RESUMEN
The GLC/TEA method for N-nitrosodimethylamine (NDMA) in beer was studied collaboratively by 13 laboratories from 7 countries. Collaborators were asked to analyze a total of 10 randomly labeled samples of beer consisting of the following duplicates: a naturally contaminated commercial beer; a beer extremely low (ca 0.1 ppb) in NDMA; and the low NDMA beer spiked with 0.5, 1.9, and 5.0 ppb NDMA. The pooled repeatability and reproducibility coefficients of variation (CV) for all samples were 17% and 27%, respectively. However, when data from 2 laboratories (outliers) were omitted, the corresponding CV values improved considerably (11% and 15%, respectively). Variance analysis showed the presence of a significant laboratory-sample interaction when all data were used for analysis, but this interaction disappeared when data from the 2 outlying laboratories were excluded. The pooled percent recovery of the overall method (omitting outliers) was 101.4 +/- 3.5. All the laboratories detected NDMA in the low NDMA beer. The method was adopted official first action.