About this site

NHS England publishes the Data Uses Register as a monthly spreadsheet. It is the authoritative source but it is also a 5,613-row, 29 MB workbook with information across three sheets. This site is the same data, rearranged for easier access.

What the register covers

The Data Uses Register records the data sharing agreements, made through the Data Access Request Service (DARS), under which organisations outside NHS England may use data it holds. For each agreement it gives the applicant organisation, the data controllers, the term, the stated purpose and expected benefits, the datasets approved, and individual files released externally by DARS.

It does not cover NHS England's internal uses of data (a separate internal register), the Data Sharing Framework Contracts that sit above individual agreements, or releases made through other routes. Links to those are on the official register page.

If you don't want your data used

Most people in England can choose to stop their confidential patient information being used for research and planning, through the National Data Opt-out. Where the register records whether opt-outs were applied to a file release, this site shows it on the agreement page, and the agreements list can be filtered by it.

Confidential data and opt-outs

For each dataset on an agreement, the register records the basis on which confidential patient information may flow under it, in a column headed "Common Law Duty of Confidentiality". The column holds one of eight fixed values, which name one or more of: support under section 251 of the NHS Act 2006, consent, a statutory exemption, and no confidential data. The Confidential data filter on the agreements list reads them from each agreement's latest version. An agreement is listed under a basis if any of its datasets names it, so an agreement with several bases is listed under each, and it counts as having no confidential data only if every dataset says so. This is what the agreement permits, not a record of what flowed.

The register also says, for every file released, whether patient opt-outs were applied to it. The Patient opt-outs filter counts the files released under every version of an agreement and puts each agreement in one of four groups: opt-outs applied to every file, to some files, to none, or no files recorded. Opt-outs do not apply to every release: data that is not confidential patient information is released without them, for example, so a file released without opt-outs is not by itself a sign that anything was wrong. The register's column is headed "Patient Opt-Outs Applied" and does not say which opt-out it means. Like every count of files on this site, it covers only files released externally by DARS, and "no files recorded" does not mean that no data was shared.

What this site changes

• One page per agreement, not one row per renewal. The register lists every version of an agreement separately — 5,613 rows for 1,950 agreements. Here versions are grouped under the agreement, with the history kept underneath.
• The three sheets are joined. Datasets and file releases are shown on the agreement they belong to, rather than in separate sheets keyed on a reference number.
• Free text is readable. Purpose, processing activities, expected outputs and benefits are laid out as prose instead of a single overflowing cell.
• NHS organisations are followed through reorganisations. Using the NHS Organisation Data Service (ODS), the site recognises when the register renames an NHS organisation or names its successor, such as an integrated care board taking over from a clinical commissioning group in July 2022. Those are shown as renames and successions rather than counted as changes to an agreement, and each integrated care board has one page, listing the sub-ICB locations the register also names it by. Names are shown without their ODS codes, except where a code explains which organisation continued which.
• Organisations are grouped into sectors. The register gives each applicant one of about 30 organisation types, and commercial applicants alone are spread across Commercial, Supplier, Consultancy, SME and others. The Sector filter groups the types into six: NHS, local government, central government and public bodies, universities, charities and not-for-profits, and companies. Some organisations are placed by hand, because their type does not say what they are: the register files universities, charities, public bodies and companies alike as "Research". Each of those organisation pages gives the reason, and the full list is in data/organisation-sectors.json. The register's own type is still shown and can still be filtered on.
• Organisations and datasets get their own pages, so you can start from "what agreements does this university hold" or "which agreements name Hospital Episode Statistics".
• Everything has a stable URL you can link to or cite. An agreement that leaves the register keeps its page, marked as no longer in the register and showing what the register last said, but it is not counted in this site's figures.
• Month-on-month changes are tracked on the what changed page.
• It publishes pages, not data files. For analysis, get the data from the register itself, or from this site's archive of every edition.

Caveats

• This is a mirror, not the source. Transformation bugs are possible. For anything consequential, check the original workbook.
• A few published records are left out because they are not data sharing, such as a test record the register published in error. Each is listed, with the reason, in data/excluded-agreements.json.
• The register describes agreements, not outcomes. "Expected benefits" is what an applicant said they intended, not an audited result.
• An agreement is not the same as an active data flow. Some agreements may simply allow organisations to continue to use data they already hold.
• Many listed agreements have ended. The register keeps an agreement after its term has passed, so 968 of the 1,950 agreements in the September 2026 edition are marked "Expired" here. An agreement is counted as in term if its latest version had not ended by the first day of the edition's month.
• Counts change between editions as NHS England corrects entries and retires expired agreements, so figures here will not always reconcile with earlier published analyses.
• Organisation names are re-capitalised for reading. The register writes most of them in capitals, so this site shows "UNIVERSITY OF OXFORD" as "University of Oxford". It is a display change only, done by rule, so an unusual name or an acronym it doesn't know can come out wrong ("Ims Health"). A page whose name was changed says how the register writes it.
• File releases are summarised by dataset rather than listed file by file. The full file-level rows are in the source workbook. The files-released figures give a view of files released. They don't cover access granted in NHS England's own systems, such as its Secure Data Environment and predecessors. Each agreement page shows the months its files went out, and files released over time counts them across the register.
• Release history and change history cover different periods. File releases are recorded from April 2016. The month-on-month change history starts with the earliest edition held here, July 2021.

How it stays up to date

By hand, and not automatically. This site shows the September 2026 edition, added on 20 September 2026. NHS England publishes a new one each month, but it won't appear here until shortly afterwards — so if a newer edition is out, this site does not have it yet. The official register page always has the current workbook, and the release archive has previous ones.

Licence

The register content is published by NHS England under the Open Government Licence v3.0 and remains under that licence here. NHS organisation codes, names and successions come from the NHS Organisation Data Service (ODS), published by NHS England under the Open Government Licence v3.0; the snapshot used is in data/ods/ in the repository. The site's own code is MIT licensed.

Acknowledgements

This site's code was built with AI assistance (Claude, from Anthropic), directed and reviewed by the site's maintainer.