Breaches of the DPA are not always about data security. I’m not sure NHS England have grasped this. Worse, I’m not sure the ICO understands public concern about what is happening with confidential medical information. They both need to listen.
Proponents of the care.data initiative have been keen to reassure us of the safeguards in place for any GP records uploaded to the Health and Social Care Information Centre (HSCIC) by saying that similar data from hospitals (Hospital Episode Statistics, or HES) has been uploaded safely for about two decades. Thus, Tim Kelsey, National Director for Patients and Information in the National Health Service, said on twitter recently that there had been
No data breach in SUS*/HES ever
I’ve been tempted to point out that this is a bit like a thief arguing that he’s been stealing from your pockets for twenty years, so why complain when you catch him stealing from your wallet? However, whether Tim’s claim is true or not partly depends on how you define a “breach”, and I suspect he is thinking of some sort of inadvertent serious loss of data, in breach of the seventh (data security) principle of the Data Protection Act 1998 (DPA). Whether there have been any of those is one issue, and, in the absence of transparency of how HES processing has been audited, I don’t know how he is so sure (an FOI request for audit information is currently stalled, while HSCIC consider whether commercial interests are or are likely to prejudiced by disclosure). But data protection is not all about data security, and the DPA can be “breached” in other ways. As I mentioned last week, I have asked the Information Commissioner’s Office to assess the lawfulness of the processing surrounding the apparent disclosure of a huge HES dataset to the Institute and Faculty of Actuaries, whose Society prepared a report based on it (with HSCIC’s logo on it, which rather tends to undermine their blaming the incident on their NHSIC predecessors). My feeling is that this has nothing, or very little, to do with data security – I am sure the systems used were robust and secure – but a lot to do with some of the other DPA principles, primarily, the first (processing must be fair and lawful and have an appropriate Schedule 2 and Schedule 3 condition), and the second “Personal data shall be obtained only for one or more specified and lawful purposes”).
Since the story about the actuarial report, at least three other possible “breaches” have come to light. They are listed in this Register article, but it is the first that has probably caused the most concern. It appears that the entire HES dataset, pseudonymised (not, note, anonymised) of around one terabyte, was uploaded to Google storage, and processed using Big Query. An apparently rather unconcerned statement from HSCIC (maybe they’ll blame their predecessors again, if necessary) said
The NHS Information Centre (NHS IC) signed an agreement to share pseudonymised Hospital Episodes Statistics data with PA Consulting in November 2011…PA Consulting used a product called Google BigQuery to manipulate the datasets provided and the NHS IC was aware of this. The NHS IC had written confirmation from PA Consulting prior to the agreement being signed that no Google staff would be able to access the data; access continued to be restricted to the individuals named in the data sharing agreement
So that’s OK then? Well, not necessarily. Google’s servers (and, remember “cloud” really means “someone else’s computer”) are dotted around the world, although mostly in the US, and when you upload data to the cloud, one of the problems (or benefits) is you don’t have, or don’t tend to think you have, a real say in where it is hosted. By a certain argument, this even makes the cloud provider, in DPA terms, a data controller, because it is partly determining “the manner in which any personal data are, or are to be, processed”. If the hosting is outside the European Economic Area the eight DPA principle comes into play:
Personal data shall not be transferred to a country or territory outside the European Economic Area unless that country or territory ensures an adequate level of protection for the rights and freedoms of data subjects in relation to the processing of personal data
The rather excellent Tim Gough who is producing some incredibly helpful stuff on his site, has a specific page on DPA and the cloud and I commend it to you. Now, it may be that, because Google has conferred on itself “Safe Harbor” status, the eight principle is deemed to have been complied with, but I’m not sure it’s as straightforward because, in any case, Safe Harbor itself is of current questionable status and assurance.
I don’t know if PA Consulting’s upload of HES data to the cloud was in compliance with their and NHSIC’s/HSCIC’s DPA obligations, but, then again, I’m not the regulator of the DPA. So, in addition to last week’s request for assessment, I’ve asked the ICO to assess this processing as well
I don’t yet have any reference number, but please note my previous email for reference. News has now emerged that the entire HES database may have been uploaded to some form of Google cloud storage. Would you also please assess this for compliance with the DPA? I am particularly concerned to know whether it was in compliance with the first, seventh and eighth data protection principle. This piece refers to the alleged upload to Google servers http://t.co/zWF2QprsTN
However, I’m now genuinely concerned by a statement from the ICO, in response to the news that they are to be given compulsory powers of audit of NHS bodies. They say (in the context of the GP data proposed to be uploaded under the care.data initiative)
The concerns around care.data come from this idea that the health service isn’t particularly good at looking after personal information
I’m not sure if they’re alluding to their own concerns, or the public’s, but I think the statement really misunderstands the public’s worries about care.data, and the use of medical data in general. From many, many discussions with people, and from reading more about this subject than is healthy, it seems to me that people have a general worry about, and objection to, their confidential medical information possibly being made available to commercial organisations, for the potential profit of the latter, and this concern stems from the possibility that this processing will lead to them being identified, and adversely affected by that processing. If the ICO doesn’t understand this, then I really think they need to start listening. And, that, of course, also goes for NHS England.
*“SUS” refers to HSCIC’s, and its predecessor, NHSIC’s Secondary Uses Service