engleski

OVERVIEW OF VALIDATION PRINCIPLES II: PRECISION, LOD, LOQ AND ANALYTICAL RESULT

The precision is a measure for the size of the random errors. From a statistical point of view, precision measures the dispersion of the results around the mean, irrespective of whether that mean is a correct representation of the true value. Therefore, it requires the measurement of the standard deviation. How this is done depends on the context. Two extreme types of precision are usually distinguished, namely the repeatability and the reproducibility. In short, repeatability is the precision obtained in the best possible circumstances (same analyst, within one day when possible) and reproducibility in the most adverse possible circumstances (different laboratories, etc.). Intermediate situations may and do occur (intermediate precision). Fourth term used in the context of precision is robustness or ruggedness. Small departures from defined details of analytical procedure often occur when one carries out the in practice. Controlling such parameters may lead to better reproducibility or to avoid sources of laboratory bias. An important characteristic of an analytical method is the smallest concentration of the analyte that can be detected with a specified degree of certainty. The fact that both blank and sample measurements are subject to error requires the problem of chemical detection to be treated in a statistical way. This implies that detection decisions are prone to the two kinds of errors associated with any statistical testing: false positive decisions (type I or α-error) and false negative decisions (type II or β-error). Therefore, three limiting levels are required to completely describe the detection capabilities of an analytical method: • the decision limit at which one may decide a posteriori whether or not the result of an analysis indicates detection, • the detection limit at which a given analytical procedure may be relied upon to lead a priori to detection and • the determination limit (or quantification limit) at which a given procedure will be sufficiently precise to yield a satisfactory quantitative estimate. The experimental design and statistical background needed for estimation of precision, LOD, LOQ and analytical result presentation will be discussed. Furthermore, practical examples will be calculated on sight using MS Excel offering easy transfer and implementation of presented know-how in participants own laboratory.

validation; precision; limit of detection; limit of quantification; analytical result

nije evidentirano