Simon Wessely statement to Leonid Schneider on data sharing

[Sasha]

Simon Wessely, professor of Psychological Medicine at the King’s College London, has forwarded me this important message on data sharing and patient privacy in UK:

“I sense that people are starting to wake up to the fact that open data/ data sharing is not as straight forward as it first seemed. In this country and elsewhere there is a complex legal framework around data sharing, and one that is also constantly changing in the light of two particular issues. The first is the fiasco of “care.data” [https://www.england.nhs.uk/ourwork/tsd/care-data/ -LS] in this country, and no doubt similar issues elsewhere, which means there has been a loss of confidence in our existing systems of data sharing and linkage, and increasingly people are wanting to have more say in what their personal data is used for. The second, not unrelated, is the increasing power of computer technology both to generate mammoth data sets, but also to permit re-identification of individuals which would hitherto not have been possible. Both of these developments mean that the old divisions between identifiable data (for which consent for sharing must be sought, unless there is specific case by case legal justification not to do so, a complex bureaucratic procedure) and anonymised or pseudo anonymised data, is being tested. There are even calls that consent is needed for sharing truly anonymised data, although that does seem to many, including myself, rather ridiculous. But such voices are definitely being listened to, following the ever increasing importance attached to autonomy, confidentiality, consent and choice by Courts and regulators alike.

What is clear is that the general atmosphere around data sharing, especially of sensitive data such as medical information, is in a state of flux. Indeed several reviews are ongoing in this country which are likely to make data linkage and sharing more onerous, and eventually there will be new legal rulings, which again are unlikely to produce a more liberal approach to data sharing/linkage.

Sooner or later I anticipate there will be a collision between the open data movement and the legal framework around data protection, at least in the UK. As an epidemiologically trained researcher I am anxious about the consequences of this, since I doubt very much it will make data sharing and linkage easier, rather the reverse. I hope I am wrong, but the fall out from care.data has been so toxic that I suspect the direction of travel will be in one direction only. At least in England”.

https://forbetterscience.wordpress....parasite-paranoia/comment-page-2/#comment-350

 

[Sasha]

Please reward Leonid with traffic and comments on his blog.

 

[adreno]

The first is the fiasco of ****

I love it when Simon Wessely talks of fiasco.

 

[Sasha]

Sasha, very minor detail. I think it's Leonid, not Leonard.

Oops! Thanks, Bob - I've corrected it.

 

[TiredSam]

I am anxious about the consequences of this, since I doubt very much it will make data sharing and linkage easier, rather the reverse. I hope I am wrong, but the fall out from care.data has been so toxic that I suspect the direction of travel will be in one direction only. At least in England”.

How reassuring that he shares our concerns, and what a pity his anxiety has obviously caused him to act in the way most likely to bring about exactly the catastrophy he is so concerned about. I recommend CBT to help him with his CFS (Chronic Fence Sitting) Disorder.

 

[sarah darwins]

Funny how these concerns suddenly occur to Sir Simes just when the PACE authors have been ordered to release data. Coincidence, I guess.

 

[Sasha]

Interesting response to that comment by a "D.G.".

 

[Invisible Woman]

To be fair (& I never thought I'd say this) I agree with him that care.data is a disaster. But whereas care.data could allow patients to be recognized from data hoovered up by hospital records and GP records where data may be just pseudo anonymized the trial results requested from PACE is fully anonymized and that makes a big difference.

To my mind the solution is that care.data should be stopped. If not, at the very least, only fully anonymized data should be used. Basically care.data (& the idea of a paperfree NHS) is the UK government's way of generating a new income stream - the sale of our personal information under the flag of improved safety and service. It's too good a revenue service for them to stop.

However, just because pseudo-anonymised data from medical records is for sale this should not stop access to far more secure fully anonymized data being released.

 

[GreyOwl]

I would like to add this comment, but I think it's too cheeky and not helpful, (though I wouldn't stop anyone else from posting it ):

Although Simon's thoughts, feelings and anxieties are very real to him, they are an example of an unhelpful belief system which enables him to avoid engaging with his peer scientific community, which does not share these same beliefs. I would suggest that he could find help from a skilled psychologist who could assist him with his clear distress, and encourage him to challenge his strong belief system. I don't think the scientific community should further engage with Simon on his concerns, as I believe that in doing so they risk validating his unhelpful beliefs, and further entrench any dependence on secondary gains.

 

[ScottTriGuy]

Please reward Leonid with traffic and comments on his blog.

done

 

[Old Bones]

Please reward Leonid with traffic and comments on his blog.

