ADI by the Autonomy Institute

Data

Download the data

The complete Annual Deprivation Index, as CSV. This is the same data the site is built from, with the same corrections applied — not the raw pipeline output.

Covering 2014 to 2025. Dataset release 2026-09-04; files generated 2026-09-03.

Each archive holds one CSV per domain (employment, crime, health), long by year, plus a README describing the columns and the known gaps.

Before you use it

  • Use each rate's own population. pop is the all-age ONS mid-year estimate (Nomis NM_2014_1, 2021 LSOA vintage) for that row. Every metric's _rate is its count divided by the adjacent metric-specific _pop column. That denominator usually represents covered residents of all ages; the new *_qof_afflicted_rate columns instead use the condition's covered eligible-age population. Do not substitute the row's pop.
  • Two health rates answer different questions. The existing *_afflicted_rate remains the estimated share of all residents and is unchanged. For nine QOF conditions, *_qof_afflicted_rate is an additional, QOF-comparable share of residents in the eligible age range. It is age-restricted, not age-standardised: it does not adjust places to a common age structure. All nine new triples are blank in 2014; seven begin in 2015, while asthma and non-diabetic hyperglycaemia begin in 2021. Never add or average the two representations.
  • Health years are offset by one. QOF runs April–March and is labelled by the year it ends, so health 2021 covers April 2020–March 2021. Employment and crime are calendar years.
  • Gaps are empty cells, never zeros. Crime coverage varies by year and police force: incomplete, materially unlocated or malformed force returns are left blank. A blank means usable data were not collected for that area-year, not a zero crime rate. To identify affected areas in a given year, filter the LAD or LSOA crime CSV for blank crime *_rate fields. National and regional rates use only areas with data.
  • Health figures for 2021 are not comparable with earlier years. QOF implementation changed during the pandemic, and NHS Digital warns that indicator data may be inaccurate. In the current England series, 10 of 20 comparable condition rates rose and 10 fell; obesity fell sharply while depression rose. Do not use 2021 as a like-for-like trend point.
  • Smoking and hypothyroidism are not included. They were one-year QOF groups and are omitted rather than shown as zero. Historical CVD primary prevention is included for 2014–2020 and blank after the register was withdrawn.

The full detail is on the methodology page.

Licence

Open Government Licence v3.0. Boundaries © ONS / Crown copyright. If you use the ADI, please cite the paper — “An annual deprivation index for neighbourhoods in England”, Kikuchi, Calvert Jump, Michell and Stronge (2024).

The paper (PDF) ↓ Pipeline on GitHub ↗