Dianthus Medical Blog Archive

UK Biobank part 4

For anyone who hasn't yet read the first 3 parts of this story, they are here, here, and here. Just to recap, my main concern is that the NHS disclosed my confidential data (and the data of about 5 million other people) to the UK Biobank team without my consent.

OK, now we're all up to date, I have just finished reading the documents I received in response to my FOI request from both the Department of Health and the ethics committee that approved the UK Biobank study.

I am not reassured.

Before the main UK Biobank study was approved, the UK Biobank team did a pilot study to investigate the feasibility of their study procedures. You can read the pilot study report here (546 KB pdf). That pilot study was submitted as part of the ethics application for the main UK Biobank project.

To give at least some credit to the UK Biobank team, they did identify in their ethics application that obtaining data from the NHS was one of the main ethical issues in the study.

Oh yes.

However, they then go on to say "Experience in the integrated pilot study showed that this approach was well received by potential participants". Hm. We'll come back to that.

As well as applying for ethics approval, the UK Biobank team also applied to the (now defunct) Patient Information Advisory Group (PIAG) for permission to obtain identifiable patient data without consent, as required by the Health and Social Care Act 2001 (now superseded, but in force at the time). Permission was sought (and granted) to obtain the data from central NHS registers. The justification given for this was that in the pilot study, in which they requested the data from individual Primary Care Trusts (PCTs), the response from PCTs was variable.

The pilot study contacted 4 PCTs to ask for the data (I believe the necessary approvals to do this were in place at the time). One of the PCTs sent the data rapidly, one sent it after several weeks' delay, and the other 2 did not send it at all. We are not told why not, but perhaps it is because the data controllers at those PCTs believed it was unethical to release confidential patient details without the consent of the patients.

If I had been on the ethics committee, I would have asked questions about that. The difficulties in persuading data controllers to release patient data should suggest to anyone who is awake when reviewing the ethics that there could be ethical problems, but that aspect was ignored, and the reluctance of PCT data controllers to release confidential data was apparently seen as no more than an administrative difficulty. The ethics committee's minutes do not contain anything to suggest that anyone asked about the reasons for the PCTs' declining to disclose patient data.

Now, both the application to the ethics committee and to PIAG stated that the recruitment method was "well received" in the pilot study. Let's see if that claim stacks up. In the pilot study, they sent nearly 60,000 invitation letters. Of those who declined to participate in the study, 57 gave the reason for declining as being concerned about their contact details being provided by the NHS without consent or about it being an invasion of privacy.  In the PIAG application, UK Biobank state that because only 57 of 60,000 invitees declined for this reason, only 0.1% of participants were concerned.

On the face of it, that sounds reasonable. Surely it's OK to go ahead with this recruitment method if only 0.1% of invitees object? But hang on a minute: only about 7,000 of the original 60,000 gave a reason for declining. So using 60,000 as the denominator is totally inappropriate. It would be more accurate to say that about 1% of invitees were concerned about the breach of privacy. Isn't it a bit worrying that the people in charge of running a major, multi-million pound epidemiology project could make such a schoolboy error in epidemiology?

But even if 1% objected, that's still not very many, is it?

Possibly not. But it's crucially important that only about 25% of invitees responded to the invitation letter at all. With a 25% response rate, the potential for response bias is huge. It is reasonable to assume that those who didn't respond to the letter at all were less enthusiastic about the project than those who did respond. So the proportion who objected to their privacy being breached among the non-responders could have been far higher than 1%.

But that's not all. The form that invitees were given to respond asked them to indicate the main reason(s) for not wanting to participate, and provided a fairly small box for the response in which it would be hard to list more than one reason. If some participants were concerned about the privacy aspects of the invitation but also had other reasons for not participating, they may well not have reported their privacy concerns, even if they did go to the trouble of returning the form.

It's important to note that participants were not specifically asked about their attitude to having their details released by the NHS.

Given the low overall response rate and the way reasons for non-participation were collected, I think the fact that as many as 57 invitees specifically volunteered the information that they were concerned about privacy should alert anyone who is awake that there could be a concern here.

But as we've already established, the ethics committee showed no signs of being awake.

Of course, obtaining data from the NHS central registers without patients' consent is a big deal, and it needed approval from government ministers to make it happen. My FOI requests have turned up some interesting documents.

The earliest is the minutes of a ministerial meeting held in 2005 at which the UK Biobank study was discussed. The minutes contain the sentence "It was agreed the first contact should come from the DH or the NHS, not UK Biobank". However, the invitation letter for the study very clearly comes from UK Biobank (although it is counter-signed by a representative of the Department of Health), so it seems that that agreement was subsequently overturned. How, I do not know.

There is also a letter, written by representatives of UK Biobank and its main funders in May 2007, to Lord Hunt, a health minister at the time, asking him to intervene to make it possible to obtain the details from the NHS centrally, and repeats the line that the invitation was "well received by the public".

The response from Lord Hunt makes interesting reading. The tone of the letter suggests that he was perfectly aware there was something dodgy about the NHS disclosing patient data, but was going to go ahead anyway. He states "we must all of us be ready to explain clearly and consistently why it is right for the Department of Health to contact NHS patients for research". He also says that he has asked the Department of Health's communications team to work with the UK Biobank and its funders "to get the right messages across".

You have to wonder why it would be important to have your communications team "get the right messages across" if you were confident that what you were doing was ethically defensible.

← Internships available Plagiarism hurts →