Sharing medical records with researchers: Assumed consent works in theory – just not yet in practice Welcome to the latest Register Debate in which writers and experts go head to head on technology topics, and you – the reader – choose the winning argument. The format is simple: we propose a motion, the arguments for the motion will run this Monday and Wednesday, and the arguments against on Tuesday and Thursday. During the …
"There may be a point at which it really does make sense for assumed – but also informed – consent to be the default model for sharing medical data"
It's not sufficient (and never will be) to merely be informed that some person or agency is doing something you disapprove of or wish to resist. If there is no choice, there are no rights.
There are typically two alternative data sharing scenarios. Either a proposal is so draconian that a majority will want to opt out, whereupon it shouldn't be implemented, or only a small minority will want to opt out, whereupon loss of that data should not seriously affect statistical results. In either case there's no justification for sidestepping personal choice.
What policy makers tend to forget is that every issue is multi-facetted. Focusing on an apparent advantage of one aspect commonly leads to unexpected adverse consequences. But those consequences rarely affect the policy makers directly, so they are frequently ignored.
>What policy makers tend to forget is that every issue is multi-facetted.
I don't think our dear leaders forget issues are multifaceted, I think they just like to steam roller over any contrary views to get whatever outcome they have want.
> only a small minority will want to opt out, whereupon loss of that data should not seriously affect statistical results
Any opt-out can potentially be removed at a later date.
The Northern Ireland Electronic Care Record (NIECR) system was launched by HSC NI (aka the NHS in NI) in July 2013 and is still in operation today. The patient records systems of all GP Practices (plus Trusts) in Northern Ireland are integrated with it so changes made by GPs to their records are automatically uploaded to NIECR at least once per day.
There was an opt-out available from NIECR's launch in July 2013 up until 13 April 2020 when the HSC unilaterally removed the ability to opt-out. At that point there were less than 100 people (out of a NI population of approx 1.9 million) opted out of NIECR yet HSC claimed opt-out removals were necessary "to support safe and effective care". So 0.0053% of the NI population being opted-out was apparently causing "problems" for HSC NI!
This unilateral opt-out removal occurred despite anyone who had previously opted-out having signed the following statement:
"I understand that if I DO NOT allow my clinical information to be viewed in the NIECR the Health and Social Care staff caring for me may not be aware of important information about my treatment. This may include my current medications, any allergies or bad reactions to medicines that I have suffered from, the results of tests, x-rays or details of treatments or episodes of care that I have had in the past anywhere in Northern Ireland.
I understand that Health and Social Care staff cannot override my decision to not allow my clinical information to be viewed in the NIECR, even in a medical emergency. I understand that I will need to submit another form to request that my clinical information is made viewable through NIECR and that there may be a delay in processing this form of up to 5 working days from receipt."
So despite individuals signing that "staff cannot override my decision to not allow my clinical information to be viewed in the NIECR, even in a medical emergency" the HSC went ahead and overrode their decision anyway. This forms an aspect of the unlawful processing cases I have had open with ICO for 9+ months now...
So people in England were worried about GPDPR's sneaky introduction, here in NI we've had similar (though not identical) systems (2011-2013 ECS,2013-present day NIECR) in place for 10+ years already.
Why is Germany described as a "collectivist country" and Brazil as an "individualistic country" in the opening paragraph?
The "individualism" bar chart here shows UK and Germany as countries scoring highly in individualism, and Singapore and Brazil substantially lower.
"Brazil has a score of 38 which means that in this country people from birth onwards are integrated into strong, cohesive groups (especially represented by the extended family; including uncles, aunts, grandparents and cousins) which continues protecting its members in exchange for loyalty."
> "I would say that Germans (and Singaporeans) have greater trust in their government's decisions than UK or Brazilian folks. With good reason."
East Germans and the retired generation may strongly disagree with that statement. Singaporeans generally don't get much of a choice in whether to trust their government or not.
They seem to have redefined collectivist as the level of people protesting or disobeying public health recommendations. That has a lot to do with the person leading the country and the politicians supporting them, as well as the trust in government. The combination of those factors is really what made the difference in how badly the pandemic was, also factoring in local problems which made adherence to the public health recommendations harder. While calling concern for others "collectivism" sounds alright, it's not what it really means, so the author has ended up with an incorrect phrase.
There may well be smart people in NHS Digital but I am fairly sure none of them are policy makers, or even listened to.
Relying on a huge organisation like the NHS to safeguard anonymised data that is reversible when they can't even develop and operate an acceptable track and trace system regardless off vast sums of money and hordes of consultants tells me I couldn't trust them at all.
