Outcome from result forms (first round): measurement uncertainty Marina Patriarca Istituto Superiore di Sanità Dpt of Public Veterinary Health and Food Safety CRL-ISS Workshop 2008 on Proficiency Tests Rome (Italy), 17 October 2008 6th WORKSHOP PROFICIENCY TESTING IN ANALYTICAL CHEMISTRY, MICROBIOLOGY AND LABORATORY MEDICINE Current practice and Future Directions Rome (Italy) 6-7 October 2008 Organised in co-operation with Chair of the Organising Committee: Dpt PVH FS, ISS www.iss.it/eurachem CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Issues • estimating uncertainty: which approach? – Affordable? – Reliable? – Fit for purpose? • assessing compliance taking uncertainty into account: which rules? CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Harmonisation? y . g 7 o 0 l 0 o 2 r : t 9 e 9 M e f d d o i n u a y r G s a t l O p u e S b I c a n c o o C V l l a a r s n e o m i n r t e e a G T n r d d Inte sic an ociate Ba Ass (VIM3) freely available from www.bipm.fr CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Measurement Uncertainty, u(x) Parameter, associated with the result of a measurement, that characterizes the dispersion of the values that could reasonably be attributed to the measurand ISO, VIM 2: 1993 (3.9) Non-negative parameter characterising the dispersion of the quantity values being attributed to a measurand, based on the information used (ISO, VIM 3: 2007, 2.26) CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Measurement result 2.9 (VIM3) set of quantity values being attributed to a measurand together with any other available relevant information 3.1 (VIM2) value attributed to a measurand obtained by measurement CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Measurement result NOTES … 2 — A measurement result is generally expressed as a single measured quantity value and a measurement uncertainty. … CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 12th PT - Lead in Milk - Sample A 0,05 Pb in milk, mg/kg 0,04 0,03 0,02 0,01 0 ab cd gh ed gg ff yy mm dd kk bb ww aa rr jj af ag aq Laboratories CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 pp Relevant documents • ISO. 1993. Guide to the expression of uncertainty in measurement. (GUM) • EURACHEM/CITAC Guide. Quantifying Uncertainty in Analytical Measurement (QUAM). 2000. www.eurachem.org • Eurolab Technical Report No 1/2007. Measurement uncertainty revisited: Alternative approaches to uncertainty evaluation. www.eurolab.org CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 A summary of approaches to MU Definition of the measurand List of uncertainty components Intralaboratory Yes Mathematical model? Evaluation of standard uncertainties Interlaboratory No Organisation of replicate measurements Method validation CRL-ISS Workshop on PT 2008 Yes PT or method performance study? Method accuracy ISO 5725 ISO TS 21748 No Proficiency testing ISO Guide 43 ISO 13528 Rome (Italy), 17 ottobre 2008 Intralaboratory approaches: Modelling The model of the measurement process (independent variables) A × Cs V f C= × As Vi e.g. or: y = f(x1, x2, …xi) The uncertainties on each variable (influence quantity) e.g. u(V), u(A), u(xi) can be combined uc ( y ) = CRL-ISS Workshop on PT 2008 N ∑ i =1 ∂f ( xi ) u 2 ( xi ) ∂x1 2 Rome (Italy), 17 ottobre 2008 Modelling approach for analytical chemistry: drawbacks • appropriate models? • measurement standards/methods? • can all effects be individually identified and quantified (e.g. matrix variability)? • how reliable is the MU estimate? CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Intralaboratory approaches: Method validation studies/IQC (1) Information from method validation studies: the best available estimate of overall precision, u(P) the best available estimate of overall bias (Rm) and its uncertainty u(Rm): calculated from the analysis of CRMs or the recovery of spiked amounts of the pure analyte; the best estimate of other components of uncertainty associated with effects incompletely accounted for in the development and validation study, e.g. residual matrix effects assessed from the analysis of matrix-matched EQAS samples, which target values are not accompanied by uncertainty statements, u(xi). All terms expressed as relative standard deviations (i.e relative standard uncertainties) and combined according to the uncertainty propagation law CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Intralaboratory approaches: Method validation studies/IQC (2) Preliminary