engleski

9TH PROFICIENCY TEST: CRITICAL REVIEW

In an inter-laboratory experiment, different laboratories determine some characteristic, e.g. the concentration of an analyte in one or various homogeneous samples under documented conditions. Several subtypes of inter-laboratory studies can be distinguished. Collaborative trials or method performance studies are used to test the performance (generally the precision) of a single analytical method. A standard method, which is routinely used in several laboratories, can also be examined collaboratively to test for a possible bias of either the method (method bias) or the laboratories (laboratory bias) that routinely use it. The comparison of different laboratories that perform comparable analyses with their own individual method is often called proficiency testing or laboratory performance studies ; sometimes, the term “round robin study” is used. It is an essential part of the requirements for laboratory accreditation and the participating laboratories have to prove their technical competence regularly. In proficiency testing, the defined test schemes are organized individually for the analysis of a certain analyte (or a group of analytes) in a certain matrix. All proficiency testing schemes are run regularly, but each scheme has its own frequency of test rounds. The most commonly used evaluation tool in proficiency testing is z- score. A z-score is the ratio of the bias estimate and the target value for the standard deviation of the results. Several estimators can be used for this target value: • A fixed arbitrary value, possibly adapted after a certain number of rounds performed if general analytical improvements allow for a better precision. • A precision estimate required for a specific task of data evaluation. • The reproducibility standard deviation observed from a collaborative study in method validation. • The (average) reproducibility standard deviation observed in the previous rounds. • The expected standard deviation for the concentration range considered according to Horwitz or the slight modification of this method for proficiency tests. • A robust estimator of spread of the data, e.g. Huber’s M-estimator. • The standard deviation of the results after removal of the outliers. Based on an adequate estimator for the standard deviation, the absolute values of the z-scores, which are standardized, allow a direct estimation of the performance of the laboratory by comparison with the following quality limits: • |z|<2: the performance of the laboratory is considered satisfactory. • 2<|z|<3: the performance of the laboratory is considered questionable. • |z|>3: the performance of the laboratory is considered unsatisfactory. As an alternative to these symmetric acceptance intervals, asymmetrical acceptance intervals are suggested for trace analysis. Based on the standard deviation of the test itself, this method takes into account that negative lower acceptance limits are not reasonable. The final report has to be send to the participants. It comprises a description of the preparation of the material and the test for homogeneity, an anonymous (coded) list of all valid data and the statistical results. The z-scores have to be tabulated or graphically expressed. Moreover, abstracts of the methods as well as the remarks of the participants should be added if it is possible.

Proficiency test; laboratory testing

nije evidentirano