Abstrakt: |
An interlaboratory study was conducted to assess two widely used procedures for estimating quantitation levels. Six laboratories participated in the analysis of artificially prepared water samples for organo-chlorine compounds by liquid-liquid extraction followed by gas chromatography-electron capture detector using USEPA Method 608. The study consisted of three phases, including six months of results from analyte free samples, the replicate analysis of fortified samples at a single concentration by the laboratory, and finally the analysis of blind fortified samples prepared by a third party. Estimated detection and quantitation limits (Currie's LC and LQ and USEPA's MDL and ML) were determined for each laboratory-method-analyte combination and then compared to the observed detection and quantitation limits. The overwhelming majority of analyte free samples had a reported value of zero. As a result, observed quantitation and detection limits were frequently zero. When they were not zero, the observed quantitation limits were sometimes less than the observed detection limits and when they were not, there was no observed fixed ratio between the quantitation and detection limits. The variability between days of analysis and the use of noise reducing techniques proved to be a significant source of the observed non-normal distribution of results from distilled water samples with a concentration of zero. Conventional procedures and their underlying analytical and statistical assumptions did not provide useful predictions of laboratory quantitation based upon the results of this study. Rather than one time statistical determinations, ongoing verification of quantitation limits may be a better approach. [ABSTRACT FROM AUTHOR] |