🚨 39% give up on mental health care: the paradox of public agents
🔍💡 Mental health of territorial agents: 45% report feeling exhausted every week, 39% have already given up on mental health care, and only 15% feel supported by their local authority. 🧭 A barometer to target actions where the needs are greatest.
📌 Nearly one in two territorial agents reports feeling exhausted every week, and the figure has been increasing since 2025. These agents are often the direct partners in prevention actions: social reception, schools, PMI, technical services, social centres. This barometer provides recent figures to objectify a situation that many feel. It also helps identify the most exposed groups (category C, supervisors, under 35s for access to care). It can be used to prepare a diagnosis, social dialogue, or a mental health project at work, including for one's own team.
Source :
📒 Barometer of mental health at work for public service agents — Edition 2026
✍️ Moodwork, idealCO and MNFCT (National Mutual of Territorial Civil Servants, partner of Macif) - September 2026 (survey from May to July 2026; 3rd edition after 2024 and 2025)
1️⃣ Analytical summary
Useful agents, engaged… and increasingly tired
The survey questions 6,513 territorial agents, across all professions and regions combined (p. 7, 9). 45% say they feel exhausted by their work at least once a week and 45% experience regular irritation, increasing by 3 points since 2025 (p. 12, 15). 35.6% perceive a risk to their mental health at work (p. 16). Category C agents, supervisors, and permanent staff seem to be the most affected (p. 12-17). Three factors emerge: workload, worry for the future, and emotional demand (p. 18). 30% of agents report experiencing incivilities several times a month (p. 21). 39% have already given up on mental health care, mainly for financial reasons (p. 34-35).
The meaning of work as a resource, collective support as a lack
The report highlights resources: 76% of agents take action for their mental health and 71% find meaning in their work, even though this figure is decreasing (−8 points since 2024) (p. 24, 30). Meaning is associated with less exhaustion, but very strong commitment can also lead to exhaustion (p. 32). Perceived support from the community is associated with about a third lower exhaustion, while only 15% perceive it (p. 39). 74% of agents have never been trained on psychosocial risks (p. 38). Each chapter proposes levers for action, and seven initiatives from communities are presented at the end of the document (p. 45-51).
2️⃣ Key points of the document
1️⃣ A slow but continuous deterioration
Weekly exhaustion and irritation have increased by 3 points in a year (45% each); the perceived risk has increased by 1.6 points (p. 12, 15, 16). Comparisons from one year to the next should be read as trends, as the samples are not identical (p. 9).
2️⃣ Marked inequalities according to category and function
Category C agents accumulate uncertainty, emotional demands and incivilities. Supervisors accumulate high workload and emotional demands (p. 18-20). A stable status does not seem to protect against perceived risk (p. 17).
3️⃣ Incivilities and sleep
Among the agents most exposed to incivilities, 43% report daily sleep disturbances, compared to 13% among the non-exposed (p. 21). This is an association, without an established causal link.
4️⃣ Renouncing treatment, a new angle
39% of agents have already renounced mental health treatment. Reasons: money (59%), lack of time (55%), access to specialists (37%), fear of judgement (17%) (p. 34-35). Women, category C and those under 35 renounce more (p. 34).
5️⃣ Meaning protects, but can also wear down
The facets of meaning statistically explain up to 52% of the gaps in burnout (p. 32). The alignment of values between agent and community is the primary source of meaning (p. 31). The sense of usefulness can also fuel over-commitment (p. 8, 32).
3️⃣ Action points for local actors
1️⃣ Target actions at the most exposed audiences
Rather than a ‘one-size-fits-all’ action, prioritise category C agents, the supervisors and, for access to treatment, women and those under 35 (p. 8, 34, 41). A local diagnosis, even brief, can check if these profiles are present in its structure.
