New national NHS patient “data lake” proposed based on STPs

NHS officials have met with IT suppliers, including Google Deepmind, to discuss a national patient data collection service that will make both pseudonymised and identifiable patient data available for research.

A draft summary of the Interoperability and Population Health Summit in December last year, outlined plans to make NHS regional organisations, likely based on the 44 Sustainability and Transformation Plans (STPs), responsible for gathering and controlling patient data. NHS England chief information officer Will Smart convened the event and wrote the summary.

The document said this data will then be fed into a new national “data lake” that will link data sets to one another and make  available either “pseudonymised of identifiable” for research purposes.

“This regional approach supports subsequent sharing of information with a national data platform for the creation of a national “data-lake” to enable nationally performed analytics.”

Attendees include representatives from IBM, Microsoft, Google Deepmind, several universities, NHS trust senior IT managers, NHS England and NHS Digital.

The summary, written by NHS England chief information officer Will Smart, is the first clear steer that NHS officials will resurrect elements of the care.data scheme, which they were forced to scrap last year amid public and clinical concerns about patient privacy.

Digital Health News reported in August last year about growing concern that central NHS organisations were planning a successor to care.data scheme.

The summary document, which does not mention care.data, said under the proposal service power and control would be devolved to the regional areas covering up to five million people, rather controlled centrally.

Overseas experience showed that collection patient data at this smaller scale would help avoid “major pushback” from public.

These regions would be expected to use an interoperable information exchange to create a real-time “longitudinal record of the patient” based on data collected from multiple sources.

A specific patient’s data could then be compared to regional population data and linked to other data sets for a “wider determinant of health and care for near real-time service planning”.

This information would then feed into national data service, used for service planning, payments, analytics and research.

All regional data collection services would use the same consent model, based on the third Caldicott report, which recommends patients be able to opt-out of how their identified data is used but not whether it is collected.

It is envisaged that patient would play “an active role in generating and using data, and controlling how health and care providers use their data”, which could include developing access to a new patient, the document said.

“This needing to happen in conjunction to protecting privacy and data rights, and improving the experience for individuals to control their data.”

However, the new model is unlikely to satisfy critics who have said existing, less extensive, health data collection programmes are already breaching patient’s privacy by supplying their data to non-NHS organisations without their consent.

MedConfidential coordinator Phil Booth said: "“It’s unclear how this will do anything to help doctors with paperwork, for whom the fax machine is the still the most practical way to pass critical information along a care pathway.”

The proposed data collection model aligns with a growing focus on developing shared cared records across STP footprints, which would help provide the technical architecture for linking patient data.

The document cited the Cheshire Care Record and the Connected Health Cities as examples of how this model might work regionally.