NHS patient data and clinical research – coming together at last?
pharmafile | January 28, 2013 | Feature | Manufacturing and Production, Medical Communications, Research and Development, Sales and Marketing | Les Rose, NHS, R&D, clinical, records, transparency
In the mid-1990s, an NHS teaching hospital close to my offices invited proposals for expanding its clinical trials unit. Hitherto it had operated a small Phase I unit with a few beds, and the idea was to develop that into a major early phase facility, and also to diversify into later phase work, with commercial sponsorship high on the agenda.
I had a long meeting with the academics leading the initiative, but they were unclear as to their unique selling point. I explained that one of the main causes of uncertainty in clinical trials was the availability of eligible patients, and that they had the key to unlock that.
They were sitting on a vast repository of patient information, not just in the form of medical records, but as databases of referrals and related data. I knew that they were already coding all this using dictionaries familiar to industry, so that they could very readily produce reliable recruitment estimates for a particular protocol, and even identify patients who could be invited, via their physicians.
They were horrified; medical records were sacrosanct, and not to be used for any other purpose. The meeting came to a quick end. The clinical trials unit was however expanded, albeit very slowly, and currently there are 24 trials in its portfolio. They are still mainly early phase, and most don’t appear to be commercially sponsored, bearing mind the original objective of attracting industry funding. Also, they are almost all in the oncology area, rather than reflecting a diversified and flexible facility.
Needless to say, patient databases don’t feature in the unit’s offerings. Curiously, at about the same time, I spoke at a conference on clinical research outsourcing, and another speaker from a different NHS trust made a presentation about negotiating contracts. It was really a sales pitch, and all about how they had access to so many patients to ensure that recruitment targets were met.
I asked how they could verify their claims, eg., by generating reports from their patient records databases. The speaker seemed dumbfounded at the idea, and said that sponsors would simply have to trust them.
Change at last
It can take a very long time for attitudes to change, but change they have. I’m not a great fan of politicians in general, but the prime minister has taken action that I suspect is much more important than even he realises.
In 2012 he approved wider release of anonymised patient medical records data for research purposes, in the form of the Clinical Practice Research Datalink (CPRD).
Several drivers for this were mentioned at the time, and I guess that high on the list will be stemming the haemorrhage of clinical trials from the UK. Surely the NHS is in a unique position, as a national public healthcare system, in having access to such a vast body of data?
Of course, this is all on a much larger scale than I envisaged all those years ago, but I was only dealing with individual NHS trusts. At the time I did see it as a pilot, which could be replicated widely by a kind of natural selection. But nobody took the bait.
Is this the real life?
I don’t know who writes Cameron’s speeches, but whoever did the one he delivered on 2 August 2012 to a big conference in London, they got something almost right. He declared that: “Drug development relies more and more on real-time data”. He probably meant real-life data, as opposed to the highly selected populations in clinical trials.
A running thread in that speech was the potential to support British industry, which underlines the view widely held by politicians that science is all about building the economy. There is of course rather more to it than that, and we must hope that eventually most politicians will understand such concepts as testing ideas and making evidence-based decisions.
But we must be thankful for this step in the right direction, however it was conceived – this could be a truly momentus change for healthcare research and the NHS.
Patient data concerns
But the plans have raised concerns – they have well and truly put the cat among the data protection pigeons. It was established some years ago that patients own their medical records, which means Cameron wants to change the NHS Constitution, such that patients will have to opt out of release of their anonymised records. I should clarify that my discussions in the mid-90s were not necessarily about patient-level data, but about aggregated statistics on diagnoses, referrals etc.
While I am not a bit concerned about release of my own de-identified data, I can understand those who might be. What if you are the only person with chronic lymphocytic leukaemia in your village? It might be possible to work out who that is, if geographical data were part of the extract, a risk known as re-identification.
For that reason, it is necessary to offer an opt out. A survey has shown that only 1% of patients would opt out, so it’s a cheap safety net that doesn’t impact greatly on data integrity. It does occur to me that certain disease types may be more liable to patients opting out, but that remains to be seen.
Overall, I am not impressed by the hysteria stirred up over data protection. Nothing is without risk, but in this case the benefits for the community are potentially enormous, in relation to very small risks of disclosure for individuals. Yes, I am using a Utilitarian argument – the best benefit for the largest number of people.
Yet I have seen people railing against publicly obtained data being put up for sale, which is rather silly. The data were obtained by our unique public sector healthcare system, this could not be done anywhere else in the same way, so it’s high time we got a better return on our investment.
Sparkling innovation
And there is already evidence of the CPRD sparking off many new initiatives. Four e-health research centres have been announced for London, Manchester, Dundee and Swansea, led by the Medical Research Council. Separately, a translational medicine programme aims to accelerate development of innovative medicines by accessing patient outcomes data on the CPRD.
Not all such ideas are at a national level however – for example in the North of England data from primary and secondary care, and pharmacies, have enabled ‘real world’ studies in respiratory disease, sponsored by GlaxoSmithKline.
Crowdsourced clinical trials?
You may have read Ben Goldacre’s ‘Bad Pharma’ (I wrote a commentary on it shortly after its release). Among his many suggestions is for large-scale head-to-head studies to be be carried out cheaply using existing medical records. This is subtly different from the CPRD, which is historical, but its framework would lend itself to a prospective design.
Ben’s idea is to enrol a patient in the trial, and then largely to forget about the protocol and rely on routine clinic visits to capture data into the medical records system. De-identified data would then be extracted and analysed. I don’t think this could replace the rigour of a randomised controlled trial, but it would certainly add an important dimension in terms of generalisation to clinical practice.
I am quite attracted to the idea. OK, there will be no drug accountability, and we won’t even know whether the patient complied with treatment. There will be missed visits and even more missing data. But all this is what happens in the real world. We will get outcome data, albeit at varying time points – which is surely what we really want.
What if we did such a trial in 1,000 general practices? It would hardly be any more effort than doing it in one. All the data would end up in the CPRD, and available to any subscribers to it. If randomisation were required, we might need an ethics committee opinion, but as this would be using marketed drugs within their labelled indications it would be considered low risk. This is only of course a quick look at the possibilities, but I think they justify in-depth consideration.
The end of the paper-chase
Despite the way the world has changed even in the last 15 years, Good Clinical Practice is still rooted in paper-based thinking. That thinking has to change. The safeguards that we thought were essential to make paper secure and reliable are no longer applicable to digital systems. Computers can do extremely boring tasks repeatedly with perfect reliability, which humans can’t.
We are now looking at vast data repositories, which we know will contain errors and omissions, but that’s real life. We can use software to crawl through the data and tell us what we can rely on. There is wider recognition of this principle, for example in a risk-based approach to trial management, where metrics are used to direct effort where it will be most valuable.
At last we will see practical implementations that will make clinical research more relevant to society.
Les Rose is a freelance clinical scientist, specialising in project management consulting.
Related Content
A community-first future: which pathways will get us there?
In the final Gateway to Local Adoption article of 2025, Visions4Health caught up with Julian …
The Pharma Files: with Dr Ewen Cameron, Chief Executive of West Suffolk NHS Foundation Trust
Pharmafile chats with Dr Ewen Cameron, Chief Executive of West Suffolk NHS Foundation Trust, about …
Is this an Oppenheimer moment for the life sciences industry?
By Sabina Syed, Managing Director at Visions4Health In the history of science, few initiatives demonstrate …
