BIOLOGICALLY RELATED NATIONAL INSTITUTE OF STANDARDS AND TECHNOLOGY STANDARD REFERENCE MATERIALS - VARIABILITY IN CONCENTRATION ESTIMATES

The extensive compilation of data on SRMs by Gladney et al. (Standard Reference Materials: Compilation of Elemental Concentration Data for NBS Clinical, Biological, Geological, and Environmental Standard Reference Materials, NBS Special Publication 260-111 (1987) provides a remarkable opportunity to check the statistical distribution of analytical results for SRMs reported under unusually favorable circumstances. The precision estimates obtained for 11 biologically related SRMs by the compilers, who exercised subjective, critical analytical judgement to eliminate data "clearly beyond the limit of acceptability" by removing values beyond 2 standard deviations from the mean, provided published consensus values that are compared to the results predicted by the Horwitz equation: relative standard deviation between-laboratories (RSD(R), %) = 2 exp (1 - 0.5 log C), where C is the concentration expressed as a decimal fraction. The compilers removed 14% of the reported values from the 117 certified analyte-matrix combinations, when at least 8 values were available, for 28 elements over a certified concentration range from 5 ng/g to 40 mg/g. Only 2 of the resulting RSD(R) values are outside the acceptable maximum of twice the RSD(R) value calculated from the Horwitz equation. However, removal of 14% of the data as nonconforming is regarded as excessive. This finding suggests that many laboratories lack a quality assurance program to investigate the occurrence of gross deviations from certified values.