What this is
The fourth report in the Health Services Safety Investigations Body’s series on mental health inpatient settings, published on 30 January 2025. The Secretary of State for Health and Social Care commissioned the series in June 2023.
This investigation examined how providers investigate the deaths of patients receiving care in inpatient units, or who die within 30 days of discharge. It covered patients detained under the Mental Health Act 1983 and informal patients. It asked three questions: whether investigations into these deaths were timely and effective, what local, regional and national oversight and accountability existed around them, and what mechanisms captured data on deaths and near misses across providers. The team visited over 40 care areas across 30 mental health care providers.
What it records
The investigation found significant challenges in maintaining safety, in conducting effective investigations, in managing data on deaths, and in ensuring learning across a system.
It found inconsistencies in data reporting, a lack of consistent terminology, and difficulty in comparing one provider with another.
It found gaps in discharge planning, in access to crisis services and in access to community therapy that were potentially contributing to poor outcomes, including deaths.
It found a culture of blame in which individuals and organisations were afraid of safety investigation processes, which slowed the investigations that did happen, so that the people who knew most about a death were the least willing to describe it.
It makes safety recommendations to the Secretary of State and to system bodies. There was no court outcome, and no finding of liability.
How to read it
This is not a report about restraint, and it must not be cited as one. Its subject is how the system investigates and counts deaths from any cause in inpatient care and in the month after discharge. Read it for restraint and you will find nothing to quote.
What it does is explain a gap that restraint training runs into constantly. If two trusts record the same death differently, and no shared terminology exists, nobody can say how many patients died, still less how many died after restraint, or whether one service is worse than another. The oversight intended to catch these deaths, at board level, through incident reporting and through regional review, was not producing information that could be aggregated or acted on.
So the honest answer to “how many people die after restraint in England each year” is not a number, and this report is the reason. Teach the gap, not a figure.
Where it stops
England, published 30 January 2025, on practice before that date. It covers inpatient units and the 30 days after discharge, all causes of death, and it does not quantify restraint deaths at all. It is a system-level investigation: its findings are about frameworks, not about what happened to any individual patient.
The report makes recommendations rather than findings, and nothing in it decides a case or apportions blame, which is part of the problem it describes.
The copy available here is the report’s opening sections and executive summary, not its evidence chapters, and the vault holds the record at `probable` because only one independent source record exists for it. A second, independent report of the same findings would lift it.
Sources
When someone asks your service for a restraint-death figure and there is none to give, our restraint reduction training covers what to record so that your own numbers can be compared.
Reviewed 20 October 2026.