NHS Digital Erroneously Reveals Data of 150,000 Patients
5.7.18 securityweek BigBrothers
On Monday July 2, Jackie Doyle-Price, the parliamentary under-secretary of state for health, delivered a written statement to the UK parliament. It explained that 150,000 NHS patients who had specifically opted out of the NHS patient data-sharing regime were in fact not opted out.
"As a result," says the statement, "these objections were not upheld by NHS Digital in its data disseminations between April 2016, when the NHS Digital process for enabling them to be upheld was introduced, and 26 June 18. This means that data for these patients has been used in clinical audit and research that helps drive improvements in outcomes for patients."
NHS Digital is the national information and technology partner to the health and social care system. It has responsibility for standardizing, collecting and publishing data and information from across the health and social care system in England. It is therefore responsible for storing and disseminating NHS patient data to those qualified to receive it.
On the same day, NHS Digital released its own statement. "We apologize unreservedly for this issue, which has been caused by a coding error by a GP system supplier (TPP) and means that some people's data preferences have not been upheld when we have disseminated data. The TPP coding error meant that we did not receive these preferences and so have not been able to apply them to our data."
It seems that a software error in an application named SystmOne, written by software firm TPP and designed to allow patients to opt out of data sharing at their local NHS surgery, failed to record the objections. Those objections were therefore not relayed to NHS Digital. Since the system relies on patients opting out rather than opting in to data sharing, NHS Digital assumed that all patients had agreed.
The software error was detected on 28 June, three years after SystmOne was released, when TPP switched to a new system. Neither Jackie Doyle-Price nor NHS Digital has given figures on how many times this data might have been erroneously shared externally during this period. However, NHS Digital compiles and publishes a register of organizations that receive patient data. The most recent publication (XLS) covers the period from December 2017 to February 18. It shows that patient data was shared more than 5,300 times in these three months.
It also shows where the data shared is considered to be sensitive or non-sensitive, and whether the data was anonymized or is identifiable. The anonymization is performed in accordance with the UK data protection regulator's requirements; but many privacy activists do not believe that anonymization is irreversible.
"As part of our commitment to the secure and safe handling of health data, on 25 May 18 [the date on which GDPR became required] the Government introduced the new national data opt-out. The national data opt-out replaces Type 2 objections. This has simplified the process of registering an objection to data sharing for uses beyond an individual's care. The new arrangements give patients direct control over setting their own preferences for the secondary use of their data and do not require the use of GP systems, and therefore will prevent a repeat of this kind of GP systems failure in the future."
It remains an opt-out of data sharing rather than an opt-in to data sharing -- the latter being generally required by GDPR.
Dr John Parry, Clinical Director at TPP, said: "TPP and NHS Digital have worked together to resolve this problem swiftly. The privacy of patient data is a key priority for TPP, and we continually make improvements to our system to ensure that patients have optimum control over information. In light of this, TPP apologizes unreservedly for its role in this issue."
NHS Digital added, "We are confident that we are now respecting all opt-outs that have been recorded in the system. We will also be contacting organizations with whom we have shared data that may have been affected, and work with them to destroy the data where possible."
In an emailed comment, Mike Smart, a security strategist at Forcepoint, told SecurityWeek, "In this case, it appears the underlying program left patient data exposed, even though each party involved in handling the data was aware of the privacy policy settings. It's a clear indicator that relying too heavily on software will cause these mistakes to happen in the future. We can't afford to leave out the human element when deciding how we protect sensitive data, and must involve creative and lateral thinking in the testing and final checking stage before software goes live."