Read on Twitter
BING BONG NEW PAPER
NHS hospital medicines data is being collected, but then needlessly withheld from analysts.
This means patients are exposed to sub-optimal treatment. And NHS money is being wasted.
The barriers are NOT technical. They are "people". https://www.bmj.com/content/370/bmj.m2607.full?ijkey=nIHRXr5GpawscdH&keytype=ref
NHS hospital medicines data is being collected, but then needlessly withheld from analysts.
This means patients are exposed to sub-optimal treatment. And NHS money is being wasted.
The barriers are NOT technical. They are "people". https://www.bmj.com/content/370/bmj.m2607.full?ijkey=nIHRXr5GpawscdH&keytype=ref
THREAD
Hospitals are where the most expensive drugs are used, and where new drugs are used.
Hospitals are therefore the place where you will find the biggest variation in quality, safety and cost effectiveness of prescribing.
So the biggest opportunities to do good with data!
Hospitals are where the most expensive drugs are used, and where new drugs are used.
Hospitals are therefore the place where you will find the biggest variation in quality, safety and cost effectiveness of prescribing.
So the biggest opportunities to do good with data!
Don't believe me?
The NHS primary care drugs budget has stayed steady at about £9bn a year for half a decade.
But the NHS hospital medicines budget has very quickly DOUBLED, from £4.2 to £8.3 billion in just six years. https://www.bmj.com/content/370/bmj.m2607.full?ijkey=nIHRXr5GpawscdH&keytype=ref
The NHS primary care drugs budget has stayed steady at about £9bn a year for half a decade.
But the NHS hospital medicines budget has very quickly DOUBLED, from £4.2 to £8.3 billion in just six years. https://www.bmj.com/content/370/bmj.m2607.full?ijkey=nIHRXr5GpawscdH&keytype=ref
That's a lot of money you say!
Yes it is... in fact the medicines spend is about 14% of the entire NHS budget.
Now stay with this thread, it gets good from here. Come into the labyrinth of NHS data flows...
Yes it is... in fact the medicines spend is about 14% of the entire NHS budget.
Now stay with this thread, it gets good from here. Come into the labyrinth of NHS data flows...
Primary care GP data is routinely shared since forever, aggregated at practice level, but still amazingly detailed. It's been open data for ten years, with no hiccups, creating a vast mountain of data science startups, research papers, and tools like our http://OpenPrescribing.net
Hospital medicines usage data... has not been shared.
Here it gets interesting. This is often discussed as a TECHNICAL problem.
But it's not.
Hospital medicines usage data is collected, nationally aggregated, normalised, through AT LEAST two systems.
Not sharing is a CHOICE.
Here it gets interesting. This is often discussed as a TECHNICAL problem.
But it's not.
Hospital medicines usage data is collected, nationally aggregated, normalised, through AT LEAST two systems.
Not sharing is a CHOICE.
First, there's a commercial system, collecting data to help pharma companies market their medicines (fine). But as we explain in our (painfully detailed) paper, the NHS contract with this supplier prevents the NHS seeing the data properly, or sharing it https://www.bmj.com/content/370/bmj.m2607.full?ijkey=nIHRXr5GpawscdH&keytype=ref
So, to be clear, pharma companies are allowed to see what each hospital is prescribing, each specialist unit, etc, in order to market at them. But http://OpenPrescribing.net & other analysts are not allowed to see the same data to help improve quality, safety, and cost effectiveness
But maybe we think that's ok: private companies, private money, private purposes, private data. Sure thing.
HOWEVER.
This hospital medicines data is also collected, aggregated, normalised, nationally, across the NHS, in an NHS funded system...
HOWEVER.
This hospital medicines data is also collected, aggregated, normalised, nationally, across the NHS, in an NHS funded system...
This system is paid for by individual NHS Trusts. Where is the data? What's done with it? All the contracts are secret. So are the dashboards. And nobody sees the raw data! Hospitals are allowed to see some "dashboards" made by the company, but only Trusts can see their own data.
Under emergency Brexit plans NHS England paid extra to see a few bespoke views onto the data. Nobody knows what these are. Why does this matter? Because analytic windows and dashboards are not the same as raw data. They can be done well or badly. You need to check working.
Otherwise you get things like this. The company collecting data and selling dashboards claims it has saved one trust £110 000 a year. But... No clear explanation of the methods used, nothing to identify the trust where this saving was reportedly made, no access to the data used..
