Proficiency Testing System of Agricultural Product for National Laboratory Accreditation

Abstract

Proficiency testing was inter laboratory comparisons. The same sample could have different test results when tested in some different laboratories. To find out and improve the performance of laboratories, proficiency testing can be done. Different evaluation of test results from participant laboratories could make different interpretations of the proficiency testing, which means it can also affect the assessment of the participant laboratories. The aims of this research are 1) to design a system for proficiency testing by using several evaluation methods of test results 2) to identify the testing methods used by the participant laboratories and possible causes of outlier laboratories. The method used in this research are: method 1: one round data selection by Grubbs and the rest of the data is calculated by Robust Z-Score; method 2: data selection by Grubbs several times until no more data can be excluded and the rest data is calculated by Robust Z-score; method 3: direct calculation by Robust Z-Score; method 4: calculation by reference value. The data used are the results of proficiency testing of cocoa powder, chili sauce and vegetable oil; reference values come from German reference laboratory; interview result from the stakeholders and field observations for related testing. The results show the median from Indonesian laboratories are not too different from the reference value; distributions of data are large enough, proved by CV robust larger than CV Horwitz (except for cadmium and the amount of dissolved solid); method 2 is chosen as the best method based on the smallest CV Robust; method 2 is the most sensitive which is giving the bigger number of outlier laboratories than methods 1 and 3; the number of outlier laboratories with method 2 are less than method 4; testing laboratories using National Standard of Indonesia (SNI) obtained satisfactory results (40% -100%) by method 2, for most of the test parameters; possible causes of outlier laboratories results come from a variety factors, particularly the implementation of testing that does not comply with the testing method.