PACE trial data

If they are saying that the specific data being asked for is too deeply/obtusely buried within the full dataset, and there is no longer anyone available who can extract it from that dataset, can we not simply just request the full dataset, albeit anonymised? Have they any good argument to not provide the complete anonymised dataset? Then no one has to spend time extracting anything from anything, and no inside knowledge needed.

Also, I think I read in the PACE protocol that all patient-identifying data would be held in a completely separate additional database. So other than that one database, isn't the rest of the data anonymised anyway? If we asked for the full anonymised database, would any effort or expertise then be needed? No anonymisation needed. No data subset to extract.
 
"QMUL has explained that the Chief Investigator of the PACE trial retired from QMUL on 31 December 2016. While QMUL remains the holder and owner of the raw data from this clinical trial, it has effectively lost the means to locate and extract it because this requires specialist knowledge. There is no longer anyone at QMUL with the ability to produce data from this trial. QMUL no longer employs anyone involved with the PACE trial."

What does QMUL " has effectively lost the means to locate and extract it ( the data)"mean? Lost the means? You can lose data if you are incompetent, but what about losing the means to locate? What does that refer to? Where is the data located? But they remain the " holder"- how does that fit in if they have lost the means to locate it?

EDIT: written while the previous posts were being written and posted.
EDIT2: OK- maybe they can locate the whole data set but not specific items within it which needs specialist knowledge. Is this it?
 
While QMUL remains the holder and owner of the raw data from this clinical trial, it has effectively lost the means to locate and extract it
Click to expand...
@Binkie4 it's not clear whether they can't work with the data, due to lack of a specialist, but know where it is, or whether they've actually in effect lost it. It appears to be both. The data set is presumably there, but no one is left who knows how to find it.

One would think though, that one of the other investigators, from another institution, might have some clue as to how it was labelled/stored. Unless all the statisticians were at QMUL. Maybe we need to ask one of them for the filenames :rolleyes:
 
I think this is a serious admission by QMUL it suggests they do not have an adaquate data management policy to support clinical trials. The spent 5m on this trial and they claim the data is so badly stored and documented that it can no longer be used. That should be considered unacceptable. The MRC should be seriously concerned.

What is more is it suggests that the data is not being professionally managed, I'm a bit shocked that the ICO did not pick up on this as an issue because the full data set does contain personal data. If the data is not well managed then that suggests that QMUL is not taking adequate care over PII and the ICO's office has glossed over this issue.

In truth though I don't think the data is complex and unless they have failed to document their schema a vaguely competent data scientist could get the relevant fields out in an hour or so. If they have failed to document the schema then they should withdraw all the results from the trial because the process behind producing them could not be of sufficient quality to be reliable.
 
Invisible Woman said:
2) This was a publicly funded trial so surely they must be accountable to some public body - 5 million quid and now the data can't be accessed?
Click to expand...

Spend 5 million on a study then someone leaves so therefore the data cannot be accessed any more. And this at a time when it is under scrutiny. I would say it's another thing for Carol Monaghan's notes to bring up with the MPs!
 
Has anyone asked for the entire anonymised dataset? I would have thought that the next logical step in response to this refusal.

I don't believe they don't have anyone who can analyse the data, it's fairly straightforward, excel does most of it for you and universities teach data analysis as a module in every science based degree. QMUL will have an intranet network and someone in IT will have access to the databases. And maybe the lead researchers have retired but are we really expected to believe that no researchers or students who worked on the trial still work there now?

Actually, this is the most bullshitty of bullshit excuses I have ever heard from a university, how did the ICO swallow such nonsense?

Edited to make the 2nd paragraph almost readable!
 
Last edited:
In paragraphs 12-13 aren't QMUL are relying on an ICO judgment that came before the tribunal verdict? Surely that tribunal verdict supersedes the ICO judgement, and showed that they were able to access the data. I wonder if any of the QMUL staff involved in releasing that data are still employed by QMUL?

edit: Yes, it was from March 2015. As Adrian goes on to say, when we then got the PACE data it took him less than two hours to dig into it and conduct the analysis QMUL claimed would cost most than £450 to do: https://ico.org.uk/media/action-weve-taken/decision-notices/2015/1043578/fs_50557646.pdf

re PACE being all over, eg: "the funding for which would have finished several years ago."

We're still waiting for papers they said would be released, eg the paper on predictors, the LTFU paper with employment outcomes... what's happening with all that?
 
Last edited:
Adrian said:
I think this is a serious admission by QMUL it suggests they do not have an adaquate data management policy to support clinical trials. The spent 5m on this trial and they claim the data is so badly stored and documented that it can no longer be used. That should be considered unacceptable. The MRC should be seriously concerned.
Click to expand...
My thinking also. In fact it suggests that the protocol (or other document?) for any trial should include provision for how the data and access to it will be maintained, long after the trial has finished. I suppose it's too long a shot to hope they may have made any such undertaking in their original documentation?
 
Binkie4 said:
."

What does QMUL " has effectively lost the means to locate and extract it ( the data)"mean? Lost the means? You can lose data if you are incompetent, but what about losing the means to locate? What does that refer to? Where is the data located? But they remain the " holder"- how does that fit in if they have lost the means to locate it?
Click to expand...

Maybe it was stored in underground bunkers that had walkways to the DWP and Whitehall and now that Peter White has retired no one else there has the clearance to walk down the channels to be the sitting Prime ministers bitch.
 
Binkie4 said:
  • " It said that SPSS and Stata software is required "(point 12)

  • Is QMUL saying that they don't have this software in the University?
Click to expand...

They have a tactical clinical trials data storage unit that cover stata and SPSS.
http://www.its.qmul.ac.uk/services/catalogue/items/142820.html

They also have licences for SPSS
http://www.its.qmul.ac.uk/services/catalogue/items/143153.html

The data was (according to the stats plan also stored on an Access database).

When the FINE data was shared on PLOS ONE in stata formats it was easy to convert with simple (python) instructions easily findable on stack exchange. So there support organisation could do that!
 
With another FoI request by @Anna Wood QMUL claimed they would need to hire a statistician to perform the task and again the ICO believed them. Once data was released it took me less than 2 hours to work out the information and that included interpreting their recovery criteria and checking the way they had modified the Oxford criteria with additional thresholds (also making dinner and with the football on TV).

So I don't think their pronouncements are trustable but the ICO doesn't seem to ask questions.
 
Barry said:
My thinking also. In fact it suggests that the protocol (or other document?) for any trial should include provision for how the data and access to it will be maintained, long after the trial has finished. I suppose it's too long a shot to hope they may have made any such undertaking in their original documentation?
Click to expand...

I think that is a requirement from the MRC now. Not sure if it was at that point. Also data sharing policies need to be set early on,
 
Valentijn said:
They might lack the technical knowledge to question what they're told. It's a big problem in judicial systems as well. Basically they have to trust that they're being told the truth, or bring in a expert, which doesn't always work out well either.
Click to expand...

Yes I was reading one book that talked about companies preferring arbitration than going to court because the arbitrators tend to be experts.

But the ICO should be experts information so they should have more of a clue.
 
You must log in or register to reply here.
Back
Top Bottom