Methods in validating tests
Precision can be expressed as: repeatability, or intra-assay precision (as a measure of precision under the same operating conditions over a short interval of time); intermediate precision within-laboratory variations (different days, different analysts, and different equipment); and reproducibility, which is the precision between collaborative laboratories.
Single laboratory precision or repeatability can initially be based on one homogeneous sample and is expressed as standard deviation.
The accuracy of an analytical procedure is defined as how close the test results of the parameters are for a specific analyte compared to the true measure of these parameters.
In the case of trace analysis, accuracy can be established through the analysis of a certified reference material, a comparison to data obtained by an independent validated method, or an interlaboratory comparison involving a laboratory accredited by the International Organization for Standardization (ISO) and compliant with the general requirements for the competence of testing and calibration laboratories (ISO/IEC 17025).
Specificity is the ability to unambiguously measure the content or potency of analyte in the presence of other components such as impurities, degradation products, and matrix components.
The way in which validity is conceptualized determines the scope and nature of validity investigations and hence the methods to gather evidence.
In the case of trace analysis, limit of detection is defined as 3*SD can be obtained by extrapolation from a plot of standard deviation versus concentration where three concentrations are analyzed ~11 times each that are at the low, mid, and high regions of interest using a matrix that matches the sample matrix.
Limit of quantitation is the lowest amount of analyte in a sample that can be determined with acceptable precision and accuracy.
As with limit of detection, depending on whether the procedure is instrumental or non-instrumental, numerous approaches can be used.
These can be based on visual evaluation, signal-to-noise, standard deviation of the response and the slope, standard deviation of the blank, and the calibration curve.