How to estimate your measurement uncertainty (MU) that is related to quantitative determinations in microbiology of the food chain? You will find answers to your questions on this complex topic in the presentation of the approach of the recently revised standard ISO 19036:2019. A link to an Excel-based tool is also provided to support you in implementing the calculations of this International standard.
ISO 19036 'Microbiology of the food chain — Estimation of measurement uncertainty for quantitative determinations' has been prepared by WG 2 Statistics of ISO/TC 34/SC 9 and was published in October 2019, replacing ISO/TS 19036 (2006) and its Amendment 1 (2009).
It is mainly applicable to the enumeration of microorganisms using a colony-count technique, but also applies to other quantitative analyses, including most probable number (MPN) techniques, instrumental methods and molecular methods.
This International standard has adopted an approach to estimate MU considering three uncertainty components:
- technical uncertainty;
- matrix uncertainty;
- distributional uncertainty.
The uncertainty contribution from systematic effects (bias) is not included in ISO 19036: in food chain quantitative microbiology and in particular with conventional microbiology techniques, assigned values or reference quantity values are usually not available so it was considered that bias cannot be reliably estimated.
The three uncertainty components
— Technical uncertainty
This component is estimated by the reproducibility standard deviation on the final measurement result, with three options in a decreasing order of priority: intra-laboratory reproducibility standard deviation, inter-laboratory reproducibility standard deviation from a method validation study and inter-laboratory reproducibility standard deviation from a proficiency test. This component is linked to the method as used in a given laboratory and the first option is the same parameter used in ISO 16140-3 for the verification of quantitative methods.
— Matrix uncertainty
This component arises from heterogeneity of matrix contamination, resulting in variability of microbial levels between test portions, which can be large for solid matrices and composite food products. For each kind of matrix, matrix uncertainty is taken as a fixed value for homogeneous matrices, or estimated by a repeatability experiment, or derived from historical values.
Taking into account that the matrix uncertainty is regarded as being independent from the target microorganism and the analytical method used, values produced by a given laboratory can be used by others.
Thus, some matrix uncertainty values have been produced:
- in a French study. You will find the context of the French approach, the flow chart to assist in the use of matrix uncertainties and the database itself.
- in a study organized by three European Union Reference Laboratories, EURL for Campylobacter, Coagulase Positive Staphylococci and Listeria monocytogenes. You will find the report of the collaborative study organised in 2021 together with a database with values and a summary of the values.
— Distributional uncertainties
These components derive from the random distribution of microorganisms. According to the feature of the analytical method, they are calculated mathematically in three potential cases:
- for colony-count techniques: Poisson uncertainty (significant at low levels) and confirmation uncertainty (when the method includes a partial confirmation step);
- for MPN techniques: MPN uncertainty.
MU estimation
ISO 19036 proposes two options for estimating and reporting MU:
- complete option: the technical, matrix and distributional uncertainty components are estimated separately from each other, then combined to calculate MU.
- simplified option: a general MU value may be reported as based only on technical uncertainty, if consistent with laboratory protocols and client requirements.
An Excel-based tool to implement the calculations of this International standard
WG 2 has verified an Excel-based calculation tool to implement the calculations described in ISO 19036: link here.
This Excel-based program contains:
- detailed instructions for use,
- a worksheet to calculate intra-laboratory reproducibility (technical uncertainty),
- a worksheet to calculate repeatability (matrix uncertainty),
- and a worksheet to calculate combined uncertainty, including the two first uncertainties as well as distributional uncertainties.
Supporting material
To facilitate the understanding and promote an harmonised implementation of ISO 19036:2019, a presentation support has been prepared by ISO/TC 34/SC 9/WG 2 in English and is available here (PDF or Power Point).
Although the utmost care has been taken with the development of this presentation support, errors and omissions cannot be entirely excluded. Members of ISO/TC 34/SC 9/WG 2 therefore accept no liability, not even for direct or indirect damage, occurring due to or in relation with the application of this presentation support. In case there are differences between the content of this supporting material and ISO 19036, the content of this International Standard always prevails.