Food Standards Agency
Sunday 5 July 2009
Safer food better business banner
AZ-Directory
What's NewRSS
What is RSS?Listen
Listen to this siteWednesday 2 August 2006
This research project aims to access current strategies used for homogeneity testing, and to identify the limitations and practical difficulties associated with these approaches.
Study Duration : June 1999 to May 2001
Contractor : Central Science Laboratory
A condition for the satisfactory execution of performance assessment schemes, including the Food Analysis Performance Assessment Scheme (FAPAS
), is that the test material should be sufficiently homogenous.
In this context, 'sufficiently homogenous'is taken as the variability of the level of analytes in the test material not being large enough to materially effect the outcome of the performance assessment.
The International Harmonised Protocol for Proficiency Testing of (Chemical) Analytical Laboratories suggests a suitable procedure for establishing the homogeneity of a candidate test material. It also recommends that the estimate of the between-sample standard deviation obtained by the procedure should be less than 0.3 times the target standard deviation for the measurement. However, if the between-sample standard deviation of a candidate test material is marginally below or equal to 0.3 times the target standard deviation then those materials will needlessly fail the homogeneity test.
The current protocol for the assessment of the homogeneity of a FAPAS
test involves the analysis of ten test material samples under repeatability conditions, with the results of the analyses being exposed to statistical tests to assess the degree of homogeneity of the test material.
Recent proposals suggest that the test for the homogeneity of test materials should be used to prove heterogeneity (at 95% confidence) before rejecting a test material, rather than confirm homogeneity before accepting a test material.
A new homogeneity test has been proposed. This protocol is a refinement of the current protocol, including a test for analytical outliers (Cochran's). The homogeneity test within the new protocol combines the F-test and the sufficient homogeneity test of the current protocol into a single test that is designed to 'fail'test materials only when 'heterogeneity'is proven, and as such is designed to avoid the needless rejection of 'sufficiently homogenous'test materials.
The aim of this study is to compare the results of applying the new test and the current test.
The performance of the protocols was examined in two ways
Data sets were produced to simulate the results of duplicate analyses of ten test material samples. This simulated data was used to assess the ability of each protocol to discriminate between data sets produced by the analysis of 'sufficiently homogenous' samples and the analysis of heterogeneous samples.
151 data sets produced, by the analysis of 62 FAPAS test materials between August 1998 and August 1999, were used to assess the likely practical effect of changing the protocol for assessing the homogeneity of test materials.
This was achieved by graphically comparing the outcome of each test on each data set.
An assessment of the quality of the data supplied to the FAPAS Secretariat was also carried out by comparing the means and variances of each series of duplicate analyses within a data set.
Analysis of simulated data sets showed that the new protocol performs well, with the rejection of approximately 5% of 'just sufficiently homogenous' data-sets, compared with the current protocol which yields rejection rate of around 50%. However this improvement was at the cost of an increased rate of acceptance of 'heterogeneous' datasets as 'homogenous'.
Analysis of simulated data also showed that the duplicate analysis of ten samples could not be used to discriminate between 'homogeneous' and 'heterogeneous' data sets if the analytical standard deviation was greater than 0.4σ.
Assessment of 149 sets of homogeneity data, generated by the analysis of 62 candidate FAPAS test material, showed that, in practice, there is little difference between the protocols. The current protocol accepted 94% of samples as homogeneous, while the new protocol accepted 98% as homogenous. However, it should be noted that the data set used in this study may not be representative of all of the homogeneity tests carried out on behalf of FAPAS, but merely representative of homogeneity tests used to make a 'final' decision about a materials homogeneity.
There does not appear to be any pressing need to replace the current protocol with the new protocol.
As it is not clear whether the data kept by FAPAS
is representative of all of the homogeneity tests carried out on its behalf, records on all homogeneity tests will now be kept, and this recommendation should be reviewed in 12 months' time with consideration given to the following:
The final report is available from the FSA Library and Information centre. To obtain a copy, please contact the Enquiry Desk, Information Services, Food Standards Agency (020 7276 8181/8182 or at infocentre@foodstandards.gsi.gov.uk ).
Contact
: For any enquiries concerning this research project, please contact the relevant Programme contact or email
science@foodstandards.gsi.gov.uk
Find out what our other sites have to offer