How we produce the construction injury statistics
Which HSE tables the construction injury statistics come from, how each headline figure is chosen, and how we check them before they go on the page.
Where do the figures come from?
From the Health and Safety Executive’s published statistics tables for Great Britain. We read HSE’s files on 3 October 2026 and use them as published: every figure is a cell from an HSE spreadsheet or, for kinds of fatal accident, a count from HSE’s own list of reported deaths.
- RIDHIST Table 1 published 20 November 2025
- RIDFATAL Table 1 published 1 July 2026
- Fatalities reportable under RIDDOR (record list), aggregated published 1 July 2026
- LFSINJIND Table-1 published 20 November 2025
- RIDIND Table 3 published 20 November 2025
- LFSILLIND Table-1 published 20 November 2025
- LFSILLIND Table-2 published 20 November 2025
The full dataset, with every row’s file, sheet and row number, is on GitHub and archived on Zenodo as doi:10.5281/zenodo.23126528.
How is each headline figure chosen?
- Worker deaths in 2025/26p
- Fatal injuries to construction workers (SIC section F), employees and self-employed together, from RIDFATAL Table 1, with HSE’s rate per 100,000 workers.
- Worker deaths over 2021/22–2025/26p
- HSE’s five-year total from the same table, and its five-year average rate of 1.78 per 100,000 workers.
- Falls from height
- Deaths HSE classes as “falls from a height” among construction workers over 2021/22–2025/26p, counted from HSE’s list of reported deaths. The share of all deaths is ours, calculated from HSE’s counts.
- Self-employed and employees
- Rates from RIDFATAL Table 1 for the latest year HSE has finalised (2024/25r), each divided by its own group of workers.
- Self-reported injury rate
- Labour Force Survey estimates from LFSINJIND Table-1: injuries per 100,000 workers, three-year averages (2022/23–2024/25 is the latest), with HSE’s 95% confidence intervals.
- Work-related ill health
- Labour Force Survey estimates of construction workers with an illness caused or made worse by work, from LFSILLIND Table-1, annual averages over 2022/23–2024/25.
- Deaths since 1981
- RIDHIST Table 1 to 2020/21, then RIDFATAL Table 1 from 2021/22, which carries HSE’s latest figures. The two tables agree where they overlap.
How are the figures checked?
- Each update is rebuilt from HSE’s files, not edited by hand.
- Every charted figure is checked against the HSE row it came from, and the kinds of fatal accident must add up to HSE’s five-year total.
- The headline figures are checked against the statements in HSE’s own summary reports for the same release.
- A figure on the page can change only when HSE’s file for it changes. Each update lists what moved and why.
What the figures do not cover
Great Britain only, not Northern Ireland. Fatal figures count deaths from injuries reported under RIDDOR, not deaths from work-related disease such as asbestos cancers. Employers report only around half of reportable non-fatal injuries, and HSE publishes no non-fatal rate for the self-employed. The long series has breaks where reporting rules and industry codes changed. Reading these figures has the detail.
Changes and corrections
- Moved into the construction data library. Charts can be embedded, and the full dataset is archived with a DOI.
When we correct a figure, we list it here with the date and what changed. Spotted one that looks wrong? Tell us which one and we will check it against the source.