Development of More Statistically Robust Measurement Quality Objectives (MQOs) for Laboratory Bias for the NATTS Network

摘要

A series of statistical analyses were performed using historical NATTS PT sample data collected from 2004 to 2010 to (1) assess the extent to which laboratories are meeting the current MQO on laboratory bias, and to (2) determine whether to recommend modification of this MQO to account for the capabilities of participating laboratories and for differences in the analytical methods associated with the different classes of compounds measured in the NATTS program. In order to assess whether more robust, compound-specific MQOs could be proposed, we investigated the performance of "variance-based" acceptance criteria that account for observed variability in the data within batches of PT samples (e.g., 2 times the standard deviation of percent difference from target values, calculated across years and laboratories). In addition, comparisons were performed to investigate differences in laboratory performance depending on whether the batch-specific target value should be determined differently, such as an average result across the batch rather than as a theoretical spike amount or value determined by analysis of one or several confirmatory samples. This investigation focused on four analytes of high importance to the NATTS program: arsenic, benzene, butadiene, and formaldehyde. Findings indicated that the MQOs should be made specific to an analyte or analyte class, that uncertainty estimates used to establish MQOs should be reassessed and modified periodically, and it may be more appropriate to define the batch-specific target concentration as the average result across all participating laboratories.