📝 1. Analytical summary

A avoidable risk, but difficult to predict

Every year, several hundred suicidal acts occur in establishments in France: in psychiatry, but also in medicine-surgery-obstetrics (MCO) and in medico-social. The HAS report of 2022 (reports of serious adverse events associated with care, 2017–2021) records 795 suicidal acts, of which about 65% are suicides; the suicidal acts are now the primary reason for reporting. These events would represent 1 to 2% of all suicides in France (estimation based on under-reported declarative data, p. 29). The suicide rate in psychiatric hospitalisation is more than ten times that of the general population (p. 29). Central point: individual prediction remains very weak — the majority of deceased individuals were not considered “at risk” during the last healthcare contact (p. 8).

A establishment plan, from detection to postvention

In response, the GEPS advocates a systemic and collective approach rather than individual responsibility. The document provides 10 recommendations covering the entire pathway: establishment plan and postvention plan, securing locations, training and availability of caregivers, regular assessment of suicidal thoughts, information transmission, liaison psychiatry in MCO, treatment of psychiatric and addiction disorders, preparation for transitions (permissions, discharges) with early recontact, systematic reporting and mortality-morbidity review (RMM). Each recommendation is accompanied by its level of evidence — most often limited or based on expert consensus — and specific actions directly transposable (p. 8–17, 51–85).

🔑 2. Key points of the document

1️⃣ The individual prediction fails — collective protection works (p. 40–42)

The ‘high risk’ questionnaires and scales do not improve the prediction and can mislead the caregiver in both directions. The winning logic is not to sort, but to reduce exposure to modifiable factors and to secure the environment during the crisis (p. 40–42, 37).

2️⃣ Three ‘false good ideas’ officially dismissed (p. 60, 69)

The constant observation of patients deemed at high risk, the ‘no suicide’ contract, and closed psychiatric units have not demonstrated their effectiveness in preventing suicide. The document calls for redirecting resources towards measures that do have an effect (p. 60, 69).

3️⃣ Prevention also concerns primary care and the medico-social sector (p. 18, 29)

If psychiatry is a priority, a significant portion of actions occur in somatic medicine, surgery, obstetrics, and in the medico-social establishment (including care homes). Emergency, paediatrics, and geriatrics are particularly concerned services (p. 29, 18).

4️⃣ Admission and discharge are the most at-risk periods (p. 31, 79–83)

A large cohort identifies the first week after admission and the weeks following discharge as the most critical. Prevention therefore starts before admission and continues well after discharge, hence the importance of a quick follow-up (ideally within 7 days) (p. 31, 8).

5️⃣ Suicide in establishments is a ‘triple shock’ that requires a prepared postvention (p. 30–31)

Families, other patients and teams are durably affected; an impact on the functioning of the service (stops, departures) is possible. The institutional postvention — a set of measures after a suicide — must be planned in advance, not improvised (p. 30–31, 52–53).

🧭 3. Action points for local stakeholders

1️⃣ Formalise a multimodal establishment plan (p. 52–53, 16–17)

Incorporate the prevention of suicide into the quality policy and risk management, in a multidisciplinary team (caregivers + management), and systematically associate a postvention plan. Use the summary table p. 16–17 as a framework for the action plan.

2️⃣ Conduct a multidisciplinary safety audit of the premises (p. 55–60)

Audit furniture and real estate to reduce access to means, then initiate the work without delay and ensure follow-up. Plan the inventory of personal belongings at entry and upon return from leave. A national audit tool (DGOS/GEPS) exists for this — see references.

3️⃣ Train the teams and secure availability (p. 61–64)

Train in the identification and management of suicidal crises, prioritising psychiatry, emergency, paediatrics and geriatrics, but also in MCO and medico-social. Ensure a sufficient staffing level and a caring attitude, particularly at night and on public holidays (more exposed periods).

4️⃣ Organise regular evaluation and follow-up contact after discharge (p. 65–68, 79–83)

Assess suicidal thoughts on admission and then regularly; prescribe a monitoring and protection plan tailored to each individual. Prepare permissions and discharges with the patient and their family, guarantee a quick appointment and a follow-up contact devise in the days following discharge (ideally within 7 days).

5️⃣ Systematise the reporting and the MRM (p. 84–85)

Report any suicidal act (mandatory procedure) and conduct a review of mortality-morbidity after each suicide or attempt: multi-professional analysis of the underlying causes leading to concrete corrective actions.

6️⃣ Combat preconceived ideas within teams (p. 42–43)

Rely on the list of preconceived ideas from the document ("talking about it would encourage acting on it", "those who talk about it don’t do it", etc.) as a support for awareness: asking about suicidal thoughts frees speech and does not create risk (p. 42–43).

Needs to be addressed / necessary adaptations

• The recommendations are designed for adult services: adapt them to the developmental specifics of children and adolescents, often hospitalised in paediatrics or adult psychiatry due to lack of dedicated beds (p. 50).

• Most of the measures fall under medical and institutional decisions: a non-clinical actor mainly intervenes upstream (culture, awareness) and downstream (guidance, support for the bereaved).

