Estimation of measurement uncertainty in food microbiology:a normative approach3rdAOAC Europe Eurachem Symposium, 3-4 March 2005, BrusselsBertrand LOMBARDAFSSA-LERQAP, Maisons-Alfort, FranceE-mail: b.lombard@afssa.fr4 March 2005 Bertrand LOMBARD 2Outline The normative approach for food microbiology Measurement uncertainty in quantitative microbiology Measurement uncertainty in qualitative microbiology Interpretation against legal limits4 March 2005 Bertrand LOMBARD 3Introduction4 March 2005 Bertrand LOMBARD 4Why ? Accreditationof laboratories Requirementsof EN ISO 17025, 5.4.6To estimate the measurement uncertainty (MU) associated to the results produced by the laboratoryIfrigourous/statistically valid calculationof MU not possible, to identify the MU components and to make a reasonnable estimation of them Implementation in France (COFRAC) for food microbiology= delayed (seeISO works)4 March 2005 Bertrand LOMBARD 5Why? (foll.) Significanceof microbiolgical analysis= direct hazard for the consumer health Quantitativemethodsin microbiology = highlyvariableNeedto quantify this variability How to express MU for qualitativedeterminations ?4 March 2005 Bertrand LOMBARD 6Thenormative approach for food microbiology4 March 2005 Bertrand LOMBARD 7The ISO structure ISO/TC 34/SC 9 TC 34 Food products SC 9 Microbiology Microbiological analysis of foods & feeds Horizontal approach4 March 2005 Bertrand LOMBARD 8Bangkok, December 2002 MU in quantitativemicrobiology= Basic approach adoptedTechnical Specification (ISO/TS)Quicker publicationUsers review& 2-year revision MU for qualitativedeterminations= In a 2ndstep4 March 2005 Bertrand LOMBARD 9Measurement uncertaintyin quantitative microbiology4 March 2005 Bertrand LOMBARD 10The approach (1) Quantitative determinations Colony-Count Techniques (+ Most Probable Number Techniques) Alternatives methods (instrumental) Decision based on a 1stseries of ISO trials (2002) Broad consensus at ISO meeting4 March 2005 Bertrand LOMBARD 11The approach(2) GUM decompositionapproach not selected MU underestimation? Heavy in food microbiology4 March 2005 Bertrand LOMBARD 12The approach(3) Global approachchosen Reproducibilitystandard-deviation (sR) on the final result of the entire measurement process In agreement withCodex Alimentarius (CCMAS)ISO/TS 21 748: 2004 Guide for use of the estimations of repeatability, reproducibility and trueness in the estimation of measurement uncertainty Establishedby ISO/TC 69/SC 6 Bridge betweenGUM & ISO 5725Global approach = special case of type A experimental estimation4 March 2005 Bertrand LOMBARD 13The main steps 2003: 1stdraft of ISO/TS 19036Microbiology of foods and animal feeding stuffs Guide for the expression of measurement uncertainty of quantitative determinations 2003/2004: 2ndseriesof ISO trials ToquantifyMU component linkedtoSamplingof thetest portionPreparationof theinitial suspension 78 participants, from 10 countries 2004: Decisionsfor final draft of ISO/TS 19036 ISO Project Group ISO/TC 34/SC 94 March 2005 Bertrand LOMBARD 14Guide ISO/TS 19036 - Presentation(1) Final DraftUnder 3-month vote (11 May)beforepublication Principles1. Global approach2. Enlarged MU = 2 sR4 March 2005 Bertrand LOMBARD 15Guide ISO/TS 19036 - Presentation(2) Principles (foll.)3. sRestimated per (consistent group of) microorganism(s) (consistent group of) matrix(ces)4.4.3 options for 3 options for ssR R Intra-laboratory sR Inter-laboratory sR(methodvalidation) Inter-laboratory sR(proficiency testing)4 March 2005 Bertrand LOMBARD 16SamplingMatrix (test portion) Initial suspensionRandom errors (repeatability)Operator/ timeBiasTest result SampleBlack boxTheblack-boxdiagramExcluded (out of theanalytical process) Excluded in general(empirical nature of microbial counts)Low levels excluded4 March 2005 Bertrand LOMBARD 17Intra-laboratory sR (1) 1st(preferred) option linked to the laboratory per se4 March 2005 Bertrand LOMBARD 18Food sample1stoperator (conditions A) 2ndoperator (conditions B)Analyse AnalyseInitial suspension Initial suspensionDifferent conditionsIntra-laboratory sR(2)Experimental protocol4 March 2005 Bertrand LOMBARD 19Intra-laboratory sR(3) The experimental protocol 10 samples per matrix type Advantage = MU at largeHeterogeneityof the sample contaminationPreparation of the initial suspension DrawbacksNeed to repeat the protocol for each type of matrix analysed by the laboratoryNeed to test naturally contamined samples4 March 2005 Bertrand LOMBARD 20Inter-laboratory sR(methodvalidation) (1) Advantages = availablevalues Drawbacks1. Conditions to meet (see ISO/TS 21 748) Laboratorys bias& precision= compatible with the methods ones All uncertainty sources (incl. test portion, sample preparation)=taken into account in the inter-laboratory trial4 March 2005 Bertrand LOMBARD 21Inter-laboratory sR(methodvalidation) (2) Drawbacks (foll.)2. Inter-laboratory sRavailable for a limited number of methods in food microbiology3. Difficulty to apply to routine analyses precisiondata obtained in limited andartificialconditions Combinaisons (matrix, strain) Annex flora (if any)4. Risk to under-estimateMU(samples homogenized and stabilized)4 March 2005 Bertrand LOMBARD 22Inter-laboratory sR(PT) Advantages Available values Large number of PT schemes in food microbiology Drawbacks1. Conditions to meet Method usedby the laboratory in PT = the same than in routine Methodusedby a sufficient number of participants PT samples routine samples2. Samples4 March 2005 Bertrand LOMBARD 23Measurement uncertaintyin qualitativemicrobiology4 March 2005 Bertrand LOMBARD 24 Preliminary works at ISO 2 possibilities foreseen CI around the Limit of Detection (LoD50) From the equivalent of reproducibility for qualitative determinations4 March 2005 Bertrand LOMBARD 25Interpretation against legal limits4 March 2005 Bertrand LOMBARD 26Legislative frame Food hygiene packageRegulation852/2004 (HI ) Draft Regulation on microbiological criteria (SANCO/4198 rev. 14, Dec. 2004) A microbiological criterion= a qualitative/quantitative limit +4 March 2005 Bertrand LOMBARD 27Food hygiene controls For operators own checks= MU not taken into account For official food controls= ?the same ruleor (Figure 2)ocase (ii) for indicatorsocase (iv) for pathogensTo be precised into EC Guidelines, to come4 March 2005 Bertrand LOMBARD 284 March 2005 Bertrand LOMBARD 29Conclusion4 March 2005 Bertrand LOMBARD 30 Global approach Pragmatic, adapted to the complexity ofFood analysisMicrobiological analysis In agreement withGUM principlesInternational references and rules MU estimation becoming widespread in food microbiology= Towards a more scientific analysis?