Validation Flashcards
When do you do a Method Verification?
confirm stated performance criteria of established assay being introduced achieved under own test conditions OR
minor change made to existing assay if change in equipment
requirement under ISO 15189
- already undergone validation by collaborative studies
- peer accepted methods published in lit w performance data
What performance parameters need to be evaluated in a full validation?
And what is a full validation ?
(maybe for an in house assay)
Full validation to ensure/demonstrate method meets necessary performance specifications for a clinical laboratory, NATA requirement
Preanalytical
- specimen type/matrix/volume
- process collection, transport, storage
- stability
Analytical (quant or quali)
- selectivity/interference
- linearity/calibration model
- accuracy (reference method) 15%, 20% near LLOQ
- precision 15%, 20% near LLOQ
- LLOQ
- LOD
- stability
- (ruggedness/robustness)
- (recovery)
- sensitivity (clinical)
- MOU
- shelf life of reagents
Post-analytical
- reference interval
- clinical sens/spec
- cut off establish for interpretation
Other
- staff training
- safety issues
- validation report
- EQAP
- approved by senior pathologist, lab manager and quality manager / retention of documents NATA inspection
What is a cross validation/comparative validation?
Describe its process.
Switching from one validated assay to another different assay
- ask for kits/reagents
- gather 20-40 samples of pos, neg, eqap, controls to run
- pre analytical procedures
- analytical procedures
- controls; 1st/3rd/PCS
- establish new LJ chart
- establish new instrument log
- technical
- sens/spec
- precision
- accuracy
- transfer RR (separately)
- new MOU establish
statistically
- linear reg
- bland altman
implementation - documentation/ SOP - staff training - alert NATA change scope - communicate to clients - parallel testing - comment on reports - EQAP notify change in method
For an in house assay being developed for full validation, how many calibrators do you need, and why do you need so many calibrators?
Minimum of 6 for full validation
linearity unknown, more data points will produce valid calibration curve
concentrations of calibrators should lie in clinically significant range of the assay
What characteristics may be evaluated in verifications?
confirm expected perf prior to implementation- needs replication
Pre-analytical
Analytical performance characteristics
- analytical sensitivity
- precision
- accuracy
- measure clinical utility (sen/spec/predictive values)
- reference range
- MOU
Post-analytical
- statistical correlation with old method/bias
What types of validations are there?
Full - non standard or in house methods, or if significantly modified
Partial - modifications to already validated methods ie. change in sample processing, matrix, species of detecting Ab
Cross/comparative- equivalent performance bw two methods
Define the term linearity.
How do you determine linearity?
Linearity- ability to induce signal/response directly proportional to concentration
Run at least 6 samples of known concentration spaced over range of interest, incl blank or close to 0, go up to 150% of conc likely encountered
Plot to visualize relationship, if non-linear
- can try elminate the cause of non linearity
- restrict conc range where it is linear
- use non linear function quadratic fits
Usually measurement range also derived from linearity studies
How do you test for matrix effect?
What should you think about if matrix effect is detected?
Can compare pure vs matrix spiked with standards- run once or twice at 3 concentrations and compare calibration plots
If <10% slope difference, ok not to compensate for matrix effect
If matrix effect
- interference
- matrix change detector response
What is selectivity? and how to do you examine for this
Ability to accurately differentiate in presence of interference ie impurity degradent other ab, matrix, microbes
specificity = in relation to one analyte
in practice, need standards, blanks, assay specimens with and without suspected interferent ig. bilirubin solution
What error does accuracy examine for?
Systematic errors
Precision- improved by increasing sample size
Mean value should be <15% of actual value except LLOQ 20%
How is precision usually expressed?
Closeness between individual test results when repeated
Random error
SD, CV (measure of imprecision)
Repeatability - intra assy
Reproducibility - inter assay
20 times per sample over 3 concentrations
What is ruggedness of a method?
How robust it is- degree to which results are unaffected by minor changes eg. temp, pH, reagent concent, flow rate
How do you establish a reference range as a part of method validation?
More often verified from published ranges/values provided by assay manufacturer
Define method validation
Validation is a process to provide assurance that an assay has an acceptable level of performance under the conditions used in a particular laboratory /and for the population serviced by that laboratory
What parameters do you check for partial and cross validation?
- performance characteristics: sens/spec/PPV/NPV
- accuracy
- precision
- linearity (?)
- reference ranges/transfer 20
- agreement/correlation; linear regression/Bland altman