This is the equivalent of me saying "my new drug cures cancer er no i won't show you how i found that out". It is INCONCEIVABLE that you'd be able to make assertions with no supporting evidence in healthcare research. Why is operational research different? https://journals.sagepub.com/doi/full/10.1177/0141076820930666
Another. NHS Improvement used this data and claimed they'd saved the NHS “£324 million in one year” on prescribing costs. No description of the analytical methods. They later shared an online webinar and one page diagram, after we raised concerns, giving minimal methods info.
This analysis is done behind needlessly closed doors. Its quality and character are unknown. It cannot be critically reviewed and improved by others with complementary knowledge and expertise in data science or prescribing.
So this closed work is HIGHLY likely to miss opportunities for improving patient care.
Conversely, if the work is excellent, nobody can learn from the methods.
This is senseless.
Conversely, if the work is excellent, nobody can learn from the methods.
This is senseless.
(I'm nearly done).
How do we know that it's worth sharing this aggregated hospital-level medicines usage data? Because the aggregated GP practice-level data has supported a huge ecosystem of data science startups, research, and quality improvement projects. Some examples:
In primary care this data has been used to identify whole new categories of cost savings for the NHS
https://bmjopen.bmj.com/content/8/2/e019643?ijkey=fc7009ec5c01b6093f2b0605e590d2395e7e3221&keytype2=tf_ipsecsha
With robust evidence that these savings were actually achieved https://www.jmir.org/2019/1/e10929/
https://bmjopen.bmj.com/content/8/2/e019643?ijkey=fc7009ec5c01b6093f2b0605e590d2395e7e3221&keytype2=tf_ipsecsha
With robust evidence that these savings were actually achieved https://www.jmir.org/2019/1/e10929/
In primary care this data has been used to develop new practical informatics and applied data science methods https://pubmed.ncbi.nlm.nih.gov/29986693/
New data science methods to research the dissemination of innovation in the health service https://www.bmj.com/content/367/bmj.l5205
New data science methods to research the dissemination of innovation in the health service https://www.bmj.com/content/367/bmj.l5205
And to explore really important questions like WHY some practices DIDN'T implement new PHE and NHS England guidance on choice of antibiotic for urinary tract infection (one of the most common reasons for seeing a GP) https://www.biorxiv.org/content/10.1101/355289v1.full
ALL OF THIS WORK IS BLOCKED IN HOSPITALS because the NHS will not share the hospital aggregated medicines usage data. There will be HUGE savings for the NHS, needlessly left on the table. Huge unwarranted variation in care, left unaddressed, exposing patients needlessly.
How can we fix this? We set out some clear, workable, practical, deliverable, detailed options in our paper, as always. Because we are here to be helpful.
https://www.bmj.com/content/370/bmj.m2607.full?ijkey=nIHRXr5GpawscdH&keytype=ref
But let me tell you what is not helpful.
https://www.bmj.com/content/370/bmj.m2607.full?ijkey=nIHRXr5GpawscdH&keytype=ref
But let me tell you what is not helpful.
We've been alerted to murmurs, since our paper was published, of people saying "this is wrong... the problem is already fixed". That is not helpful. The problem has not been fixed. This data remains inaccessible. There is only one test: do I have this data in my laptop today? No.
RIGHT that's me done for now.
Please read our paper. It's open to everyone on the link attached. Lots of extra technical goodness in there, and detail on the labyrinth of contracts and peculiarness that has kept this data closed for so long. https://www.bmj.com/content/370/bmj.m2607.full?ijkey=nIHRXr5GpawscdH&keytype=ref
Please read our paper. It's open to everyone on the link attached. Lots of extra technical goodness in there, and detail on the labyrinth of contracts and peculiarness that has kept this data closed for so long. https://www.bmj.com/content/370/bmj.m2607.full?ijkey=nIHRXr5GpawscdH&keytype=ref
As always: lots of GREAT people, fighting great battles for good, behind the scenes, in NHS E, I, X and elsewhere. You know who you are. We know who you are. We sing your praises! But the bad people, the blockers, especially the denialists: we see you too https://www.bmj.com/content/370/bmj.m2607.full?ijkey=nIHRXr5GpawscdH&keytype=ref
See the replies for people who are similarly trying to get this data to do good for the NHS, improving care, spotting savings, reducing patients' risk, with data. https://twitter.com/joefd/status/1290963505886171136