2️⃣ Remove financial and practical barriers to access to treatment
Inform about existing support, negotiate low-cost consultations, identify internal relays (occupational medicine, nurse, social service assistant) (p. 34-35). Make visible “ who to contact, for what, in how much time” (p. 8).
3️⃣ Train on psychosocial risks from the start of the position
Integrate a RPS awareness session during the welcome of new agents, then renew it, and monitor its effects through a questionnaire (p. 38). The supervisors, “first preventers”, are a priority target (p. 40).
4️⃣ Treat incivilities as an organisational risk
Set up an incident register, immediate support (for example first aid in mental health) and follow-up support (p. 22). Do not refer this subject to individual feelings alone.
5️⃣ Open structured speaking spaces
Speaking groups, time for discussion on difficulties, social dialogue dedicated to mental health (p. 17, 20, 39). On addictive behaviours, ensure anonymity and confidentiality (p. 28).
6️⃣ Draw inspiration from field initiatives, with caution
Participatory QVCT forum (Sceaux, p. 49), multidisciplinary approach (Marseille, p. 45), identification of “brown-out” and “bore-out” (Pas-de-Calais, p. 51). Their results are reported by the local authorities: plan for your own evaluation before transposing.
Unmet needs: the report does not describe a reusable diagnostic method or validated tool. It also does not detail the situations of agents in contact with vulnerable populations (social, PMI, early childhood) as a specific group. Local adaptation remains necessary.
4️⃣ Additional references
External resources
1️⃣ INRS — Campaign "Let's talk about mental health at work!" (March 2025)
Guide "9 tips for daily action" (ED 6250), prevention brochure (ED 6349) and RPS factor assessment tool (ED 6403), aimed at managers and HR. Methodological complement directly related to the levers of the barometer. inrs.fr — Let's talk about mental health at work
2️⃣ FIPHFP — Toolkit "Mental health at work: understanding, detecting and acting sustainably" (May 2026)
Resources oriented towards the public service: brochure, awareness materials, peer support, helplines. Useful for the "access to care" and "de-stigmatisation" tracks. fiphfp.fr — Toolkit (PDF)
3️⃣ EU-OSHA — "France: The pulse of OSH 2025 – Mental health and digital transformation at work" (September 2025)
National report from the European Agency for Safety and Health at Work: exposure to psychosocial risks and prevention measures in France. It allows for situating the barometer figures in a broader context. osha.europa.eu — OSH Pulse 2025 France
5️⃣ Frequently asked questions (FAQ)
1️⃣ Who responded to the survey, and are the results representative ?
6 513 territorial agents responded online, voluntarily (p. 9). The results are adjusted according to the professional category only. They provide a strong trend, but other biases (age, profession, type of community) remain possible (p. 8-9).
2️⃣ Which agents seem to be the most in difficulty?
Category C agents and supervisors report more exhaustion; the permanent staff report more irritation and perceived risk (p. 12-17). For access to care, it is mainly women, category C and those under 35 (p. 34).
3️⃣ Why do agents give up on mental health care?
Firstly for financial reasons (59 %), then due to lack of time (55 %) and access to a specialist (37 %). Fear of judgement (17 %) and the lack of information (13 %) also play a role (p. 35).
4️⃣ Does the meaning of work really protect?
Meaning is associated with less exhaustion, especially when autonomy is high (p. 32). But feeling very useful can also lead to overcommitment. These are statistical associations, not a proof of protection (p. 8, 32).
5️⃣ Are agents trained in psychosocial risks?
No, for the majority: 74 % have never been trained and 60 % say they are little or not at all aware (p. 38). The average knowledge score they give themselves is 4.4 out of 10.
6️⃣ What role do supervisors play?
They are both exposed (high emotional load and demand) and essential: support from management is the second factor associated with less burnout (p. 19, 39). Management is the most cited theme in open comments (20 %) (p. 40).
7️⃣ Are the initiatives presented at the end of the report evaluated?
