The corresponding section under chemical laboratories discusses what analysts need to know about uncertainty measurements and about the principles and basis of estimating uncertainty. This document describes the application of uncertainty measurement in microbiological testing.

When repeated measurements are made on an analyte using an identified test method, there can be a dispersion of measured values attributed to the quantity that is being measured. In making estimations it is possible to think of a "true" value for the analyte in the matrix. Individual estimations however, will not show the "true" value. The difference between the "true" value and the observed value is described as the error in estimation. The uncertainty indicates how big the error might be.

Some of the key features that need to be considered in estimating uncertainty during microbiological analysis are identified in the document European Co-operation for Accreditation EA 4/10 as:

• Microbiological tests generally belong to the category of tests that preclude rigorous, metrologically and statistically valid calculation of uncertainty.

• It is generally appropriate to base estimation of uncertainty on repeatabilityand reproducibility data alone, but ideally including bias (from PT results).

• Some individual components like pipetting, weighing and diluting may be evaluated to demonstrate their negligible contribution to overall uncertainty.

• Sample stability and sample preparation cannot be measured directly and evaluated statistically. However, their importance should also be considered.

On the basis of the above, the best approach a microbiological testing laboratory could take is to first identify and document the causes of uncertainty of measurements associated with testing. There are causes among those listed below that cannot be handled in the laboratories with their present level of practice. They need to be understood. The laboratory then needs to take steps to estimate the measurement uncertainty associated with activities that could be controlled and devise a means to minimize these uncertainties.

The more important activities leading to uncertainty of measurement in microbiological testing are,

• Size of the original sample from which sub-samples are taken for testing

• Sampling and sub-sampling in selection of the required quantity of matrix for testing

• Time taken to homogenize the sample with diluents in a blender

• Volume of homogenate used for blending

• The volume of diluents used in serial dilution of the blended sample

• The number of dilutions carried out before subjecting the sample to tests

• Variations in incubation temperature and time

• Counting of the colonies

• Techniques of preparation of microbiological media and reagents

• The sterilization temperatures, the time and the total duration of exposure of media to heat

It is possible for an analyst to identify the uncertainties arising from the above and take steps to minimize the degree of uncertainty. The following factors are not used in estimation of uncertainty in microbiological testing as there are no means to address them.

• Non-homogenous distribution of analytes in food

• Sample preparation and sample stability

• Plate readings of less than 100 CFU / ml

• Qualitative results [presence / absence]

• Methods needing additional step for confirmation

• Methods based on counting positive tubes / cells (MPN, micro-titer plates )

The estimation of uncertainty in microbiological measurements is based mainly on the repeatability and reproducibility data generated by intra-laboratory and inter-laboratory testing. Proficiency testing makes a very important contribution to this by establishing the total bias of the results. The following table gives the routine analytical activities carried out in a microbiology laboratory in generating data to in quality management and their relationship to measurement uncertainty through repeatability and reproducibility data.

Information to establish measurement uncertainty could be generated partly along with testing and partly by special activities. Some of the more important information to work on is:

• Information on intermediate precision generated continuously along with sample testing

• Information from proficiency testing and inter-laboratory comparisons

• Data related to accuracy and precision at the individual steps of testing

• Data generated from quality control information during testing

The concept of intermediate precision is established in relation to two other levels of precision that may be observed in the performance of a laboratory, namely repeatability precision and reproducibility precision. In microbiological testing, comparison of test results for highly homogenous samples may be carried out by repeatability studies, where estimations are done under closely monitored conditions such as the same analyst working on the same day using the same equipment. This is described as repeatability precision. Reproducibility precision may be achieved when a similar set of samples is tested in different laboratories. These are the two possible and reasonable extremes encountered in testing. The term ‘intermediate precision’ is used to describe a set of test results generated within a laboratory by different analysts, on different days using different equipment for the same sample. The intermediate precision represents a more realistic analysis situation and the data could be used to form the basis for establishing measurement uncertainty in microbiology.

The final measurement uncertainty in analysis is a combination of uncertainties arising from each step of the testing activities. An approach that could be taken here is to calculate the uncertainty budget for each step so that the values can be pooled and the distribution pattern of values in relation to each analytical activity taken into consideration. The results generated over a period during testing operations could be used to calculate the means and standard deviations in relation to each of the activities. The data could be used to understand the extent of uncertainty and develop practices to minimize its level. Some of the activities that the laboratory could get involved in are: