What this is
A figure from the Care Quality Commission’s statutory monitoring report on the Mental Health Act 1983, laid before Parliament and published on 14 March 2025. The report covers how the Act was used for 2023/24, on a reporting base of 823 monitoring visits covering 870 wards.
The count sits inside the report’s account of deaths notified by providers. It is the year before the 2024/25 count of 14 that has its own entry on this shelf.
What it records
The report records 288 notified deaths for 2023/24: 225 deaths of detained patients and 63 of patients subject to a community treatment order.
Of those 288, 13 were deaths of detained patients “who had experienced control or restraint within 7 days of death”, which the CQC puts at 6 per cent.
Two qualifications sit inside that number. One notification directly described the restraint of a patient before a death from natural causes. Six of the 13 were from unnatural causes.
Four detained patients who had been secluded died.
How to read it
Proximity is not causation. The measure records that control or restraint took place in the seven days before a death. It is not a finding that restraint caused any of the 13, and the report makes no assessment of mechanism.
“Control or restraint” is wider than restraint. Holds, seclusion and mechanical restraint all sit inside one number, and the report does not break the 13 down by type of intervention. Quoting it as “13 restraint deaths” is a different claim, and the document does not support it.
It is a numerator with no denominator. There is no rate per admission, per detention or per restraint episode, so it cannot say whether restraint is getting safer, only that 13 deaths were recorded in that relation in that year.
The 288 also mixes populations, detained patients and community treatment order patients, while the 13 are detained patients only.
And the definition moved. The 2023/24 report counts patients who “experienced control or restraint within 7 days of death”; 2024/25 counts those recorded as dying “within 7 days of restraint”. The 13 and the 14 are close, and they are not like for like. They are not added or averaged here. The defensible sentence is that the count stayed in the low teens across two years on a slightly shifting definition, and any trend claim has to carry that with it.
Where it stops
England, one year, one notification route. The count depends on a provider recording that control or restraint happened in the seven days before a death, and notification practice varies.
It covers detained patients. Deaths of informal patients, of people in the community, and of people restrained in other settings are not in it.
The report does not break the 13 down by setting, age, ethnicity, sector or restraint type in the text summarised here, so none of that should be inferred. The copy available stops inside the summary, so the counts rest on the published table as the vault record read it, and the publisher is the only source. No causal claim is made, and this page makes none.
Sources
If you are asked in a board paper whether restraint deaths are rising, our training on vulnerable and volatile people covers what a service can honestly say from its own records.
Reviewed 20 October 2026.