Pathology

Alexandria [email protected]

Page 2: Laboratory Method Verification, March 2017

Quality

Doing the right thing right, from the first time and every time!

Page 3: Laboratory Method Verification, March 2017

Accreditation

An institution or a program meets standards of quality set forth by an accrediting agency

Page 4: Laboratory Method Verification, March 2017
Page 5: Laboratory Method Verification, March 2017

Validation / Verification??Method validation: (Manufacturer concern)

Establishing the performance of a newdiagnostic tool.Confirmation, through the provision ofobjective evidence, that the requirements fora specific intended use or application havebeen fulfilled’ (doing correct test)......ISO 9001:2005

Page 6: Laboratory Method Verification, March 2017

Validation / Verification??Method verification: (Lab / user concern)

A process to determine performancecharacteristics before a test system is utilizedfor patient testing.

Confirmation, through the provision ofobjective evidence, that specifiedrequirements have been fulfilled’ (doing testcorrectly)……ISO 9001:2005

Page 7: Laboratory Method Verification, March 2017
Page 8: Laboratory Method Verification, March 2017

According to WestgardØ The inner, hidden, deeper, secret meaningof method validation is error assessment.ØHow much error might be present in thetest result within your laboratory ?ØCould this degree of error affect theinterpretation and possibly patient care ?

If the potential error is large enough to lead tomisinterpretation, then the method is notacceptable.

Page 9: Laboratory Method Verification, March 2017
Page 10: Laboratory Method Verification, March 2017

Ø Random error, RE, or imprecision isdescribed as an error that can be eitherpositive or negative, whose direction andexact magnitude cannot be predicted, wherethe distribution of results when replicatemeasurements are made on a singlespecimen.ØUsually, due to error in Pipetting

Page 11: Laboratory Method Verification, March 2017
Page 12: Laboratory Method Verification, March 2017

Ø Systematic error, SE, or inaccuracy is anerror that is always in one direction,displacing the mean of the distribution fromits original value.ØIn contrast to random errors, systematicerrors are in one direction and cause all thetest results to be either high or low.ØEither constant or proportionateØUsually, due to error in calibration

Page 13: Laboratory Method Verification, March 2017
Page 14: Laboratory Method Verification, March 2017
Page 15: Laboratory Method Verification, March 2017
Page 16: Laboratory Method Verification, March 2017

Internal Quality Control (IQC) is used, on dailybasis, in the decision to accept or rejectresults of patients samples and enables thelab to describe and monitor the quality of itswork.-Usually it has 2 levels (Sometimes 3);representing the “Normal” and the“Pathological” analyte level.- Judged according to Westgard Multi-QCrules

Page 17: Laboratory Method Verification, March 2017

External Quality Control (EQC) =Proficiency test is used, on monthly or Bi-weekly (Or others) basis, where labs from allover the world join the program and send theirused Method / Analyzer.- A statistical comparison is made and eachlab result is compared to the result of its peergroup in each analyte.

Page 18: Laboratory Method Verification, March 2017
Page 19: Laboratory Method Verification, March 2017

The following items need verification

ØAnalytical Specificity: Interference studiesØAnalytical Sensitivity: Calibration curve

Detection limitØReportable range: Linearity experimentØPrecision: Replication studyØAccuracy: Bias / Recovery studyØReference Intervals

Page 20: Laboratory Method Verification, March 2017
Page 21: Laboratory Method Verification, March 2017

Analytical SpecificityThe ability of an analytical method to detect

“ONLY” the analyte of interest.

Freedom from interference by any element or compound other than the analyte of interest

Page 22: Laboratory Method Verification, March 2017
Page 23: Laboratory Method Verification, March 2017

Analytical Specificity- Analytical Specificity is verified using

interference studies.- A validated method, known to be free of

the interfering substance is used. A seriesof samples containing increasedconcentrations of the interfering substanceare analyzed using that method, and themethod under study, then both results arecompared

Page 24: Laboratory Method Verification, March 2017
Page 25: Laboratory Method Verification, March 2017

Some Automated systems have a “HIL” index

Page 26: Laboratory Method Verification, March 2017
Page 27: Laboratory Method Verification, March 2017

Analytical Sensitivity- The ability of an analytical method todetect a low concentration of a givensubstance in a biological sample. The lowerthe detectable concentration, the greaterthe analytical sensitivity.- Detection limits studies

Page 28: Laboratory Method Verification, March 2017

Analytical SensitivityOR,-The ability of an analytical method todetect (respond to) a change inconcentration of the analyte. The smallerthe detectable change (the change inconcentration that can result in a definitechange in the reported signal), the greateris the analytical sensitivity.- Slope of the Calibration Curve

Page 29: Laboratory Method Verification, March 2017
Page 30: Laboratory Method Verification, March 2017

REMEMBER!!

Page 31: Laboratory Method Verification, March 2017

STANDARD

• It is a solution of known concentration.• Formed by dissolving a known amount of

an analyte in a specific volume of anaqueous solvent.

