Purpose: Quantitatively analyze variability of analysts, laboratories and methods enumerating coliforms in shellfish waters based on 2013 and 2014 FDA PTs.
Methods: Eight growing area and eight process water simulated shellfish samples were inoculated with coliforms at 10-80 CFU/100 ml. The inoculation procedures, mixing methods and homogeneity/stability were verified according to ISO 17043 and 13528. Participants utilized NSSP-approved methods (most probable number (MPN) or membrane filtration). A total of 119 analysts (48 laboratories) in 2014 and 121 analysts (48 laboratories) in 2013 tested simulated shellfish samples for total and/or fecal coliforms.
Results: Accuracy and precision were calculated according to ISO 13528. Analyst z-scores were based on robust consensus means. Fecal coliform reproducibility SD was 0.279 and 0.177 in 2013/14 for MPN, and 0.263 and 0.165 in 2013/14 for membrane filtration. Similar patterns were seen in process water samples where total coliform reproducibility SD was 0.305 and 0.197 in 2013/14 for MPN, and 0.275 and 0.158 for membrane filtration. Test on equivalency according the Q-Hampel method showed MPN and membrane filtration did not have equivalent performance across 8 growing and 8 process water samples in 2013. Means for MPN and membrane filtration were not equal, but the mean deviation fell within tolerated limits across 8 growing water and 8 process water samples in 2014.
Significance: Analysis of data from two consecutive annual PTs demonstrates method variability plays a role in coliforms testing and could affect enforcement of shellfish regulations. This study shows how large-scale PTs in relevant matrices can be used to generate method performance data and evaluate overall performance of laboratories.