Feature

Care.data: why are Scotland and Wales doing it differently?

BMJ 2014; 348 doi: https://doi.org/10.1136/bmj.g1702 (Published 20 February 2014) Cite this as: BMJ 2014;348:g1702
  1. Margaret McCartney, general practitioner, Glasgow
  1. margaret{at}margaretmccartney.com

NHS England has delayed the launch of its national system to link general practice and hospital records after concerns about patients’ privacy. Care.data should look to the devolved nations, says Margaret McCartney

The system that offered researchers linked primary and secondary care records for the first time is on pause. The hesitation comes after polls suggested that less than half the population had heard of the scheme1 and concerns about the security of the data, the process for gaining consent,2 and data being available to commercial organisations.

After NHS England sent every household a leaflet that didn’t explicitly mention care.data, the Information Commissioner’s Office said, “We don’t feel that the opt out has been explained as clearly as we expected.”3

The Health and Social Care Act 2012 gives the Health and Social Care Information Centre, an executive non-departmental public body, the power to require all general practices to upload their data. Citizens can opt out by asking their general practitioner—but only if they know they are otherwise opted in.

A rich and accurate dataset of referral patterns, diagnosis, treatment episodes, drug allergies, and trends in admissions has advantages for research and planning health services. But negative publicity may mean many people opting out and less accurate and useful data.

The scheme will allow approved researchers to obtain raw, identifiable, “red data,” including diagnosis codes, prescription details, and demographic information, without the consent of individual patients. Such data can be released only in a public health emergency or with the secretary of state for health’s approval after scrutiny by the independent Confidentiality Advisory Group. So far, most of the applications have been from NHS researchers, but applications have also been made by commissioning organisations to facilitate payments4; more applications are expected once care.data is rolled out.

Aggregated “green data” are fully anonymised and available freely. Pseudonymised “amber data” have the main patient identifiers changed and access has to be approved by the independent Data Access Advisory Group. Requests have so far been mainly from academics but also from the Institute of Fiscal Studies and the private commercial healthcare data companies Dr Foster and Civil Eyes Research.5 6

Different approaches

There has been much less controversy in Scotland and Wales. The Welsh system does not hold any fully identifiable data, and researchers need consent from individual patients to access red data. And, unlike in England, there is no law forcing practices to participate; each practice chooses, and almost half have voluntarily signed up.

In Wales data are split at source. Demographic data are “pseudonymised” by the National Wales Information Service (NWIS), with fields such as postcodes removed. The Secure Anonymised Information Linkage project (SAIL, www.saildatabank.com) combines them with clinical data without demographic identifiers so that neither SAIL or NWIS can independently decrypt patients’ identity.

The Scottish Primary Care Information Resource (SPIRE), due to launch at the end of 2014, stipulates pseudonymisation at source before data can be transmitted to NHS National Services Scotland.7 SPIRE will be able to use “reversible” pseudonymisation, which can identify the patient and link—for example, GP and hospital records. The resulting information will then have “one way” pseudonymisation performed before it is used for research. Unlike in England, general practices will be able to opt out—entirely, from specific uses, or case by case. Practices can review each request before data are released to SPIRE, with no response taken to mean no. When asked about identifiable, red data, a spokesman for the Scottish government said, “This will only be extracted with the approval of the SPIRE steering group and the explicit consent of each patient.”

In Northern Ireland, the new electronic care record (NIECR) makes clear in an information leaflet to patients that “only the staff caring for you will be able to look at your information on the NIECR and you will be asked for your permission first. An audit trail will check this.”8

Currently, researchers throughout the UK, including marketing and drug companies, can access aggregated, green data for free. For red data, individual consent must be obtained from Scottish and Welsh patients—but not those in England.

So what of amber data? The chief data officer of the NHS, Geraint Lewis, has said that “a determined analyst could attempt to re-identify individuals within amber data by linking them to other data sets.”9 This risk is to be mitigated by access being governed by the Data Access Advisory Group and a contract which states that it is illegal to attempt to identify individuals.

SPIRE recognises that even pseudonymised individual records may contain sufficient detail to risk disclosure of patients’ identities, so limits access to users and purposes approved by a steering group.

So why has NHS England done things differently? A spokesperson for NHS England told me that because “organisations across these different care settings use different information sources, pseudonymisation at source is not currently possible.”

But Scotland plans to do it, and Wales already does, so why is it not possible in England? And why was a law passed in England that allows for fully identifiable data to be made available to researchers without individual consent, when this will not be allowed in Scotland or Wales?

In 2011, when announcing the intention to change the NHS Constitution to enable care.data, Prime Minister David Cameron said, “The endgame is for the NHS to be working hand-in-glove with industry as the fastest adopter of new ideas in the world.” This would act as a “huge magnet to pull new innovations through, right along the food-chain—from the labs, to the boardrooms, to the hospital bed.”10 If this is the true intention of the English version of big data then it should be in the leaflet.

Notes

Cite this as: BMJ 2014;348:g1702

Footnotes

  • Competing interests: I have read and understood the BMJ Group policy on declaration of interests and have no relevant interests to declare.

  • Provenance and peer review: Not commissioned; externally peer reviewed.

References

View Abstract