• Ex: Dissolving 100 mg glucose in 100 mlD.W gives a standard of a 100 mg / dlconcentration

Page 32: Laboratory Method Verification, March 2017

CALIBRATOR

• A solution or a device of known quantitative or qualitative characteristics

(eg: concentration, activity, intensity)• Used to calibrate or adjust a measurement

procedure.• Matrix is preserved.

Page 33: Laboratory Method Verification, March 2017

• The calibrator material is reconstituted &introduced to the analyzer before using anew lot of reagent, and its concentration isassigned.

• The calibrator is treated like samples andthe absorbance of the developed color isdetermined.

Page 34: Laboratory Method Verification, March 2017

Using the provided concentration, the analyzer constructs a calibration curve

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0 50 100 150 200 250

Page 35: Laboratory Method Verification, March 2017

• If the given concentration is 100 mg / dl,the absorbance of this point is determined.

• By providing the analyzer with the rangewhere this method is linear, the calibrationcurve is extended so that any sampleconcentration lying within that linearityrange can be deduced from this curve

Page 36: Laboratory Method Verification, March 2017

Detection Limits1.Lower Limit of Detection (LLD)= Limit of

the blank (LOB)

2.Biologic Limit of Detection (BLD)= Limit of Detection (LOD)

3.Functional Sensitivity (FS)= Limit of Quantification (LOQ)

Page 37: Laboratory Method Verification, March 2017
Page 38: Laboratory Method Verification, March 2017

Detection Limits1.Lower Limit of Detection (LLD)= Limit

of the blank (LOB)Most analytical instruments produce a signal evenwhen a blank (reagent without analyte) isanalyzed. This signal is referred to as the noiselevel. The LLD is the analyte concentration that isrequired to produce a signal greater than two (orthree) times the standard deviation of the noiselevel.

Page 39: Laboratory Method Verification, March 2017

Detection Limits1.Lower Limit of Detection (LLD)= Limit

of the blank (LOB)Limit of Detection (LLD) is estimated as, the meanof the blank sample plus 2 or 3 times the SDobtained on the blank sample:

LLD = mean(blk) + Zs(blk)

where the Z-value is usually 2 or 3

Page 40: Laboratory Method Verification, March 2017

Detection Limits2.Biologic Limit of Detection (BLD)= Limit

of Detection (LOD)-It is the limit of blank, after the addition of theanalyte of interest (Its lowest concentration)- BLD is estimated as the LLD plus 2 or 3 times thestandard deviation obtained from the "spiked"sample

BLD = LLD+ Zs(spk)

where the Z-value is usually 2 or 3- Results between BLD & LLD should be reportedwithout quantitation

Page 41: Laboratory Method Verification, March 2017

Detection Limits3.Functional Sensitivity (FS)= Limit of

Quantification (LOQ) -The lowest concentration of target compoundsthat can be quantified confidently, that meets somepre-specified targets of imprecision, commonlyCV=20%- Several spiked concentrations must be studied todetermine the precision profile at the lowconcentration range and to select the lowestconcentration at which a 20% CV is obtained.

Page 42: Laboratory Method Verification, March 2017
Page 43: Laboratory Method Verification, March 2017

To Sum Up!1.LoB is the highest apparent analyte

concentration expected to be found whenreplicates of a blank sample containing noanalyte are tested.

2.LoD is the lowest analyte concentration likely tobe reliably distinguished from the LoB and atwhich detection is feasible.

3.LoQ is the lowest concentration at which theanalyte can not only be reliably detected, but atwhich some predefined goals are met. The LoQmay be equivalent to the LoD or it could be ata much higher concentration.

Page 44: Laboratory Method Verification, March 2017
Page 45: Laboratory Method Verification, March 2017

Reportable Range=

Analytical Measurement Range

It is the range of numeric results a method can produce without any special specimen pre-treatment, such as dilution.

Page 46: Laboratory Method Verification, March 2017

Ø It should be verified, for the manufacturer’sclaim, using the linearity experiment.Ø It is performed using either calibrators,proficiency samples, or samplesØ Serial dilutions will be made covering thewhole analytical measurement range, andreaching as close as possible to the claimedvalues of the manufacturer.ØEach dilution is to be processed in duplicateto remove the element of imprecision.

Page 47: Laboratory Method Verification, March 2017

Ø The observed measures (on the X-axis) areplotted against the expected measures (onthe Y-axis), and a line point to point graph isconstructed for each analyte.ØThe line is judged visually for its linearityand according to each experiment, theanalytical measurement range of eachmethod is verified, where any patient resultobtained in the future, outside the verifiedrange, cannot be released without furtherprocessing (Dilution or concentration).

Page 48: Laboratory Method Verification, March 2017

Dilution Expected conc Observed conc

D1 8.27 8.27

D2 4.56 4.7

D4 2.5 2.6

D6 1.5 1.52

D7 1 1.07

D8 0.5 0.68

Correlation: r 0.999886844

Analytical range: fT4 : 0.1 - 12 ng / dl

Page 49: Laboratory Method Verification, March 2017

0

1

2

3

4

5

6

7

8

9

0 2 4 6 8 10

Observed measure

Expe

cted

mea

sure

Page 50: Laboratory Method Verification, March 2017
Page 51: Laboratory Method Verification, March 2017
Page 52: Laboratory Method Verification, March 2017
Page 53: Laboratory Method Verification, March 2017

PrecisionCloseness of agreement between quantity values obtained by replicate measurements of a quantity, under specified conditions.