📚 4. Additional references

• Grid self-assessment of environmental security (DGOS, GEPS, HAS, FORAP, ANAP), January 2026 — the tool operational that accompanies recommendation no. 2. Presentation : santementale.fr

• Note information no. DGS/SP4/2025/114 of 28 July 2025 regarding the postvention (collective) — recent national framework, complementary to the postvention section (rec. no. 1). PDF : bulletins-officiels.social.gouv.fr

• Plan and deploy coordinated actions and services in suicide prevention — Practical guide for establishments (MSSS Quebec / CIUSSS de l'Estrie‑CHUS, 2024). Transferable "establishment plan" approach. PDF : publications.msss.gouv.qc.ca

Internal references Practices in Health around the theme of suicide - https://pratiquesensante.odoo.com/blog/tag/suicide-138

❓ 5. Frequently asked questions (FAQ)

1. Can we predict which patient will act?

No, the individual predictive capacity is very low: the majority of deceased individuals were not deemed at risk at the last contact. This is why the document prioritises collective protection rather than triage (p. 8, 40). [p. 8, 40]

2. Are risk questionnaires useful in practice?

Few: no "high risk" classification based on questionnaires has proven clinically useful; they are time-consuming and can be misleading. They still have a place in research and teaching, and as an interview guide (p. 42). [p. 42]

3. Should units be closed and patients monitored continuously?

No : closed units and constant observation have not demonstrated effectiveness, and hospitalisation also carries risks. An individualised monitoring plan and securing the locations are preferable (p. 46–47, 60, 69). [p. 60, 69]

4. Is it a subject reserved for psychiatry?

No : a significant portion of incidents occurs in MCO and in medico-social (including care homes). Emergency, paediatrics and geriatrics are particularly concerned; the plan must include these services (p. 18, 29). [p. 18, 29]

5. When should vigilance be at its maximum?

At the entry (first week) and around the discharge (following weeks), as well as at night and on public holidays, often in times of understaffing. Prevention continues after the administrative discharge (p. 31). [p. 31]

6. What should be done concretely at the patient's discharge?

Prepare the discharge with the patient and their family, ensure a quick appointment, plan a protection strategy and organise a follow-up contact in the following days, ideally within 7 days (p. 79–83). [p. 79–83]

7. What should be planned after a suicide in the establishment?

Activate the postvention: welcome and support the relatives, be attentive to other patients and caregivers, report the incident (mandatory) and conduct a mortality-morbidity review leading to corrective actions (p. 30–31, 84–85). [p. 30–31, 84–85]

🟢 6. Rewriting in Easy-to-Read Language

Easy-to-read and understand version of the summary and key points.

What is this document about?

• This document talks about suicide in hospitals and care facilities.

• Every year, people commit suicide during a stay in care.

• This happens in psychiatry. But also in other services.

• It’s very hard for families, other patients and caregivers.

• The GEPS is a group of suicide experts. They wrote this document.

What the experts say

• One cannot guess who will act. It’s almost impossible.

• So we must protect all people. Not just a few.

• We must make places safer.

• We must properly train caregivers.

• We must be there, gently, especially at night.

• We must communicate well among caregivers. And with the family.

• The moment of arrival and the moment of departure are the most dangerous.

• After discharge, we must call the person. Quickly. If possible within 7 days.

3 false ideas to forget

• Constantly monitoring a person: this does not work well.

• Having them sign a paper saying "I will not commit suicide": this does not work.

• Locking the doors everywhere: this does not provide better protection.

An important message

Talking about suicide with a person does not give them the idea.

On the contrary: asking the question helps the person to talk.

We can save many lives. We must know this and believe it.

🧭 7. Cross-sectional analysis — Values of Health Practices

▸ Literacy — The document proposes explicit work on vocabulary (non-stigmatising terms) and a list of misconceptions to combat, useful for informing teams and the surrounding community (p. 7, 42–43).

▸ Empowerment — Shared decisions between caregivers and patients and protection plans constructed with the patient are recommended; the patient is involved in preparing transitions (p. 64, 68, 80).

▸ Participation — Co-construction mainly focuses on the multidisciplinary team and the surrounding community; the direct participation of the individuals concerned in designing the recommendations is not described.

▸ Community health — The collective dimension is central within the establishment (shared culture, multimodal plan), but the territorial/community anchoring outside the walls remains secondary.

▸ Ethics — The authors highlight the ambivalence of the crisis and dismiss the moral register (courage/cowardice, freedom); they remind us of the risks of hospitalisation itself (p. 42–43, 47).

▸ Human rights — Attention to children/adolescents, the elderly, and suffering populations, and the refusal of stigma, reflect a concern for equity and inclusion (p. 7, 50).

▸ Intersectorality — Involved are management, caregivers, liaison psychiatry, architects, establishment committees, and the surrounding community; alignment with HAS 2025 criteria links quality and prevention (p. 18–19).

▸ Partnership — Models of collaboration are formalised: liaison psychiatry in MCO with protocols, structured information transmission, continuity between city and hospital (p. 70–73, 82).

▸ Fight against discrimination — The document explicitly combats judgments and the stigma of suicidal individuals, and promotes a non-judgmental and caring stance (p. 7, 42–43).

🔎 8. Evaluation of the reliability of the resource

Scientific relevance

High on the method: recommendations from a reference scholarly society, systematic literature review, reading group, HAS 2020 methodology, dated and referenced sources (bibliography p. 90 and onwards). Honest and acknowledged reservation: no recommendation reaches level of evidence 1; the majority falls under a limited level (3) or expert consensus (4), due to the difficulty of conducting randomised trials in this field. The incidence figures in establishments are partly based on declarative data (EIGS) recognised as under-estimated.

Operational relevance

Strong for establishments: actionable summary table, specific actions per recommendation, alignment with HAS certification, and associated national audit tool. For a prevention actor outside of establishments, the use is more indirect (culture, awareness, guidance, postvention). Clinical measures are subject to medical decisions: in accordance with the CEM V4 framework, clinical and normative points must be validated by the relevant professionals, this document serving to structure and prepare, not as a single source.

🏷️ 9. Strategic hashtags

#healthpractices #SuicidePrevention #MentalHealth #RiskManagement #Postvention #MedicoSocial #CareSafety #EvidenceBased

This article was developed in accordance with the Charter of the use of artificial intelligence of Health Practices. Click on the image