UK's 'anonymous' health records are wide open -

Ross Anderson, professor of security engineering at the University of Cambridge Computer Laboratory, told MedConfidential's crowd how proposed 'anonymised' data is anything but.

Data on databases and anonymised will be stored without names, per se, but will still retain the patients' post code and age, meaning that it is actually pretty simple to identify patients and those with sensitive records, such as people who are HIV positive.

Anderson pointed out that it doesn't take much more than one family member or friend with access to patient records to get to that data, either. Considering how many people the NHS employs there are a lot of potential weak links. The emergency care system in Scotland allowed access to the health records of prominent political figures like Gordon Brown and Alex Salmond. The culprit was not prosecuted - as it was not 'in the public interest' - however, it could also be an embarrassing metric of the open nature of supposedly anonymised health records.

Although British Prime Minister David Cameron has promised health records will be anonymised, the anonymisation process seems like a cop out as it is still possible to access very private information. There is indeed an opt out, however, it is "like Facebook" - the defaults are wrong, the privacy mechanisms are "obscure" and they get changed whenever a lot of people learn to use them.  Efforts to exempt medical data from European data protection regulation are also underway thanks to the health market lobby.

Anderson concluded that a national system holding 50,000,000 records is too big a target, will be cumbersome, fragile and unsafe, and failures to properly protect privacy will have real costs in safety and access - particularly for the most vulnerable or at risk sections of societies.

Anderson also pointed out that hard data 'thinks' about people in a different way - it doesn't take into account measured human behaviour, and can "look at you in a different, unmoderated way". According to Anderson, for example, it can take just four Facebook 'likes' to determine the sexuality of a user.