Discover more from Rory’s Always On Newsletter
Data - tapping the NHS reserves
The Christie shows the way
Rory’s Always On Newsletter is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
“Data is the new oil” is a tiresome cliche but there is some truth to it. If data leaks you can end up with a mess as costly as any oil spill and, just as putting diesel in a petrol car won’t get you very far, trying to extract value from incompatible data sets is a fool’s game.
The NHS is in theory the Saudi Arabia of health data with vast reserves which could make it very wealthy indeed. But, as I have written before, too often that data is trapped within one institution with fears about privacy or technical issues preventing it from being shared for the common good.
But the data team at one pioneering hospital has been showing what can be achieved with a little boldness and imagination. The Christie in Manchester is a hospital that specialises in one thing, cancer - not only treating it with the most modern approaches such as Proton Beam Therapy but running a major research operation employing hundreds of scientists looking into new treatments.
In theory, this should be an excellent model for a datacentric organisation - the hospital collects vast amounts of data on its patients, their treatment and outcomes which can then be used in its research wing.
But when I spoke to data advisor Phil Bottomley and Chief Information Officer Alistair Reid-Pearson they made clear that had not been happening: “We're very good at capturing data, but we're not very good at using it for research,” said Phil. The two, both relatively new to the Christie and in Reid-Pearson’s case to healthcare, decided to find out was going wrong and come up with a new strategy.
They found that over time the hospital had acquired a plethora of IT systems that individually might have been “best of breed” but didn’t actually speak to each other. That made sharing data, even within the hospital, pretty hard.
Working with a big data organisation, Better, they adopted a policy known as OpenEHR, a specification that makes it easier to exchange data between systems and put it in an electronic health record.
What they decided not to do was buy one of the giant American “all singing all dancing” IT systems which are good at helping larger hospitals go digital in a hurry, not so good at allowing the extraction of data. Alistair explains that a key principle of the new NHS “Data Saves Lives” strategy, built on the realisation of the crucial role played by data in the Covid crisis, is that you separate your data from your applications: “You have full control over your data, but that is separate from your application layer. So one of the things we're acutely aware of is if we buy one of these big Leviathan systems, we wouldn't be doing that.”
The first big success of the new strategy at the Christie builds on the fact that many of its patients are now collecting data remotely on their smartphones. They download an app and when they are at home after chemotherapy fill in questionnaires about their symptoms and how they are feeling. “But then getting that back into the hospital system is really difficult,”says Phil Kendall. “So that loop between patient and hospital is not joined up at all.” Given that taking care out of the hospital and into the home is a key NHS objective, it seems extraordinary that getting the data from the patient to the doctor is so hard.
But the Christie has an answer - ePROMS, electronic Patient Reported Outcome Measures. It uses Better’s platform to join up the information tapped into their smartphones by patients and the data that doctors have already collected.
“The transformational bit is how we’re structuring the data when it's captured so we can use the data instantly for research, “says Alistair. “That's the openEHR part of it.” One result may be that there will need to be fewer phone calls to patients to collect information. The data will also be more readily available for machine learning applications which could detect patterns in symptoms and perhaps give early warnings of problems.
But the most impressive thing I heard in my conversation with Alistair and Phil was a story which said much about the potential of health data and the challenges involved in using it. One of their colleagues, a brilliant young data scientist called Lauren Scanlon, had taken a batch of blood test results and come up with an algorithm which could detect AKI (acute kidney injury) 30 days in advance with 95% accuracy. As AKI is a serious threat to people undergoing chemotherapy this sounded like an amazing advance and I wanted to know when it would be put into practice.
There was a lot of shaking of heads. Alistair explained:
“She's done it in the pilot, but putting it into live operation to compare it to high volumes of data to a point where it could be clinically used as a judgement call is what we can't do at the moment.”
And why not, I asked.
“Because the data is not structured in a way where we can do that accurately,” he replied.
Clever scientists like Lauren Scanlon are beginning to tap the NHS data reserves but using this valuable resource to transform healthcare is going to be a lengthy and complex business.