Go into hospital nowadays, and you will do well to escape without having a blood test of some sort. Very often these are routine tests, which give doctors an overview of the state of play. There might be a few wayward figures here or there – but the doctors will ignore them, or explain them away as part of the normal variation of homeostasis.
In the PMJ this month the spotlight turns to one biomarker that is commonly requested when patients are admitted to hospital. Indeed, the troponin is one test which I see regularly used completely out of context, and providing information which is often difficult to assimilate into the clinical picture. The paper – an analysis of >11000 admissions to a large medical facility in Dublin, Ireland has examined troponin results for all admissions under the medical (but not cardiology) service from January 2011 to October 2012.
Now, the troponin is a test that has undergone a change over the time that it has been available to clinicians in everyday practice. I can remember taking serial CKs in patients with suspected myocardial ischaemia, and my joy at the troponin becoming available for use in my potential CCU patients. I can also remember the many patients who have been admitted to hospital for 12 hours just to see what their troponin will be – a clear case of a biomarker dictating practise, rather than been a tool for me to use. And I have many memories of strained conversations with colleagues about the meaning of a mildly raised troponin which had been requested as part of a bundle of tests at the point of admission – without any real thought being given to how one might interpret the results.
These strained conversations have altered in tpne over the years as the blind faith in the value of troponin to indicate ischaemic heart disease which accompanied the hype of the test when it was first released, has been eroded by realisation that troponin is no way near as specific as we were once led to believe – and interpretation now requires quite a lot of Bayesian reasoning to clear the waters.
The article looking at troponin tests on the acute medical take makes a fascinating read, and helps provide some data to the consideration of the not uncommon problem – “well what do I do with this result now?”
The answer in the case of an unexpected elevated troponin is to consider the overall clinical context, and attempt to understand where the physiological stress has proceeded from, as this study shows a significant association between elevated troponin and mortality:
So – a helpful paper looking at a common clinical scenario, and providing a fairly robust argument for how to approach the problem.
But one of the most fascinating parts of this analysis is the determination of what is ‘normal’ and why do we love to have such binary answers to complex questions?
The manufacturers of the assay employed recommend a cut-off of 14ng/L for the normal range. But, given that the test isn’t as specific for myocardial injury as they would like – a figure of ≥53ng/L should be used to indicate myocardial ischaemia. For the purposes of the published study a figure of <25ng/L is used as the cut-of for normal, and ≥25 as ‘positive.’
The persistence of a desire to classify a test result that the outcome of this large observational study indicate is a sliding scale, indicating physiological stress, rather than any specific disease process (in this study that effectively excluded cardiac disorders as the presenting complaint) into normal and abnormal categories belies a huge cognitive bias that we all carry around with us. Essentially we like to make judgements based on prior experience, heuristics, and easily interpreted chunks of information – what David Khaneman would call a ‘System 1″ or ‘fast” process. We do this regularly with a high degree of accuracy when on the acute take.
What this paper could be seen to do is boil down a clinical problem into another readily available answer, that can be applied in everyday practice – to me, it is a reminder of the blind faith I used to have in a test that I and it’s manufacturers understood poorly, and drove clinical protocols and pathways, rather than me applying some critical thinking to my actions, and their results – and using the test to its best effect. I wonder how many more biomarkers we will see undergoing this sort of evolution.