Page 54: Laboratory Method Verification, March 2017
Page 55: Laboratory Method Verification, March 2017

PrecisionØPrecision should be assessed usingquality control material (A minimum of 2levels), or pooled serum (Of minimum twoconcentration levels)ØEach level of the QC material / pooledserum is measured 5 times per day (Withinrun), for 5 days (in between runs)

Page 56: Laboratory Method Verification, March 2017

PrecisionØThe measures obtained from this precisionstudy are to be collected, and the mean, SD,and CV are calculated for each parameter,for the used QC levels / pooled serum.ØThe obtained CVs, are statisticallycompared to the manufacturer’s claim usingANOVA test of significance, to determine ifthere is a significant different between theobtained CVs (and their verificationintervals), and the manufacturer’s claim at acertain CI; usually 95%

Page 57: Laboratory Method Verification, March 2017
Page 58: Laboratory Method Verification, March 2017

AccuracyCloseness of the agreement between theresult of a measurement and a true value ofthe measurand.

Page 59: Laboratory Method Verification, March 2017

TruenessCloseness of the agreement between thereplicates of result of a measurement, anda true value of the measurand.

Page 60: Laboratory Method Verification, March 2017

ØThe difference between the mean ofreplicates of a measurement, and its truevalue is the BIAS.Ø The CLSI calculation of bias is based onthe results of 7 – 11 PT samples; each ismeasured in duplicate, and then comparedto the true value (Peer’s mean) usingstudent T-test.

Page 61: Laboratory Method Verification, March 2017
Page 62: Laboratory Method Verification, March 2017

A concept!!!

Page 63: Laboratory Method Verification, March 2017

A concept!!!

Page 64: Laboratory Method Verification, March 2017

A concept!!!

Page 65: Laboratory Method Verification, March 2017

Why is Z-score used in PT results??

Page 66: Laboratory Method Verification, March 2017

Bias is verified for the tested method whenthere is no significant difference between:

q Mean Z-score of PT results (7 – 11) is notsignificantly different from Zero

q Mean Z-score of PT results (7 – 11) is notsignificantly different from peer’s mean

Page 67: Laboratory Method Verification, March 2017
Page 68: Laboratory Method Verification, March 2017

Reference intervalsØRemember that we are just “Verifying” thereference intervals stated by themanufacturer or published in the literature,and “transferring” them to the lab using themethod under study.Ø “Establishment” of reference intervals isanother issue.

Page 69: Laboratory Method Verification, March 2017
Page 70: Laboratory Method Verification, March 2017
Page 71: Laboratory Method Verification, March 2017

Reference intervalsØAcceptability of the transfer shall beassessed by examining 20 referenceindividuals, from our subject population, andcomparing the obtained test results to thoseof the manufacturer/ Literature.ØThose 20 individuals should be selected insuch a way that will satisfy the exclusion andpartitioning criteria.ØThe test results should be examined tomake sure that none of the results appearsto be an outlier.

Page 72: Laboratory Method Verification, March 2017

Reference intervals

ØThe manufacturer's / Literature referenceintervals are considered verified if no morethan two of the 20 tested subjects' values (or10% of the test results) fall outside thoseranges.

Page 73: Laboratory Method Verification, March 2017

Reference intervals

ØExclusion / partitioning criteria include:age, sex, fasting status, disease history,drug history, previous surgeries, and time ofthe cycle / pregnancy for females.

Page 74: Laboratory Method Verification, March 2017
Page 75: Laboratory Method Verification, March 2017
Page 76: Laboratory Method Verification, March 2017

Method ComparisonØ According to CLSI, at least 40 samplesshould be assayed on both methods underexamination (two field methods), or betweenone tested method and a reference method.Ø Several statistical approaches can beused, one of them is to calculate thecorrelation coefficient “r”Ø“r” should be more than or equal 0.95

Page 77: Laboratory Method Verification, March 2017
Page 78: Laboratory Method Verification, March 2017

Total Error

It is the summation of both Random andSystematic error.

It is calculated as follow:TE = Bias (%) + 2 CV

It is compared to Biological Variation (or anyother specifications) for Total AllowableError

Page 79: Laboratory Method Verification, March 2017
Page 80: Laboratory Method Verification, March 2017

UncertaintyØ It is an interval around a reportedlaboratory result that specifies the location ofthe true value with a given probabilityØ It takes into consideration both theimprecision (SD), and the inaccuracy (Bias)Ø It is calculated from the data of 6 monthminimum

Page 81: Laboratory Method Verification, March 2017

What performance characteristics are usually validated?

ØReportable range (Linearity)ØPrecision (or imprecision)ØAccuracy (or inaccuracy, bias)ØReference interval

Page 82: Laboratory Method Verification, March 2017
Page 83: Laboratory Method Verification, March 2017