Method Validation in a Clinical Lab

What is the difference between validation & verification?

validation: is the establishment of performance specifications

verification: is the comparison of performance specs w/ that of the manufacturer

What 2 tests are also included in high complexity testing?

Lab modified moderate complexity tests & all lab developed tests (LDT)

What 3 materials are used when performing a method verification?

Cal Ver Material (when you’re gonna have a new device or need to change setup or range adjustment)

Calibrator material (used to run b4 QC)

QC material

What are the requirements for validation of waived tests between CLIA, CAP, & TJC?

CLIA: follow the manufacturer’s instructions

CAP: not required except for verification of reference range

TJC: validation of accuracy, precision, reportable range

What are the requirements for validation of non-waived tests between CLIA, CAP, & TJC?

CLIA: accuracy, precision, reportable range, reference interval for moderate, sensitivity & specificity for high complex

CAP: All 6 of the above

TJC: all of the above

What are the requirements for validation of POCT for CAP?

SOP’s, Personnel training/competency, proficiency tests, instrument maintenance, specimen handling, QC/QA, reporting

In regards to accuracy & precision, what results in a random error, systematic error, & acceptable error?

random: good accuracy, low precision

systematic: good precision, low accuracy

acceptable: good for both

What is the CLSI protocol in regards to measuring precision?

suggests 2 levels, run 3x per run, over five days (15 replicates in all)

What is the equation for random error?

RE=2*SD

What are the 2 ways in which analytical accuracy is measured for a method?

1. comparison of results b/w new method & reference method

2. results using new method on certified reference materials

When validating for analytical accuracy, what are the requirements for comparison of method (COM)?

minimum 20 test samples over span of entire testing range (does not exceed measurement range)

run both new & reference methods

evaluate bias b/w two tests using linear regression