Rma. Hahne et Bd. Landenberger, THE USE OF SPIKED ASBESTOS SAMPLES TO DETERMINE ANALYTICAL PROFICIENCY OF COMMERCIAL LABORATORIES DOING FIBER COUNTING, American Industrial Hygiene Association journal, 54(9), 1993, pp. 560-563
A series of proficiency testing samples containing one of two levels o
f chrysotile fibers was prepared by an outside laboratory for use in a
ssessing the competence of commercial laboratories doing asbestos fibe
r counting for various corporate locations throughout the world. Sampl
es were prepared at levels of 768 and 203 fibers/mm2, with standard de
viations of 65 and 18 fibers/mm2, respectively, according to the prepa
rer. A third set of samples with only a background material loaded on
it was also prepared. The samples, which were submitted blind along wi
th unknowns, were analyzed with phase contrast microscopy used by comm
ercial laboratories in various parts of the United States. The results
reported for the knowns were then analyzed statistically. Very few of
the results were identified as outliers at a 98% confidence level, ei
ther within a given laboratory using the Dixon test or among laborator
ies using the Grubbs' test. The performances of the individual laborat
ories were also compared against each other using Tukey Pairwise Compa
risons. Among those laboratories that analyzed at least four samples a
t a particular level, there was no difference among the laboratories'
performances (at a 95% confidence level). Despite the failure of the s
tatistical evaluation to distinguish among the laboratories' performan
ces, the misidentification of spiked sample results by two laboratorie
s proved useful in eliminating laboratories with procedural flaws from
future use by the corporation.