Just to be clear, the NHS didn't run the track and trace system. That was an outsourced shit show overseen by that master of data security Dido Harding.
She oversaw the spending of billions of your quids and then simply stuck an NHS flag in the resulting turd and washed her hands of it all.
These scientists wriggle like eels to get big pharma filthy hands on our data.
Gaslighting, defying logic and blinded by its own greed and self-importance.
How people can respect scientific community when they lost touch with what they are for?
The position of authority will only work for so long, once it wears off, the "I am a scientist" is going to mean nothing.
Years ago I opted out of having my medical records shared beyond the local surgery. Recently I find that the hospital, the duty doctor that 111 put me in touch with, all have access to my medical records, they can read my prescriptions, tell me the results of blood tests, etc. So much NHS regard for patient opt-out.
I was wrong to have opted out, for the local sharing of information is very much to my advantage. However, this shows how poor and meaningless privacy is to the NHS and our choice matters not a fig to them.
"our choice matters not a fig to them"
Which is better if you are the government?
A. Create an opt-out system, and enroll the ones who choose to opt-out. Or ...
B. Create an opt-in system, and enroll the ones who did not choose to opt-in.
I say "A" is clearly better for the government, because they can claim the optee missed a step, or missed a deadline, or it was a clerical error, or whatever. Whereas in "B" the only possible reasons are government incompetence or government malfeasance.
It depends what you opted out of. I've opted out and would take action if the NHS shared my data with external researchers.
However the hospital has data access and processing needs relating to the service they provide that mean they don't need my consent, so my opt out is irrelevant to them.
Difficult (for me at least) to fathom how someone who specifically points out the potentially lethal results of data of/about Afghanis collected under the previous regime falling into the hands of the Taliban and still comes the conclusion "Not Yet" rather than "Not Ever" in answer to the question being polled.
You made a sort of half assed fake counter argument there, Joe, a sort of "sure its great but". Oh fook off, transparently false attempt to argue this.
The plan: take "aggregate" data, build outcome models from that data, the models overrule the doctors decision, NHS saves money by not paying for expensive procedures that the model says it doesn't need to pay for, even if the doctor says it does.
The problems:
They'll just end up paying data modellers like Palantir instead.
The doctor knows better than generic model, second guessing doctors prior decisions.
That data gets used for uses it was never intended for. Stuff like: "Why should insurance company cover people living unhealthy lives? [+play one person off against another], they get the data anyway, lets pass identifiers, but just for this particular case and not for other cases....".
Stuff like "wouldn't it be great if the police had a list of everyone who might be a drug user, we have that data anyway, its just another query, and the police can already RIPA query that database anyway, so lets make it official".
Stuff like, "if we cancel this health policy, how many people of [opposition demographic] do we kill?". The analysis the Republicans were running when trying to cancel Obamacare and keep all the Republicans on board. Hey, sure we kill 60,000 but they're mostly Democrats, here, look at our data models.
Stuff that causes the basic purpose of healthcare to be subverted from caring for health, and instead becomes about what gains [group] can get from mining that PRIVATE DATA.
Germany, you've never experienced German healthcare clearly, it's expensive and fragmented, and try to get a cancer specialist and you run around one polyclinic to another to get the bits and pieces of healthcare needed. A clinic here for the CT scan, a clinic there for the cancer specialist, a clinic for the medicine, run run run. Chase chase chase.
Germany life expectancy, 81.57 years, UK 81.52 years, no difference. All that extra money (and you claim data) makes no difference.
If German aggregate data is so useful, WHY NOT USE THAT? They are a similar european demographic complete with a Lidl food diet and similar health care. If *anonymous* aggregate data is so useful, then a similar *anonymous* dataset will be fine. Germany: insanely overpriced, inefficient system, overfunded by taxes. The taxes are so high, 50% of the people cannot afford to move out from their mothers home. Clearly an out of balance economy there.
Singapore, you get that its 50% subsidy for a kids visit and yet still hundreds of dollars, for something as trivial as a pulled elbow? Or you can live with the pulled elbow if you cannot afford it. They have nothing, nobody goes to Singapore for their hospital treatments, they head to Thailand. Singapore: all smoke and mirrors hyper inflated economy, pretends to be hi-tech run by data and robots and is actually run by cheap labour from neighboring Asian countries. Why not just say, you're data driven and run by robots and import Polish labor like before, if you want the smoke and mirrors economy?
You hold up two of the crappiest overpriced systems available as models to be pursued.
UK has a Pharma industry because it has NHS and a stable marketplace. If you strip billions from NHS, you will lose your Pharma industry.
