|
|
|
Proficiency Tests
These tests aim at providing continuous information on the ability of participant laboratories, in achieving good performance scores through time. Nowadays they are almost mandatory for laboratories accredited in accordance with ISO 17025 or other accreditation schemes. Regardless of formal requirements from the part of accreditation bodies, international or national regulations, or laboratory clients, proficiency tests are a useful tool for laboratories that work for continuous improvement, and are an important aid to internal quality control. There are many different possible schemes for conducting interlaboratory tests for this purpose, such as those from FAPAS, NATA, or the Harmonised Protocol, devised by IUPAC and AOAC International (see the References, for a complete listing of these institutional instructions). A recent published document, standard ISO 13528:2005, is a very thorough document regarding statistical techniques for proficiency tests, and contemplates most of the traditional techniques that have been used by many Proficiency Test (PT) providers, in the course of the years. InterLab2.0 was initially devised for collaborative studies. As many PT coordinators use ISO 5725-2 in order to evaluate their tests, some features were added, in order to help those organisations in the statistical treatment of results from PTs. Basically, InterLab2.0 uses data from the PT round to derive the assigned value (as defined in ISO Guide 43-1) of each analyte in the sample. This value may be obtained as a general mean of all participants’ results, or it may be obtained after elimination of outlying results. The outliers are evaluated using the tests available in ISO 5725-2 (Cochran’s and Grubbs’ tests), in a number of different ways, depending on the design of experiment used for the particular test. It is also prepared to follow Prescott’s test, as described in ISO/TC 102/TCR 5, as an alternative to Grubbs’ test. This test was included based on research by Prescott (1978) that shows the weaknesses in Grubbs’ test, when there are more outliers than expected in the data set. It is also possible to use InterLab2.0 with a pre-defined assigned value, once the result may be registered to each sample in the initial definition and recording of the samples for the tests (the Maintenance – Samples Menu). This result may be originated from expert laboratories, from a certificate in case a Certified Reference Material is used, or from Formulation experiments. For setting the value of the standard deviation for proficiency assessment (or the target value for standard deviation), InterLab2.0 offers two choices: consensus value obtained from the participants results in each round (with or without eliminating outliers), and by using the Horwitz Function to predict the reproducibility standard deviation as a function of concentration. This kind of application is very specific, and for that purpose InterPro 1.0 web system is available. |
Questions or comments about this website, please contact us at:
webmaster@kymi.com.br.
|