“If we need to do all this work we need to see the context that we can use it in…”

Quote: Everyone……….everywhere!

That’s a common statement in Pathology laboratories at the moment, and particularly in those that are trying to tackle Measurement Uncertainty (MU).

At a time when change is occurring seemingly from everywhere MU is just one more new thing we have to do. The Kubler-Ross change curve above describes this very nicely.

Firstly shock and denial, closely followed by the harsh realisation that it is not going away. We try to get out of doing it and then when we realise we can’t, we feel depressed about it. Finally……..eventually……….there is an acceptance that it is the way forward.

So I decided for the first post on this site to try to jump straight to the end. I want to try to put it in context and see a positive end to where we are going, and ultimately show why we should be actually positive when we think about MU.

What can we use it for?

The first and most obvious thing is accreditation. There is also the improved quality management of our services. But……we can use it for much more scientific, and clinically useful things than that.

We can use it to “put a number” on what we already know as pathology laboratory scientists. We can answer questions we get asked every day, questions like:

  • Is there a difference between Lab A and Lab B?
  • Are the results in the same lab different on different occasions?
  • What do the results actually mean in this patient?
  • Are they clinically significant?

We think best using examples so we will walk through a case study (fictional i hasten to add) where i have used some data to illustrate how we can use MU to help us answer the above questions.

The scenario:

A patient has a prothrombin Time tested at hospital A

The result is reported as 13.6 seconds

Hospital B receives a sample and has a result of 14.6 seconds reported

What does this tell us?

This only tells us that the measured result was 14.6 seconds……..

Hospital B has an MU of +/- 1 second for the PT

If the result is reported as 14.6 +/- 1.0 seconds what does this tell us now?

We measured the value as 14.6 seconds but, actually, the true value could be anywhere between 13.6 and 15.6 seconds

A Prothrombin Time result is reported as 14.6 +/- 1.0 seconds at 95% confidence

..and now

There is 95% probability that the “true” value will lie within 13.6 and 15.6 seconds

(PT reference range = 10.2 – 13.4 seconds, coincidently at both sites!)

Questions:

Are the results analytically different?

We can calculate this if we know the uncertainty of our assay and the uncertainty of the assay from Hospital A

By combining the MU from both sites we can use many different methods to assess what is analytically different (these will be covered in upcoming posts). In our example here we can conclude that the analytical difference between the two sites is 0.9 seconds

Therefore any result between 13.7 seconds and 15.5 seconds cannot be distinguished as analytically different from our result of 14.6 seconds.

The next article will show how we can use this to see if there is a clinically significant difference between the results and if there has been a change in a patient based upon two consecutive results.

 

After that we will go back to the start and show how we came up with the uncertainty estimates in the first place