Once again, a Twitter dialogue has started me thinking and hastened me to another blog post. This time it’s around automation and the removal of the clinical check by a pharmacist.
https://twitter.com/PharmacyInnLab/status/1521792025334583298?t=-I6jreKkE1Yn5uWuvlSHdA&s=19
There’s an interesting back and forth that follows this initial tweet. But a fellow pharmacist raised concerns about the missing clinical check, and early attempts to use ML/AI in supporting pharmacists to undertake the ‘most appropriate’ (my words) clinical check.
I think it is well known that I am pro-technology, but I am also very much focussed on patient safety too (as a clinical safety officer it is a prime concern). So I do think there may be a place for technology to support healthcare processionals (HCPs) and in this case, around the clinical checks of prescriptions, but we need to be certain that it is safe for us to do so.
The arguement for recent developments being, if we see the same items for the same patient, treating the same conditions, with no changes and no additions - in fact a carbon copy of the previous prescription, then what does the pharmacist bring to the clinical check at that point? Bear with me on this one before you starting citing your reasons against my statement.
Could we base the clinical appropriateness of the current prescription on recent, multiple previous occurrances?
I genuinely believe that Machine Learning/Artificial Intelligence will go some way to remove the repetitive and low value inputs of the pharmacist in to the clinical checks.
But before we get carried away with thinking we are freeing up lots of pharmacist time, and resource - think again.
The medicine supply journey, from prescribing intention to actual receipt by the patient is a complex and interconnected one. The separate parts are owned and controlled by separate professionals and organisations, each with their own safety profiles and mitigations. The only contestant in the ‘process flow’ appears to be the patient.
So if we consider our own little pharmacy section of the process, and start to amend the paramaeters in which we work, we make some assumptions on the inputs and output of our own little world.
So a pharmacist might consider that a prescriber has reviewed a patient’s notes, confirmed a lack of changes in their condition, and therefore the continued issue of the prescription is appropriate. Of course the bloods have been done, and of course the hosspital investigations considered.
But in the meantime, perhaps the GP is utilising new prescribers within their team, utilising remote consultations and triaging patients - I mean, they can rely on the pharmacist who recieves their prescriptions to be clinically checking their prescribing, and reporting back their concerns - even though this is the third repeat of the same batch of prescriptions for the patient.
So to try to come to a conclusion before I lose your attention - what I see is a complex system, with interconnect parts, evolving independently but having an impact on each other.
So I see ML/AI supporting the pharmacist to make more appropriate clinical assessments, and providing the time and mental space to have good conversations with prescribers and patients alike, to make the supply of medicine as safe as possible.
So my ideal future is the AI alerting me to the fact that Mr Malik has had 5 supplies of his salbutamol in the last 6 months but only 2 supplies of his Clenil. Or that the ordering of Mrs Jamroz’s Metformin MR doesn’t align with the current dosage she’s prescribed, oh and by the way her treatment doesn’t align with current NICE guidance in terms of her type 2 diabetes.
It will reassure me that the prescriptions it has passed for dispensing meet the concordance criteria, and U&Es and TFTs are seen in the additional information of the Electronic Health Record.
I think we are a way off from that point, BUT, we need to start having the professional discussions around the topic, start formulating some guidance or standards to how we use this technology - because at the moment, it is a bit of a blank sheet.
Current guidance on ML/AI in pharmacy