Modelling of the assays involves identification of the most significant uncertainty contributors associated with each assay – as identified by a suitably experienced scientist. There is no requirement for any calculations at this point.

It is important to note that the extent of the modelling required, and the documentation associated with that, is dependent on the assay. For example, if, in the expert opinion of the scientist performing the assessment, it is considered that a suitable uncertainty assessment is possible from monitoring Internal Quality Control only, then that can be the approach taken. However, there are many situations where that will not be the case so other methods are required. In such cases, more in depth, and fully documented modelling is performed.

Recommendations are to select ONLY the 3-4 most significant contributors to uncertainty, whilst still documenting other sources as insignificant. This approach is statistically valid due to the rules of the propagation of uncertainty, and their contributions, through the uncertainty calculation. Briefly, uncertainty contributors will be insignificant if the contributors are <30% of the largest contributor as long as the contributors are uncorrelated. If the contributors are correlated contributors can be excluded if they equate to <10% of the largest uncertainty contributor. However, which ones are most significant cannot be determined until the respective standard uncertainties have been calculated. As such, all uncertainty contributors are included initially.

As with the definition of the measurand, it is recommended to complete this phase for all assays. As before, all data is assessed, and documented, using a spread sheet where the most significant contributors are identified. From these contributors the most appropriate MU approach is determined by the scientist performing the assessment.


Move on to Identifying Contributors

%d bloggers like this: