1. RECENT DEVELOPMENTS AT THE CODEX ALIMENTARIUS COMMISSION AND THE INTER-AGENCY MEETING WITH RESPECT TO FOOD ANALYSIS Roger Wood Food Standards Agency, c/o Institute of Food Research Norwich Research Park, Colney, Norwich UK NR4 7UA

2. CODEX ALIMENTARIUS COMMISSION Inter-governmental organisation setting food standards on a world-wide basis. Pre-eminence because referenced in the WTO agreements.

3. RESPONSIBILITIES CCMAS is primarily responsible for: developing quality standards for analysis and sampling of foods (including laboratory quality standards) recommending various technical aspects of methods of analysis (eg measurement uncertainty, recovery considerations) endorsing methods of analysis submitted to the Committee from the various Codex Commodity Committees. CODEX COMMITTEE ON METHODS OF ANALYSIS AND SAMPLING

4. GUIDELINES FOR THE ASSESSMENT OF THE COMPETENCE OF TESTING LABORATORIES INVOLVED IN THE IMPORT AND EXPORT CONTROL OF FOOD Adopted by the Commission at Step 8 in June 1997 SCOPE 1.These guidelines provide a framework for the implementation of quality assurance measures to ensure the competence of testing laboratories involved in the import and export control of foods. 2.These guidelines are intended to assist countries in the application of requirement for trade in foodstuffs in order to protect the consumers and to facilitate fair trade.

5. REQUIREMENTS 3.The following criteria shall be adopted by laboratories involved in the import and export control of foods: Compliance with the general criteria for testing laboratories laid down in ISO/IEC Guide 25:1990 “General requirements for the competence of calibration and testing laboratories”; [i.e. effectively accreditation], Participation in appropriate proficiency testing schemes for food analysis which conform to the requirements laid down in “The International Harmonised Protocol for the Proficiency Testing of (Chemical) Analytical Laboratories”, Pure and Applied Chemistry 65 (1993) 2132-2144; [already adopted for Codex purposes by the CAC at its 21 st Session in July 1995]

6.  Whenever available, use methods of analysis which have been validated according to the principles laid down by the CAC, and  Use internal quality control procedures, such as those described in the “Harmonised Guidelines for Internal Quality Control in Analytical Chemistry Laboratories”, Pure and Applied Chemistry 67 (1995) 649-666 4.The bodies assessing the laboratories referred to above should comply with the general criteria for laboratory accreditation, such as those laid down in the ISO/IEC Guide 58:1993: “Calibration and testing laboratory accreditation systems - General requirements for operation and recognition”.

7. Method Criteria in Codex accuracy applicability (matrix, concentration range and preference given to ‘general’ methods) detection limit determination limit precision; repeatability intra-laboratory (within laboratory), reproducibility inter-laboratory (within laboratory and between laboratories), but generated from collaborative trial data rather than measurement uncertainty considerations recovery selectivity sensitivity linearity

8. Current Topics of Note The use of sampling and analytical results (compliance) Proposed Draft General Guidelines on Sampling Criteria for Evaluating Acceptable Methods of Analysis for Codex Purposes Endorsement of Methods of Analysis Provisions in Codex Standards Single-Laboratory Validation Proposed Draft Guidelines on Measurement Uncertainty GMO Methods of Analysis Recovery

9. MAJOR ISSUES IDENTIFIED The basic principles of the sampling procedure used by The Member States; the treatment of analytical variability (normally known as the measurement uncertainty) in the interpretation of a Codex specification, and the use of recovery corrections when calculating and reporting analytical results.

10. The effect of different countries taking different approaches for each of the issues identified are described. It must be appreciated that there may be other enforcement issues which have a similar effect.

11. Two countries may have different national rules for the interpretation of results from lots: Country A requires: that each and every item in the lot meets the specification. In this example it means that all 1,000 units, if analysed separately, would have to be less than 2.0mg/kg. Here a significant number of units are greater than 2.0mg/kg so the lot would be deemed to be in non-compliance with the Codex specification and so would be rejected, but Country B requires: that the mean value of the characteristic in the lot is to be less than the Codex specification. In this case the mean value is 1.9mg/kg so the lot would be deemed to be in compliance with the Codex specification.

12. Consequence: the two countries A and B will make different judgements as to the compliance with a Codex specification on essentially the same lot. This is unacceptable and can only be avoided if the sampling procedures are elaborated at the same time as the commodity standard is elaborated in the Commodity Committee. In addition it should also be noted that he number of units to be analysed also influences the decision on compliance.

13. ANALYTICAL CONSIDERATIONS The treatment of analytical variability (normally known as the measurement uncertainty) in the interpretation of a Codex specification The use of recovery corrections when calculating and reporting analytical results

14. It is essential that interpretation of analytical results is similar if there is to be equivalence across Codex; without it there is no uniform interpretation of legislation. Some of these points now explained.

15. REPORTING OF RESULTS WITH RESPECT TO THEIR MEASUREMENT UNCERTAINTY All analytical results should be reported in the form “a+b” where “a” is the best estimate of the true value of the concentration of the measurand (the analytical result) and “b” is the range within which the true value is estimated, with a given probability, to fall. The value of “b” is known as the “measurement uncertainty” and may be estimated by the analyst in a number of different ways.

16. The estimation of the value of “a” is dependent on: the accuracy of the method of analysis used how well the analyst uses that method, ie whether the analytical system is “in control”. The value of the measurement uncertainty “b” is dependent on: the inherent precision of the method of analysis used the number of analytical replicates that are carried out. The more replicates, the less the value of the measurement uncertainty.

17. REPORTING OF RESULTS BY FOOD CONTROL ANALYSTS IN DIFFERENT COUNTRIES The procedure adopted by some food control analysts is to report samples as containing “not less than “a” – “b”” in situations where the statutory limit is a maximum permissible concentration. Thus, in any enforcement situation the maximum benefit is given to the food producer. This is consistent with the requirement to prove beyond reasonable doubt that a limit has been exceeded, if the case should come to Court. This does mean that the effective enforcement limit is, in such countries, not identical to the numerical value given in legislation. Other food analysts may report the value “a” without taking into account any measurement uncertainty considerations.

18. CONSEQUENCES OF REPORTING RESULTS IN DIFFERENT WAYS There are potential problems with the reporting of results for which there is a Codex specification. This is best explained by example: Let us assume that there is a specification of 4 µg/kg for the analyte being analysed. It would be anticipated that the measurement uncertainty for the analysis will be of the order ± 45% of the analytical result, ie the analyst would determine for nominal concentrations of 3, 6 and 10 µg/kg, the following concentrations including their uncertainties: a.3.0 ±1.3µg/kg b.6.0±2.6µg/kg c.10.0±4.4µg/kg

19. Situation a Here the level reported is below the specification. All countries would take the same view and accept the material.

20. Situation b Here the level reported is above the statutory limit but the true value lies in the range 3.4 to 8.6 µg/kg. The level and its uncertainty would be reported. Here some countries would report the sample as containing not less than 3.4 µg/kg of the analyte and because it is not beyond reasonable doubt that the limit has been exceeded, no action will be taken. However, other countries may take action on the 6.0 µg/kg result, without taking uncertainty into account. For these countries, the material will be deemed to be non-compliant.

21. Situation c Here the level reported is above the specification and the true value lies in the range 5.6 to 14.4 µg/kg. All countries will state that the material is non-compliant with the specification.

22. Conclusion In situation b, there is the possibility that different countries will make opposite decisions as to whether the material conforms with the specification. Various projects have shown this to be the situation.

23. THE USE OF RECOVERY INFORMATION IN ANALYTICAL MEASUREMENT A real example may result in the mycotoxin area where there may be a limit of 4µg/kg for total aflatoxin in nuts. Here the following situation may arise: Country A will analyse a consignment and find a result of 3.5µg/kg total aflatoxin using a method which, in the analytical run, has a recovery of 70%. Country A does not correct for recovery corrections as a matter of policy and so the reported result will be 3.5µg/kg and so the sample will be in compliance with the 4µg/kg limit.

24. Country B, however, uses recovery corrections as a matter of policy. That country could analyse the “same” sample using the “same” methodology and obtain the “same” analytical result but will report not 3.5 but 5µg/kg on a recovered basis. Here there is the possibility that because the 5µg/kg level is greater than the limit of 4µg/kg limit for total aflatoxin, that country may deem the sample not to be in compliance with the limit. As in the previous situations it is important that Codex stipulates the basis on which the specification is to be enforced.

25. It is stressed that this is not an analysis or sampling problem as such but an administrative problem which has been highlighted as the result of recent activities in the analytical sector, most notably the development of International Guidelines on the Use of Recovery Factors when Reporting Analytical Results, and various Guides prepared dealing with Measurement Uncertainty.

26. Note: Some of these issues are now being addressed by the EU Commission, particularly the Measurement Uncertainty and Recovery issue.

27. These aspects directly affect the interpretation of results in countries which use Codex Standards and so may be regarded as “food control”. At the present time there is no common interpretation of analytical results across the Codex Community so significantly different decisions may be taken after analysis of the “same sample”.

28. It is essential that interpretation of analytical results is similar if there is to be equivalence across the Codex Community; without it there is no uniform interpretation of Codex Standards.

29. Criteria for Evaluating Acceptable Methods of Analysis for Codex Purposes a)Proposed Draft Guidelines for Evaluating Acceptable Methods of Analysis Government comments b)Proposed Amendment to the Principles for the Establishment of Codex methods of Analysis in the Procedural Manual Government comments

30. Will enable performance-based approach to be introduced rather than a prescribed method approach – use any method provided: Laboratory is proficient Method meets the performance criteria identified above

31. In order to do this we do need method performance information, laboratory proficiency testing, all of which is aided by CRMs. Also helps in enabling the measurement uncertainty of a result to be estimated.

32. Assessment of the Acceptability of the Precision Characteristics of a Method of Analysis The calculated repeatability and reproducibility values can be compared with existing methods and a comparison made. If these are satisfactory then the method can used as a validated method. If there is no method with which to compare the precision parameters then theoretical repeatability and reproducibility values can be calculated from the Horwitz equation (M. Thompson, Analyst, 2000, 125, 385-386).

33. Single-Laboratory Validation a)Consideration of harmonized IUPAC Guidelines for the in- House Validation of Methods of Analysis Government comments b)Requirements for single-laboratory validation for Codex purposes c)Validation of methods through the use of results from proficiency testing schemes

34. In situations where methods of analysis can only be validated within a single laboratory then they should be validated in accordance with IUPAC Harmonised Guidelines Methods of analysis adopted under this Regulation should be edited in the standard layout for methods of analysis recommended by the International Organisation for Standardisation

35. MEETING OF INTERNATIONAL ORGANISATIONS WORKING IN THE FIELD OF METHODS OF ANALYSIS AND SAMPLING (INTER-AGENCY MEETING)

36. Single-Laboratory Validation Electronic Compendium of Analytical Methods (e-CAM) IUPAC/ISO/AOAC INTERNATIONAL Harmonisation Programme The Continuous Assessment of Standard Methods Analysis – Effects of Proficiency Testing Blind Proficiency Testing Incorporation of Change of Methods/Method Corrections in the Codex Alimentarius Commission Harmonisation of Analytical Terminology in Accordance with International Standards Policies and Practices of IAM Members Regarding Proprietary Laboratory Techniques