Virtual reality and psychosocial risks: training de-escalation before you need it

A legal assessment duty, not a "soft" topic

In many companies psychosocial risks are still treated as an optional subject, to be addressed when there is time. The legal framework says otherwise: D.Lgs. 81/2008, art. 28 paragraph 1, requires the risk assessment to cover all risks, including those linked to work-related stress, in line with the content of the European agreement of 8 October 2004.

On the methodological side, INAIL has published a methodology for assessing and managing work-related stress risk (2017 edition), organised into a preliminary phase based on objective indicators and an in-depth phase collecting workers' perceptions.

Many organisations get this far. The critical point is what comes next: which measures are adopted, and how you train the people who every day handle a customer raising their voice, an aggressive patient, a user who has lost control.

What the European data say

The First findings of ESENER 2024, published by EU-OSHA in 2025, are based on a survey conducted from May to October 2024 across 41,458 establishments in 30 countries. Some results:

• "Having to deal with difficult customers, patients or pupils" is reported by 56% of establishments in the EU-27 (it was 59% in 2019).
• Time pressure is reported by 43%.
• Among establishments with at least 20 employees that report the difficult-customer issue, only 46% have a procedure in place to deal with threats, abuse or violence (it was 51% in 2019). In health and social work the share rises to 66%.
• Establishments with an action plan against work-related stress account for 39% (33% in 2019).
• The main declared obstacle to addressing psychosocial risks is reluctance to talk openly about them (59%).

The Italian anomaly

In Italy 47% of establishments state that they have no psychosocial risk factor at all: the highest share in the EU, against an EU-27 average of 25%.

This is hard to read as an organisational achievement. More likely it is a problem of recognition and language: if the risk is not named, it does not enter the assessment and generates no measures. The topic connects directly to work-related stress prevention.

Why traditional training fails here

The skills required here are not knowledge: they are behaviours under pressure. Keeping your voice low when the other person raises theirs. Increasing physical distance instead of reducing it. Recognising the point at which it is better to call a colleague than to carry on alone.

A slide listing "the five rules of de-escalation" trains none of this. Classroom role play works better, but it depends on participants' willingness to expose themselves in front of colleagues and on the trainer's ability to vary the scenario credibly.

That is also why these contents are often reduced to an hour of awareness raising: they are hard to practise, not hard to explain.

What virtual reality can do

Here VR has a solid rationale, because the risk is not physical but relational, and it cannot be reproduced through a classic practical drill.

Repeatable, adjustable scenarios

• A customer or patient moving from complaint to threat, with adjustable escalation levels.
• The possibility of repeating the same situation several times, trying different strategies with no consequences.
• Contexts that are hard to simulate in a classroom: a service desk at the end of a shift, a crowded ward, a home visit.

Controlled exposure

The participant faces the tension in a protected environment, with the option to stop. It is the opposite of on-the-job learning, where the first experience of aggression is also the real one.

Behavioural observation

Unlike a quiz, the session shows what the person does: where they position themselves, how they respond, how long before they call for support, whether they keep the exit route clear. These are far more informative for the trainer. Anyone who has already worked with VR and 360° video in safety training recognises the difference.

Limits and cautions

This needs saying plainly: recent studies exist on the effectiveness of VR for de-escalation and aggression management training, but the evidence is still limited and methodologically heterogeneous. It is not correct to present it as a proven solution. It should be treated as a complementary tool, to be validated against your own internal indicators.

Other operational cautions:

• Risk of reactivating traumatic experiences. Someone who has suffered an assault may react intensely. Voluntary participation, advance information on the content and the option to stop at any moment are necessary.
• Privacy of behavioural data. Gaze, voice and movement traces are personal data: define purpose, legal basis and retention periods, and rule out individual appraisal uses.
• Debriefing led by a competent person. Without a guided discussion the session remains a strong experience and nothing more: the criteria described in debriefing for safety serious games apply.
• It does not replace organisational measures. Training de-escalation without reviewing staffing levels, desk layouts, alarm procedures and post-incident support shifts the burden onto the individual worker.

How to integrate it into the programme

1. Start from the assessment: identify roles and contexts where contact with difficult users is frequent.
2. Define observable target behaviours, not topics to cover.
3. Check that a procedure for threats, abuse and violence exists: the 46% figure says it often does not.
4. Use VR for repeated practice, not for one-off awareness raising.
5. Close every session with a debriefing and tracked corrective actions.
6. Measure over time: incident reports, results of perception questionnaires, cases handled without escalation.

In short

VR does not solve psychosocial risks: it makes part of the skills that govern them trainable. The precondition is still naming the risk, which according to ESENER 2024 almost half of Italian establishments claim not to have to do.