They are described by the communities based on feedback from experience (p. 44-51). For example, the 4-day week in Lyon reports 87 % of agents feeling better (p. 47). No independent evaluation is cited: these are sources of inspiration.
6️⃣ FALC version (Easy to read and understand)
What does this document talk about?
This document is a survey.
It talks about the mental health of agents from municipalities, departments and regions.
Mental health is feeling good in your mind.
6 513 agents answered questions online.
The survey took place from May to July 2026.
What the agents say
Almost one in two agents feels very tired each week.
This tiredness increases a little each year.
The causes: too much work, fear for the future, having to hide their emotions.
Three agents in ten often suffer insults or violence from the public.
Many of these agents sleep poorly.
Four agents in ten have already given up seeing a psychologist or a doctor.
The main reason: it costs too much.
What helps the agents
Many agents do sports or relax to feel better.
Seven out of ten agents find their work useful.
Feeling useful helps to cope.
But investing too much can also be tiring.
When the community supports the agents, they are less tired.
However, few agents feel supported.
Three out of four agents have never received training on these topics.
What can be done
Help prioritise the most tired agents.
Help pay for appointments with a psychologist.
Train the agents from their arrival.
Train the team leaders.
Create times to talk about difficulties.
7️⃣ Cross-sectional analysis — values of Health Practices
Literacy : the operational summary (p. 7-8) and the boxes "What needs to be maintained" facilitate reading; the document does not provide tools suitable for agents with a low reading level.
Empowerment : the agents are surveyed and 76% are already taking action for their health (p. 24), but they do not participate in the design or analysis of the survey.
Participation : more than 1,700 free comments have been analysed (p. 40); discussion groups and a dedicated social dialogue are recommended (p. 17, 39).
Community health: the collective dimension appears through support among colleagues and informal collective times (p. 31); the approach remains focused on the organisation of work.
Ethics : the methodological limits are noted (p. 8-9); the interests of the publishers (prevention provider, mutual) are not discussed in the document.
Human rights: the inequalities in access to care based on gender, age and category are highlighted (p. 34), which supports an equity approach.
Intersectorality : internal links (occupational medicine, social worker, psychologist) and culture-health links (SolYmusées, p. 50) are mentioned.
Partnership : the barometer associates a company, a platform and a mutual; the multidisciplinary governance of Marseille proposes a charter of collaboration (p. 45).
Fight against discrimination: harassment and sidelining are mentioned in the comments (p. 40); discrimination related to origin, disability or orientation is not addressed.
8️⃣ Evaluation of the reliability of the resource
Scientific relevance — moderate
Strengths : large sample (6,513), 3rd edition allowing for tracking of trends, explained adjustment, reading precautions clearly stated (p. 8-9), item formulation in notes (p. 18).
Limits : voluntary participation via partner networks (risk of over-representation of concerned agents), adjustment on a single criterion, statistical methods not detailed (“explained variance”, p. 32), no scientific references cited, no validated scale identified.
Vigilance : the conclusion speaks of a « demonstrated » protective effect of collective support (p. 41), whereas the method only allows for establishing associations (p. 9). The publishers offer services for prevention and social protection: the document also highlights a visibility approach.
Operational relevance — average to good
Concrete action levers in each chapter, useful for building a prevention plan or an argument.
Varied experiential feedback that is transferable (p. 45-51), but without independent evaluation data.
Recommended use: as a numerical reference and dialogue support, in addition to a local diagnosis and methodological resources (INRS, FIPHFP).
9️⃣ Strategic hashtags
#healthpractices #MentalHealthAtWork #LocalPublicService #PsychosocialRisks #QVCT #Prevention #RenunciationOfCare #MeaningAtWork
You can support Health Practices: https://www.pratiquesensante.com/blog/annonces-28/pour-soutenir-pratiques-en-sante-3253
This article was developed in accordance with the Charter
for the use of artificial intelligence by Health Practices.
Click on the image 