Comments on “Measurement uncertainty is not synonym of measurement repeatability or measurement reproducibility”

Of course, I agree with this statement (1) and here are some comments.

 

First, some definitions and observations.

 

1.       A repeatability condition is defined as a “condition of measurement, out of a set of conditions that includes the same measurement procedure, same operators, same measuring system, same operating conditions and same location, and replicate measurements on the same or similar objects over a short period of time”

2.       A reproducibility condition is defined as a “condition of measurement, out of a set of conditions that includes different locations, operators, measuring systems, and replicate measurements on the same or similar objects”

3.       GUM includes two sources of uncertainty both expressed as standard deviations: type A, which is characterized by measurements and type B, which is characterized by either measurements or more commonly by experience or assumptions.

4.       In the real world of commercial diagnostic assays, reproducibility is almost always larger than repeatability.

5.       It is logical to assume that the reason for #4 is uncorrected systematic effects.

6.       A reason that some effects are uncorrected (or not better corrected) is economics.

 

Now if one could take an infinite set of measurements for a diagnostic assay, would there be a difference between reproducibility and uncertainty of measurement? I maintain the answer is no, they will be the same. All of the type B effects (and type A) will be expressed in an infinite set of measurements.

 

For a shorter set of measurements, reproducibility and uncertainty of measurement will be different, although for diagnostic assays, one routinely has large sets of quality control data (reproducibility) that span relatively long times.

 

One problem with this quality control data is that it is not patient data and thus some effects cannot be sampled such as patient interferences. (Postulating them through assumptions in GUM is not easy either).

 

Another consideration is the types of effects that manifest themselves, such as calibration bias and non calibratable reagent lot effects. In principle, these systematic effects could be made smaller but aren’t since economics prevent it. Since these effects can be relatively large, given a long enough sampling time, these effects and the other unknown effects which are expressed over time will approximate an uncertainty of measurement approach although as stated above, the quality control results will never account for effects such as patient interferences.

 

So although measurement uncertainty is not synonym of measurement repeatability or measurement reproducibility, a reproducibility experiment, conducted over a long enough time will probably give a result that is similar to an uncertainty of measurement approach (save for the patient interference problem).

 

And finally, it is a lot easier to calculate a standard deviation on some quality control data, than to go through a proper uncertainty of measurement procedure.

 

References

 

1.       Paul De Bièvre Measurement uncertainty is not synonym of measurement repeatability or measurement reproducibility Accred Qual Assur (2008) 13:61–62

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: