Doctors across the NHS rolled out dexamethasone within hours of clinical trials showing a significant measurable effect on Covid. The difference wasn’t the trial or the doctors – it was finding something materially better for a major pressing need. Most trials don’t.
The problem is not a lack of trials. The problem is all the incentives that influence which trials get done, and who gets to decide. Those incentives are not always aligned with the NHS, and not entirely aligned with patients.
Step changes from clinical trials primarily come from focus and attention. The fifteenth male ‘baldness’ pill with a marginally smaller side effect is an improvement, but it’s not something that makes anyone care. It’s the large step changes that usually get hoped for when “more clinical trials” are set as the measure, and when data infrastructure is being changed to support them. RECOVERY’s dexamethasone result didn’t need to convince anyone to care.
A bigger question is what “more trials” means. Success needs new entrants, and radically different actors, otherwise the status quo fails in exactly the same way that “business models for data” has failed for a decade. The latest incarnation is counting clinical trials (again) – a measure pushed by industry as it’s easy to measure and helps larger members more than the smaller ones who actually do the innovation.
People matter
A measurable step change needs someone with the commitment and ability to keep working on the topic.
Katalin Karikó continued with mRNA research when no one else cared, and succeeded. The people behind the cystic fibrosis drugs had one job, and succeeded. Cure Parkinson’s is still going.
The same pattern holds at scale. Smallpox was eradicated because Bill Foege and Nicole Grasset picked a target and stayed on it; Guinea worm is nearly there, though Jimmy Carter didn’t quite outlive the last one. Polio, malaria and HPV are still being chased by people who have made each their life’s work.
Flu strain B/Y appears to have gone extinct as a side effect of the world’s choice to suppress covid – nobody set out to do it. But accidents can’t be commissioned.
Cure Parkinson’s wants to, well, cure Parkinson’s. A dementia one is starting up.
Maybe they’ll succeed, maybe they won’t, but the single common feature is a primary goal other than profit. Those are the projects that stand the best chance of step change on the diseases – simply because they’ll keep going until they’re done, and not only until they make money off a different grant.
There is survivorship bias. Only a few try (and fewer of those can do it while paying for food and somewhere to sleep). Most obsessives are simply wrong, but well funded true obsessives can be wrong for longer and lobby more successfully than the public interest arguments survive.
Incentives matter
The problem is not a lack of trial ideas. The problem is all the incentives on which trials get done, and those who decide which trials will get done. Those incentives are not always aligned with the NHS and not entirely aligned with patients.
Much of the time, those trials will be of competitors or derivatives of existing drugs (often to simply extend the patent protections).
A clinical trial is a necessary component. It’s not enough, but it is an easy way for Industry to blame the NHS for not making things easy enough, and it is an easy fiction for data brokers to say they can solve, if only given ever more patient data. Some of the legitimate need is less overhead for similar clinical trials. But the race to the bottom on data will not help anyone who expects to continue to care about the disease after the trial is done.
Companies complain that their me-too drugs with marginal improvement over existing treatments take a long time to be rolled out across the NHS. The time taken is generally inversely proportional to the benefit to patients – big benefits are faster than minimal ones. But that explanation doesn’t fit in an investment pitch.
Creating a race to the bottom
That clinical trials overlap with Direct Care makes it especially attractive to get data on everyone that can then be sold on for other purposes.
The NHS can offer around 60 million records with strong safeguards (supposedly). Although it is becoming a race to the bottom since Mr Streeting decided he could do whatever he chooses now that he’s got the recipe data; that ‘flexibility’ will also apply to all future trial data.
The NHS will engage with anyone who pays the data fees and can fill out the forms satisfactorily. Opportunists will go to the NHS if they can, or any data broker that’ll sell them access when they can’t (hello again, UK Biobank and friends).
There are far more ways to make money than there are to make working treatments. Adding yet another data broker will only attract those who are doing more derivative work. It comes down to the Disney ethos – some people make treatments to make money, and some make money to make more treatments.
The system is set up to ensure everyone is treated equally, civil servants don’t want to pick winners and keep their distance from measures of commercial commitment. Even if there was a measure, the losers always complain loudly, and as NHS England found with the creation of DigiTrials, a well-connected loser gets to push ahead anyway – the same cronyism that infested HDRUK.
In DigiTrials, that distance sent an official NHS letter on behalf of a company, one that was read by a recipient as saying “we think you might have dementia or cognitive dysfunction, would you like to join this commercial study?” It was a legitimate and approved project, it just didn’t look like one and there was (and is) no way for someone holding the letter to check. (The company recently “merged” in the way data brokers always do).
Most VCs and investors want returns without particularly caring about the health outcomes – so the system set up for commercial use will be a mess of contradictory promises based on split incentives and fail to deliver; then the leadership will get fired, and promises to patients will get ripped up as they are treated as irrelevant while corporate promises are treated as sacrosanct.
The NHS will treat applicants equally but is not a vehicle for enthusiasm.
Which logo is at the top of the form makes no difference to anyone. The new Health Data Research Service is expected to take over the NHS data access service and both offer data for direct care clinical trials and sell data for research and other secondary uses. Existing policy already expects HDRS to process data for research that patients have opted out of. It’s inevitable that whoever replaces current leadership will fold under pressure to copy the direct care data and sell that – as the “OneLondon” project did when it came under pressure after failing to meet its objectives.
The “Single Patient Record” also cites OneLondon as an inspiration – have all your mental health notes, prescriptions, and diagnoses copied and sold, and their wording on the National Data Opt Out is unclear. OneLondon leadership made decisions that were aligned with their incentives, as leadership always does.
The problem isn’t the researchers – the issue is the decision process assumes that every trial is the cure for cancer, not another male baldness microstep, both of which are indistinguishable from a novel GLP1 expansion or combination therapy.
Where will opportunists go?
DH/E will use the legitimate argument that everyone should be eligible for a trial (if they wish), to argue that either HDRS should go fishing in the Single Patient Record database (effectively the Single Palantir Record), or instead, to copy the entire SPR database in its entirety to HDRS, including notes of patients who have opted out of research uses. Established companies already believe this is going to happen. Both outcomes, to some degree, are consequences of the short-sighted and temporary spin-out of HDRS from the NHS.
Our Future Health was conceived to help the other interests of its Chair, Sir John “no Nobel” Bell. Benefits to the members of OFH/Industry are a side effect, and it’s unclear how OFH can manage the conflict of interest if materially competing companies want access to their cohort (it may be they can’t, and Sir John’s undisclosed interests will have to offer sufficient cash with strings attached). Bell’s investments may use (and fund) OFH, but their competitors will be dissuaded.
Seeing HDRS on the horizon, MHRA’s GP database CPRD has panicked and abandoned its promises to patients and to the GPs they rely on for data and with HDRUK are trying to jerry-rig an NHS-adjacent service without the governance and accountability (but with the fees) – a service that wants to expand to cover many population databases and place all individuals within them at risk.
If you want to enhance NHS data with something else, you’re either looking at OFH, or you’re on your own for now.
Where should good organisations go?
Where should someone like Cure Parkinson’s (etc) go? They’ll go to the NHS for scale.
Much of the NHS “slowness” is down to having applicants who aren’t incentivised to respond quickly or deliver rapidly. Cure Parkinson’s and friends can. The system can respond quickly if applicants avoid the neglect of convenience scheduling, in the same way most NHS scheduling is several weeks out, but two-week cancer pathways and A&E operate much faster. Delays measured on an industry calendar don’t disclose that the ball was in the applicant’s court while the NHS was waiting for them to answer basic questions.
Such a fast turnaround will be gamed by those who want to iterate on a patent-extending variant of their existing treatment – they have the revenues to support such games.
Hence the need for someone who wants to fix a topic with the ability to devote time and cash that isn’t bound to third party deadlines. Considering disease as a policy failure misses the point. Fundamentally what’s needed is someone who actually gives a shit about a material difference.
That’s what RECOVERY had, and a line that’s “more trials” never can. The real success will be from working with an existing organisation committed to success, as the CF team were – and are.
Unless patients can see every way their data is used, the race to the bottom will be secretive and the benefits to patients will remain undelivered – there’s no meaningful pressure to do better.