Normal ranges are papered to the door of almost every clinical medical student’s lavatory door or fridge, inside the cover of every notebook in the wards – accompanying every result on the EHR – everywhere we are told confidently what normal is. But as this paper studying the laboratory findings of several thousand inpatients at a hospital in North London highlights – ‘normal’ is not as clear cut as it may initially seem.
A paper from the hospitals looked at in this study was the subject of a previous blog which highlighted the variation in practice and often poor implementation of ivestigations into the cause of low sodium values in patients acutely admitted to the three hospitals involved.
This paper has taken a signal from a previous one and has now produced data that questions the validity of the 135-145 range for serum sodium.
The authors noted during their previous studies that many of the patients acutely admitted to the hospital had low sodium results, whilst a cohort of patients from care homes had higher values, and seemed to be dehydrated. The mortality for patients being admitted rose with increaing sodium concentrations – but the break-point in the graph was within the normal range. So we have a population whose results don’t fit the ‘normal’ range, and a ‘normal’ range that seems associated with increasing mortality:
Clearly these retrospective observational studies shouldn’t have lab managers running around redefining normality and encourage us all to drive our patients’ sodium to the lower half of normal in an attempt to save lives…
BUT and it is a big but that deserves capital letters – we do need to work out who defined normality. Thankfully Prof McKee and his colleagues have done a bit of digging for us and give a potted history of the normal range for sodium measurement. And it turns out that this range – embedded in millions of memories the world over is actually based on comparatively few data points – the first papers used about a hundred healthy volunteers using flame photometry – a technology that is largely superceded by more accurate methods. The subsequent studies they refer to us up to a 1000 measurements (often in multiple sub-groups) from which they drew their conclusions.
How can this be? Surely we don’t just take decades old evidence and allow it to heavily influence our treatment plans, delay discharges and so on?
In this case the answer seems to be… yes. However, this is not the only sphere of medicine where old data continues to heavily influence current practice.
Oxygen is one of the most commonly administered, but not prescribed, drugs in the formulary. In COPD it is one of the few drugs that has evidence for influencing mortality, rather than simply altering a trajectory of decline…
And the evidence for this? It is predominantly based on an MRC funded study from the late 1970s that included 87 patients. That evidence was enough to change practice, and alter lives I am sure, but it probably would not stand up to scrutiny for the basis of a major shift in practice nowadays. The linked paper on sodium measurements, for example looks at more than 100000 samples and trials of therapy in COPD looking to demonstrate a mortality benefit now need to have thousands of patients (the TORCH trial enrolled 6200)
So what is truly normal, are any of our favourite ‘common sense’ treatments justified in modern medicince, do we do anything right in our every day practice?
Clearly yes, there have been huge improvements in survival from many diseases over the decades, and common medical practices are clearly successful at identifying pathology, seeking out the underlying disease, and then targeting that. However, when confidently stating that something is the correct strategy to pursue, we should also be mindful that our convictions might just be based on less than solid ground. And this uncertainty is at the heart of a healtyh academic examination of our medical practice on a daily basis.
We should not be paralysed by doubt, but we should have a healthy degree of scepticism when appraising both existing practices (the PANTHER IPF trial is perhaps one of the most significant turnarounds of recommended practice triggered by high quality trial evidence) and when new technology comes along (see this blog on troponins in acute medicine.)
So next time you are on a ward round, and find yourself struggling to guide a patient towards ‘normal’ for a biochemical test, or some other finding that we all ‘know’ to be true – you should perhaps make a mental note and work out from the evidence if all we are doing is tilting at windmills, because that is what we have always done, or if there is a genuine reason to strive for that particular outcome.