Not only is Leonid's blog excellent, the comments section shows, as usual, how educated our patient population is. I do have one question, however, regarding a post by schweitz75 which describes the CPET test as follows:

"The CPET has long been used in cardiology – and athletics – and is highly regarded as objective. It can’t be gamed. The tester puts a mask over the patients nose and mouth (or over the mouth and clamps the nose shut) so that the air going into and out of the patients’ lungs is measured by a machine while the patient rides a stationary bicycle or walks on a treadmill that gets progressively more difficult. When the patient reaches maximal cardiac effort [a pulse of (200-age) x .8], the machine records the amount of oxygen intake and carbon dioxide release – that is, it measures how much oxygen the patient inputs and how much carbon dioxide has been produced."

I was shocked at how low my "maximal cardiac effort" would be based on this calculation. I'd be finished the test before they even attached the equipment to my body. I then remembered that the HR maximum I've been using in an attempt to "pace" was based on (220-age). Would "200-age" be a typo in the comment, or is this actually the calculation used at Staci Stevens' Workwell facility?

 

[Sasha]

I think it's a typo for 220, @Old Bones, but the comment is incorrect about how they determine maximal effort - the person has to have an O2/CO2 ratio over 1.2 (or something, or the other way around).

 

[SilverbladeTE]

Only things I want to hear that man say are:
"I utterly and unreservedly apologize to the many thousands of M.E. and other patients who have suffered as a result of my actions."
and
"I was wrong, utterly completely wrong."
and
"Guilty, m'lord"

 

[SilverbladeTE]

Funny how these concerns suddenly occur to Sir Simes just when the PACE authors have been ordered to release data. Coincidence, I guess.

yeah, funny that, isn't it?

Always remember folks:
there are NO coincidences when power and money are involved

 

[user9876]

I seems to me this is another shockingly weak response from Wessely as to why PACE data shouldn't be shared. The care data thing is entirely different because it is next to impossible to anonymise detailed medical records (although there could be attempts by transforming the data whilst trying to retain certain statistical or other properties).

The notion that computing power would help reidentify people is also wrong. If you have a many to one function then it is impossible to track back and many of the questionnaires basically have this structure. An all knowing oracle (a concept used in crypto proofs) may well be able to identify particular people in the recovery data as they would already know the data and hence the distances walked. However, I suspect that an oracle that knew the general health of 250000 ME patients in the UK could not associate any reliably with the data due to the many to one relationships in the survey data and fluctuations in the disease which could lead to variation in how much someone could walk in 6 minutes on any given day.

No amount of computation will allow a recreation from the data to pick out 640 patients from the 250000 in the UK.

 

[Esther12]

I seems to me this is another shockingly weak response from Wessely as to why PACE data shouldn't be shared.

Was he even trying to argue against the release of any of the data requests made? I saw a few people describing this as just muddying the water: I wasn't sure that it even reached that.

I sense that he is beginning to realise how dire a situation PACE is in, and how dependent they are on being able to make pronouncements from on high without any real public debate or scrutiny.

 

[GreyOwl]

Was he even trying to argue against the release of any of the data requests made? I saw a few people describing this as just muddying the water: I wasn't sure that it even reached that.

Yes, he is conflating two separate issues, because that strategy has worked so well for him in the past.

 

[Aurator]

"The second [issue] not unrelated, is the increasing power of computer technology both to generate mammoth data sets, but also to permit re-identification of individuals which would hitherto not have been possible....Sooner or later I anticipate there will be a collision between the open data movement and the legal framework around data protection, at least in the UK. As an epidemiologically trained researcher I am anxious about the consequences of this, since I doubt very much it will make data sharing and linkage easier, rather the reverse"

This is truly contemptible in its insincerity. Given how much is clearly riding for him on the PACE data remaining under wraps, who in the world does he imagine is ignorant of where his real apprehensions lie? To feign concern at restrictions potentially being placed on data sharing generally, whilst doing everything in your power to prevent it happening in a specific case you are intimately connected with gives a whole new meaning to shameless duplicity. Such behaviour argues for a pathological absence of scruples.

 

[worldbackwards]

To be fair (& I never thought I'd say this) I agree with him that care.data is a disaster.

Opted out

This really is ludicrous. And given that other people like Richard Smith (hardly a friend of ours) is saying that the direction of travel is entirely in the other direction makes one wonder why someone with such a clear vested interest in not releasing PACE data should be attempting to hijack the debate with such patent bollocks, already rejected by the information commissioner and consistently thrown out as a blind once the whole 'vexatious' thing started getting old.

 

[Invisible Woman]

Opted out .

Don't want to side track things too much here but you cannot opt out of care.data entirely. Opting out only blocks GP records from being uploaded but all records from outpatients, A&E, walk in centres and the like will be gathered and used under care.data. There is no opt out option for that.

Apart from that I agree with you. In fact given care.data means that pseudo-anonymized data from medical records has and will be sold on why would you object to fully anonymized data being released to researchers? care.data just strengthens the argument to release fully anonymized PACE trial data in my view.