We can reinvent healthcare and drive up the quality of care across the system once we start digitising patients’ data, argues Matthew Swindells
Information technology seems to be coming back into fashion. First the mandate in December said that by March 2015 everyone will be able to access their GP record and plans will be in place to link electronic health and care records which then follow the individual across NHS or social care. Furthermore, everyone will be able to book a GP appointment online and conduct an e-consultation with their GP.
‘If we really are going down the route of digitising healthcare we also ought to be thinking about second-order effects’
Then, PwC published a report commissioned by the DH saying that £4.4bn in savings per annum could be made by rolling out e-prescribing; better use of information to aid post-operative care; use of acute performance data by commissioners to achieve contractual savings; and giving clinicians access to complete and accurate clinical and attendance information.
The secretary of state and Tim Kelsey, national director for patients and information at the NHS Commissioning Board, threw their weight behind this report and, on top of the mandate commitments, instructed hospitals to have an operational electronic patient record system by 2014 and committed the NHS to being paperless by 2018.
Mr Kelsey’s rationale is that the volume of real-time clinical information that hospitals will need to provide to the commissioning database Care.data can only be achieved through an EPR and that information and digital is “the single biggest answer to the QIPP challenge”.
If we really are going down the route of digitising healthcare we also ought to be thinking about the second-order effects. These are far more transformational than anything Mr Kelsey, Mr Hunt or PwC are highlighting right now.
A promising field
In his fascinating book, Automate This: How algorithms came to rule our world, Christopher Steiner explores how complex decision making that was once the sole purview of people − their knowledge and instincts − has been replaced by algorithms trawling through mountains of data in a split second.
‘Clinical data doesn’t belong to a venue like a hospital or a surgery, it belongs to the patient’
He begins by looking at how in the 1970s innovators hacked into the electronic data feed going to terminals and developed models for trading which outperformed experienced traders. Once electronic trading made data generally available, algorithms replaced traders in all the big trading houses. He doesn’t gloss over the damage that was done to the world economy when the machines became cleverer than the people.
Mr Steiner then looks at more subtle uses for algorithms and big data, such as personality profiling by NASA to choose compatible space shuttle crews, and by online sales companies to match customers and sales people. He also examines voice recognition software that doesn’t decode every word but looks for common combinations of words to make its best guess.
What has any of this to do with healthcare? Jeff Hammerbacher, the man whom Steiner credits with developing the algorithms that sift through Facebook’s vast store of user data and who is now chief scientist at Cloudera − leading big data/cloud search innovators − considers the most promising field for people like him as “medical diagnostics”.
Health systems need to break free from the idea that digitising healthcare data is simply to provide managers with graphs and clinicians with easy access to clinical information. Once you digitise the data, you can reinvent healthcare − first with nudges to doctors and nurses and ultimately replacing whole layers of people and delay.
Clinical data doesn’t belong to a venue like a hospital or a surgery, it belongs to the patient. If we link it in the cloud − as Mr Kelsey proposes − this can’t just be for management, it must be to drive quality of care across the continuum.
It is a challenge for any clinician to draw the right results from data collected over a long period: a lab test a year ago, a scan three months ago and vital signs recorded a minute ago.
‘Refusing to digitise healthcare is like refusing to adopt the car because we love our horse’
If you present all the relevant data at the same time, a doctor will be confident to diagnose, but if they don’t know to look for the crucial lab result, buried under 12 months of paper or off the bottom of the computer screen, they can’t make a diagnosis. An algorithm could.
And when a diagnosis is missed by a clinician who chooses not to use decision support technology and a lawyer shows that an algorithm would have spotted a correlation that no human could be expected to see, is that any better than practising medicine for which you’re not trained or refusing to use modern equipment in theatres?
In the US, where many of the best hospitals are fully digitised, real-time algorithms are already being used to forecast the occurrence of sepsis, prompt which patients are suitable for a clinical trial, incorporate genomic information to advise on medications and predict which admitted patients are most at risk of a future emergency readmission.
Refusing to digitise healthcare because it’s hard, expensive and challenges the status quo is like refusing to adopt the car because we love our horse. The future always arrives in the end.
Matthew Swindells is vice chair at BCS Health, the Chartered Institute for IT, and senior vice president for population health and global strategy at Cerner.