requirements 1. Quality assurance measures concerning: staff measurement procedures measuring equipment standards and matrix-matched reference materials internal quality control / control charts 2. Any significant bias should be eliminated, corrected or otherwise taken into account CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Uncertainty components from in house validation studies integrazione del segnale taratura A Bias interferenze u(Rm) della matrice efficienza della nebulizzazione efficienza della nebulizzazione efficienza della atomizzazione Precision u(P) ripetibilità T T Vf T operatori taratura Vi operatori riproducibilità taratura T riproducibilità 2 A × Cs V f C= × As Vi ripetibilità T T 2 riproducibilità ripetibilità u(C ) u( xn ) u ( P) u( Rm) = + + ... + C ripetibilità C Rm C taratura taratura diluizione operatori 2 C T riproducibilità operatori operatori T riproducibilità incertezza sul titolo MR Other components: u(xi) ripetibilità e.g. sampling, matrix inhomogeneity, robustness integrazione Cs As del segnale efficienza della atomizzazione efficienza della nebulizzazione Interlaboratory approaches Method performance studies ISO/TS 21748 provides guidance for the use of information from collaborative interlaboratory studies on the performance of a defined and detailed measurement procedure, designed and carried out according to ISO 5725:1994: U = 2sR ISO/TS 21748 Guidance for the use of repeatability, reproducibility and trueness estimates in measurement uncertainty estimation. ISO, Geneva, 2004. CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Interlaboratory approaches Proficiency Testing (1) Use results of PT – as a check on the evaluated uncertainty – to obtain/assess information on bias and its uncertainty – as a basis for a good estimate of the uncertainty arising from those parts of the measurement procedure within the scope of the scheme [under stated conditions & provided that systematic deviation and any other sources of uncertainty are also taken into account] CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Interlaboratory approaches Proficiency Testing (2) • An approach based solely on PT data is not recommended, however: – PT/EQAS requiring laboratories to adopt the same or very similar methods provide a rough indication of MU as: u = SD PT / EQAS _ one _ method – preliminary evaluation of MU as the SD of the differences between the laboratory reported values and the assigned values in repeated rounds CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Which coverage factor? • • • • • • 11 labs: 2 3 labs: did not report U 2 labs did not report k 1 lab: 0.8346 1 lab: 2.306 1 lab: 3.18 CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Expanded uncertainty U=ku Which coverage factor? CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Choosing K (1) Issues to be considered – The level of confidence required – Knowledge of the underlying distributions of values – Knowledge of the number of values used to estimate random effects CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Choosing K (2) • For most purposes it is recommended that k is set to 2. • This value of k may be insufficient where the combined uncertainty is based on statistical observations with relatively few degrees of freedom (less than about six). The choice of k then depends on the effective number of degrees of freedom. CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Choosing K (3) k = 2, why??? – The level of confidence required • 95% – Any knowledge of the underlying distributions of values. • normal – Any knowledge of the number of values used to estimate random effects • always >6, unless based on strict modelling approach CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 D.1.3. Measurement uncertainty The analytical result shall be reported as x ± U whereby x is the analytical result and U is the expanded measurement uncertainty, using a coverage factor of 2 which gives a level of confidence of approximately 95 % (U = 2u). COMMISSION REGULATION (EC) No 333/2007 CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Is my MU estimate reliable? 