Table of Contents

Return to previous page

View Article

Report by the EBC Analysis Committee on the Determination of NDMA in Beer by GC-TEA or GC-MS Detection (2009 Collaborative Trial) - Approved by the Analysis Committee for inclusion in the EBC Analytica.
H. G. J. Welten and F. R. Sharpe

The determination of NDMA in beer by the method of GC-TEA or GC-MS was collaboratively tested by the EBC Analysis Committee according to ISO standard 5725-2: 1994. Repeatability (r95) and reproducibility (R95) values are presented. There is statistically significant evidence (Spearman rho = 0.83, p = 0.008) of dependence of repeatability on the mean value (repeatability is worse for bigger mean), there is also statistically significant evidence (Spearman rho = 0.94, p = 0.001) of dependence of reproducibility on the mean value (reproducibility is worse for bigger mean). This relationship between s and m can be represented by a straight line s = a + bm. Final repeatability standard deviation = sr = 0.022 + 0.034 m. Final reproducibility standard deviation = sr = 0.054 + 0.117 m. Note that the precision becomes poor at concentrations below 0.5 ?/l.The EBC Analysis Committee tested a new method for the determination of NDMA in beer that uses the principle of GC-TEA or GC-MS. An instrument ? the Thermo Energy Analyser (TEA), is commercially available and is finding use within the brewing and malting industry. Beer is extracted with dichloromethane and after concentration a GC analysis is performed. A nitroso-sensitive detector (TEA or MS) is used. The NDMA concentration in the sample can be calculated with the use of calibration samples.Twelve identical bottles of each of seven beers containing prepared concentrations between 0.18 to 2.42 ?g/litre of NDMA were distributed to 12 participants nominated by the EBC and ASBC Analysis Committees. Twelve sets of data were received for samples one to seven.Participant 5 appeared to have a significant problem with the analysis and the EBC Analysis Committee decided to eliminate their results prior to statistical evaluation. This still left sufficient laboratories presenting results to justify statistical analysis. All the data used in the statistical analysis is presented in table 1 for each sample and participant. Statistical evaluation was carried out using Mandel?s k and h statistics to test the consistency of results.- k refers to within-laboratory consistency and repeatability and the results are shown in figure 1- h refers to between laboratory consistency and reproducibility and the results are shown in figure 2O Outliers have been evaluated using the Cochran?s test and stragglers using the Grubbs?test. Data have already been checked and corrected. Looking at the collation of values (standard deviations, indicating the spread of the values), we see that for the precision of sample 7 results from participants 2 and 4 are very poor. Two combinations are excluded: Participant 2, bottle 7 and participant. 4, bottle 7. ...

Descriptors: NDMA in beer, Analytics, EBC Analytica, GC-TEA-Analysis, GC-MS-Analysis

BrewingScience Monatsschrift fr Brauwissenschaft, 64 (May/June 2011), pp. 46-51