Hello - I'm interested in understanding whether there can be justifications for not evaluating uncertainty measurements as a testing lab. I understand there is the option not to report it unless relevant, but I'm wondering if it needs to be done at all.
The testing that would be performed in my lab does not produce pass/fail outputs or those that meet a certain requirement/range of acceptance. They are typically to see how a customer material might react with a released product when run in an assay. Could we incorporate the risk process into this for factors such as user error, material handling, etc. and showing that they are in control rather than trying to determine a specific measurand?
There are several points to note regarding the requirements. Firstly irrespective of whether the testing is reported as a pass/ fail result or not, all laboratories must identify contributions to the measurement of uncertainty (MU). Secondly, a distinction is made between identifying contributions and evaluating measurement uncertainty, which involves quantifying the MU. The standard states in a note that when a well-recognized test method is used (i.e. standard methods) and controls are in place to limit the major sources of uncertainty, then the laboratory need not evaluate measurement uncertainty. This applies to both qualitative and quantitative results. Bear in mind that even if a result is reported qualitatively, for example, a positive or negative for a diagnostic test, it is still necessary to identify contributions to uncertainty to ensure that they are being controlled sufficiently. Typically a laboratory will state that a standard method is being used and performance parameters are met (verified); therefore, measurement of uncertainty wasn't fully evaluated. However, they need to list the major contributions to uncertainty, such as temperature affecting the volume of solutions in volumetric glassware and the uncertainty of calibrated mass balances. These controls can be considered as measures to minimise risk as part of addressing the risk that results may not be fit for purpose.
If the test involves a measurement rather than an observation, and the method is not a validated standard method, the measurement of uncertainty must be fully evaluated.