0,05 Sample A solid line: Avg A, 0.025 mg/kg dotted lines: Uf A, 0.0054 mg/kg Pb in milk, mg/kg 0,04 0,03 0,02 0,01 0 ab cd gh ed gg ff yy mm dd kk bb ww aa rr jj af ag aq Laboratories CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 pp C.3.3.2. ‘Fitness for purpose’ approach Maximum standard measurement uncertainty (Uf) Uf = (LOD/ 2) + (αC) 2 2 LOD = limit of detection of the method (µg/kg); C = the concentration of interest (µg/kg); α = numeric factor - for C<50 µg/kg, use 0.2 COMMISSION REGULATION (EC) No 333/2007 Lead in milk CRL-ISS Workshop on PT 2008 Uf = (4 / 2) + (0.2 × C) 2 2 Rome (Italy), 17 ottobre 2008 Is my estimate of MU fit for purpose? 0,02 Lead in milk - Sample A Lead in milk - Sample B U, mg/kg 0,015 Uf A, 0.0054 mg/kg Uf B, 0.0066 mg/kg based on avg conc 0,01 0,005 0 ab cd gh ed gg ff yy mm dd kk bb ww aa rr jj af ag aq Laboratories CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 pp Interpretation of results CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Maximum levels COMMISSION REGULATION (EC) No 1881/2006 of 19 December 2006 setting maximum levels for certain contaminants in foodstuffs Section 3: Metals 3.1 Lead 3.1.1 Raw milk ( 6 ), heat-treated milk and milk for the manufacture of milk-based products 0,020 (mg/kg wet weight) CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 D.2.1. Acceptance of a lot/sublot The lot or sublot is accepted if the analytical result of the laboratory sample does not exceed the respective maximum level as laid down in Regulation (EC) No 1881/2006 taking into account the expanded measurement uncertainty and correction of the result for recovery if an extraction step has been applied in the analytical method used. COMMISSION REGULATION (EC) No 333/2007 CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 D.2.2. Rejection of a lot/sublot The lot or sublot is rejected if the analytical result of the laboratory sample exceeds beyond reasonable doubt the respective maximum level as laid down in Regulation (EC) No 1881/2006 taking into account the expanded measurement uncertainty and correction of the result for recovery if an extraction step has been applied in the analytical method used. COMMISSION REGULATION (EC) No 333/2007 CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Harmonised decisions? 0,05 N Y N Pb in milk, mg/kg 0,04 N N Y N N Y N N Y 0,03 Y Y 0,02 Y 0,01 Y 0 ab cd gh ed gg ff yy mm dd kk bb ww aa rr jj af ag Laboratories CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 aq pp Guidance documents • ILAC-G8:1996 Guidelines on assessment and reporting of compliance with specifications • Eurachem/CITAC Guide. Use of uncertainty information in compliance assessment. (2007) www.eurachem.org • DG SANCO Report on the relationship between analytical results, measurement uncertainty, recovery factors and the provisions of EU Food and Feed legislation, with particular reference to Community Legislation concerning - contaminants in food (Council regulation (EEC) no 315/93 of 8 February 1993 laying down Community procedures for contaminants in food) - undesirable substances in feed (Directive 2002/32/EC of the European Parliament and of the Council of 7 may 2002 on undesirable substances in animal feed) CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 ILAC G8 - Criteria for inerpretation of results reject 120 Accept/reject only if requirements expressed as < ; > accept 100 80 1 2 3 4 5 accept/reject with a less than 95% level of confidence CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 EURACHEM/CITAC Guide - Decision rules • Prescriptions for the acceptance or rejection of a product based on the measurement result, its uncertainty and the specification limit or limits, taking into account the acceptable level of the probability of making a wrong decision. • “Acceptance” and “Rejection” Zones • A decision rule that is currently widely used is that a result implies non compliance with an upper limit if the measured value exceeds the limit by the expanded uncertainty. CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008 Recommendation Report on the Relationship Between Analytical Results, Measurement Uncertainty, Recovery Factors and the Provisions of EU Food and Feed Legislation, In practice, when considering a maximum value in legislation, the analyst will determine the analytical level and estimate the measurement uncertainty at that level. The value obtained by subtracting the uncertainty from the reported concentration, is used to assess compliance. Only if that value is greater than the maximum level in the legislation is it certain “beyond reasonable doubt” that the sample concentration of the analyte is greater than that required by the legislation. CRL-ISS Workshop on PT 2008 Rome (Italy), 17 ottobre 2008
© Copyright 2025 Paperzz