It's private data, you do not get to argue the pros and cons of accessing people's private data, without them being in the argument. THEY decide on a case by case basis whether YOU can use THEIR private data.
Opt in, or undermine healthcare.
"Collectivist countries such as Germany and Singapore are seen as having better served their citizens during the pandemic than more individualistic societies such as the US, Brazil, or the UK."
Really? let's spin that the other way:
Individualistic societies such as New Zealand and Israel are seen as having better served their citizens during the pandemic than more collectivist countries such as Russia and North Korea."
So really, mate, there wasn't even a straw argument for you to knock down. Kudos for knocking it down anyway, though.
What we're weighing up here is the potential benefits of it getting to the right people Vs our confidence it won't get to the wrong people.
The problem is that trust in Government in the UK is at an all time low. With a PM who unashamedly lies to us, goes back on his word with our neighbours as a badge of honour and is already in discussions about rolling back data protection laws for profit, it's seems we're only we ever one casually waved though regulatory change from anything we might have though was satisfactory being swept away.
How can you expect people to to take a leap of faith in that context?
If it's anonymised, you should not have the right to withhold your data. Consent should not be required.
If it ain't anonymised properly, then obviously consent is required - but, just anonymise it properly. and you don't have to worry about people too selfish to consent to something with clear benefits and no obvious (realistic) risks.
I know the local conspiracy fiends will insist someone is going to screw them over if they give out so much as the first character of their postcode, but in practice we're talking about data that will undoubtedly be used to successfully save lives, and which, again in practice, no-one will be allowed to get up to any monkey business with it. That there is the slightest debate about consent here is a sign of the utterly sick selfishness of a certain portion of our society. They are literally putting a purely notional right to privacy ahead of people's lives. Shameful and disgusting.
> "They are literally putting a purely notional right to privacy ahead of people's lives."
Like when we send troops to war to preserve "freedom"? Seems like the debate over lives vs freedom has been had and your side lost miserably.
But seriously, it's not about a notional right to privacy, it's about better health outcomes and I contend that government stealing, sharing and abusing our medical data will lead to people avoiding giving relevant details or seeking out appropriate treatments for fear of negative social outcomes and that will lead to more damage than the sharing of health data ever could cover. It is a net negative for health outcomes long term.
If you wanted to find Stephen Hawking's medical records all you would need is his date of birth and the fact that he had amyotrophic lateral sclerosis (ALS) a form of motor neuron disease.
People with this disease do not normally live as long as Stephen did so there would be very few records with a similar date of birth. If it was known that he had certain complications at certain times, this would narrow down the choice considerably.
Leave out this information and it destroys the usefulness of the record.
You don't need much medical information to identify a person, considering everyone is unique.
And if someone has few or no medical problems they may well not be be in the database.
Anonoymisation is more complex than its given credit for. You get back to the same problem; who decides what's good enough?
As scrubber notes, positioning this as saving lives vs not is a false representation. It's macro research vs local service effectiveness, with an added bonus of providing an avenue for sought-out prejudice.
The practice of assumed consent could work if the data was truely anonymous, as in providing grouped results only and no individual lines of data.
However, that means medical researchers have to know what questions that they want to ask, and someone will have to constantly process those requests. Even then, that also means that in the very act of gathering and centralising that data, there's the likelyhood that someone will leak it or hack into the system.
Thus, I am unconvinced that any approach to gather data into a single place is a good idea, and would personally be far happier if my medical record was something I, personally, kept with me and took to appointments.
The simplest thing is to process the data on your computer, into a set of patterns that are impossible to understand. But these patterns will be quite enough to search for information, provide you with personalized ads, etc.
This how a fragment of model on a document on technology looks like:
datum - be - in : 1794
user - be - in : 1552
profile - be - in : 1441
datum - be - remote : 1335
system - be - remote : 1193
datum - be - plural : 1131
system - be - in : 1110
one - be - least : 1066
Your data will remain yours, and commercial companies will get what they need. For example a pattern “datum - item - plural : 883”.
The wolves are full, and the sheep are whole.
One so-far unmentioned problem is de-duping. You can either anonymise data enough to be really anonymous, in which case you can't de-dupe the database and it's medically unreliable, or you can anonymise just up the point where de-duping is possible - at which point the anonymisation is easily defeated.
Also, the anonymisation can be reversed if there is a legal reason to do it? That way lies disease (and tyranny). We do not want criminals to avoid getting medical attention for communicable diseases, or getting vaccine shots, and they will avoid that if the records of it are used to catch them.
Peter Fairbrother
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
