Before we start looking at how we calculate Measurement Uncertainty in our assays, it is useful to review what it is and why we are being asked to do it.
Firstly let’s define a measurement. As simple as that sounds it does help make sense of what is to follow.
From the Cambridge Dictionary:
- the act or process of measuring
- a value, discovered by measuring, that corresponds to the size, shape, quality, etc. of something
The first definition seems obvious. However, the second definition gives us a bit more to think about.
“…a value, discovered by measuring…..” – This makes the distinction between the value we assign as the result of a measurement and what we would call the “true value” of what we are measuring. This is important. We cannot EVER establish the “true value” of something when performing an experimental measurement. There will be an element of error contained in all measurements.
The purpose of any measurement is to try to obtain a value as close as we can reasonably get to a representation of the true value. Now that we know this, we must make a statement of how close our measured value is to the expected true value, and how sure we are that it is that close; in essence this is what Measurement Uncertainty is.
A distinction must be made at an early stage between “error” and “uncertainty”. These are very different things.
Error is the difference between the “true” value and the measured value. Error is minimised in the laboratory by optimisation of the measurement system and calibration (when necessary).
Total error is a combination of random error and systematic error. Removal of systematic error (Bias) be it by calibration or other methods, leaves the standard deviation (or Coefficient of Variation (CV)) as the uncertainty contributor from random error.
Uncertainty tells us information about the quality of the result. It allows us to give the measured value in the context of the range we are confident of it being within. As such, how the assay “behaves” is included in the final result when MU is included. MU always associates results with a probability. It’s the probability that the “true” value is in the range we quote around the measured result.
Applications of MU
Once we have determined the MU for an assay, the results can be applied in many different ways. These range from a quality management perspective i.e. we need to do it to achieve accreditation, to the application to patient results and its ability to help us answer many questions that we encounter on a daily basis in a clinical laboratory. These questions include:
- Is there a difference between results from different laboratories?
- Are results from the same laboratory different on different occasions?
- Are results clinically significant?
Answers to some of these questions will be covered in the Examples section.