VR therapy for psychosis (gameChange)

1. Introduction

Version followed: the gameChange automated VR cognitive therapy as specified in the trial protocol (Freeman et al., BMJ Open, ISRCTN17308399) and the multicentre trial report (Freeman et al., Lancet Psychiatry 2022;9(5):375–388), and the commercial product gameChangeVR (Oxford VR) as described in NICE early value assessment HTE15, now listed on the NICE site as HTG701. Later or local service versions may differ in session count or staffing; these are noted where sources state them.

gameChange is an automated, immersive virtual reality (VR) therapy for people with psychosis who avoid everyday places mainly because of anxiety (agoraphobic avoidance). It was developed by Daniel Freeman's team at the University of Oxford with people with lived experience of psychosis, clinical psychologists, designers and software developers, and programmed by Oxford VR, a University of Oxford spin-out [source: Freeman et al. BMJ Open protocol; Lambe et al. 2020]. A virtual coach (an on-screen avatar) guides the person through computer-generated everyday situations. A mental health worker (in the trial: a peer support worker, assistant psychologist or clinical psychologist) is present in the room or supports remotely, sets up the equipment, and links what is learned in VR to real-world practice between sessions [source: Freeman et al. 2022; NICE HTG701 ch. 2].

Core idea of change: it is framed as cognitive therapy, not simple exposure. The person carries out repeated behavioural experiments in which they drop defence (safety-seeking) behaviours, so that threat beliefs about other people and situations can be tested and updated and a sense of safety can be learned. The person is not asked to stay in a situation until anxiety falls [source: Freeman et al. BMJ Open protocol; Freeman et al. 2022]. In the trial, reductions in threat cognitions and within-situation defence behaviours each partly explained the treatment effect [source: Freeman et al. 2022].

Who delivers and status. The coach is software; the human role is a supporting mental health professional. In the trial, supporters needed no prior cognitive therapy experience; they had half a day of training, a written manual and weekly supervision [source: Freeman et al. 2022]. NICE notes one NHS trust trained band 3 to band 5 staff in about half a day with supervision inside clinical teams [source: NICE HTE15]. A psychotherapist may take the supporter role, or work alongside the person and the service that delivers it.

Regulatory and NICE status (as of October 2026 per NICE): NICE early value assessment HTE15, published 15 November 2023 and now listed as HTG701, says gameChangeVR can be used in the NHS while more evidence is generated, for severe agoraphobic avoidance in people with psychosis aged 16 and over, with support from a mental health professional. For mild to moderate agoraphobic avoidance in psychosis it is recommended only in research, with access through company, research or non-core NHS funding. The company must have evidence-generation agreements in place and confirm progress to NICE yearly; the evidence-generation period is 3 years, or shorter if enough evidence emerges, after which NICE decides on routine adoption. All the technologies may be used only once they hold appropriate regulatory approval [source: NICE HTG701 recommendations; NICE HTE15 pdf]. Whether NICE had finished a review of routine adoption by October 2026 [uncertain: not verified]. At the time of the trial the software was a CE-marked class I medical device [source: Freeman et al. BMJ Open protocol; Freeman et al. 2022]; current UKCA/CE marking [uncertain: not verified]. An NIHR-funded real-world waitlist randomised trial (IRAS 351247; 200 participants with psychosis and the most severe avoidance, often housebound; practice over eight weeks; ethics favourable opinion 17 March 2025) is generating evidence for NICE [source: HRA research summary]. Outside the UK, regulatory status [uncertain: not verified].

Evidence in brief (attributed). In the trial (reported as 346 participants across nine NHS trusts [source: GMMH news, 2023]), gameChange plus usual care reduced agoraphobic avoidance and distress at 6 weeks compared with usual care alone (small effects); overall differences were not significant at 26 weeks; benefits were concentrated in, and maintained for, people with severe baseline avoidance; most secondary outcomes did not differ. The authors note the trial cannot separate the VR effect from contact with the staff member [source: Freeman et al. 2022]. NICE lists evidence gaps on long-term benefit, who benefits most, relapse and extra sessions, adverse effects, staff grade and time, and comparison with standard NHS care [source: NICE HTE15].

2. Framework

gameChange is a short, structured, software-led programme with a human supporter. The structure below follows the trial version [source: Freeman et al. BMJ Open protocol; Freeman et al. 2022; Lambe et al. 2020] unless marked.

Phase A: Referral and assessment (before session 1).

Goals: confirm a psychosis diagnosis and anxious avoidance of going out that the person wants help with; gauge severity; screen physical and VR suitability; agree goals; obtain consent.
Trial eligibility: aged 16 or over; schizophrenia-spectrum diagnosis (ICD-10 F20–F29) or an affective disorder with psychotic symptoms; self-reported difficulty going outside mainly because of anxiety; able to consent. Exclusions included photosensitive epilepsy; substantial visual, auditory or balance impairment; current intensive psychological therapy; insufficient English; forensic or psychiatric intensive care settings; organic syndrome; primary substance use or personality disorder; clinically significant learning disability; active suicidal intent with plans [source: Freeman et al. 2022]. These were research criteria; services set their own, and NICE asks clinicians to assess individual safety and suitability with the person and, where appropriate, carers [source: NICE HTE15].
NHS use under NICE is for severe avoidance; NICE does not publish a severity cut-off on the pages seen [uncertain: not verified].
Move on when: eligibility and suitability are confirmed, the person has tried a headset or seen a demonstration, and has agreed goals and consent [synthesis].

Phase B: VR treatment sessions (about six sessions over about 6 weeks).

Format: about six sessions with roughly 30 minutes in VR each, usually weekly; three sessions was the minimum adherent dose in the trial; people progress at their own pace [source: Freeman et al. 2022; BMJ Open protocol]. A clinical expert told NICE sessions last about 20 minutes [source: NICE HTE15]. The current evidence-generation trial describes practice over eight weeks [source: HRA research summary].
Setting: the person's home or an NHS clinic room; the person usually stands and can walk a few paces [source: Freeman et al. 2022].
Session 1: the coach meets the person in the coach's virtual office, teaches the basic controls, explains the rationale, and the person chooses a first situation [source: Freeman et al. 2022].
Each session: belief rating at start; choose a situation (new or repeated) and work through levels with the coach; belief rating at end; supporter sets real-world practice and reviews it next time [source: Freeman et al. 2022].
Six situations (in general terms): a café, a GP waiting room, a pub, a bus journey, opening the front door onto the street, and a small local shop. Each has five difficulty levels that increase the number and closeness of people and the amount of social interaction [source: Freeman et al. 2022; Lambe et al. 2020].
Moving between levels and situations: the person chooses, guided by the coach, who asks for feedback to tailor progression; they can repeat earlier situations [source: Freeman et al. 2022].

Phase C: Real-world practice (between and after sessions).

Goal: transfer learning to the person's actual streets, shops and transport. The supporter sets practice tasks and reviews completion at the next session [source: Freeman et al. 2022].

Phase D: Ending and follow-up.

The trial measured outcomes at 6 weeks (end of treatment) and 26 weeks [source: Freeman et al. 2022]. A modified satisfaction questionnaire was completed after the final session [source: BMJ Open protocol]. Structured ending, relapse planning and booster or extra VR sessions are not described in the sources seen; NICE lists extra sessions and relapse as evidence gaps [source: NICE HTE15]. The ending steps given in section 4 are [synthesis].

3. Guiding principles

Cognitive, not just exposure [source: BMJ Open protocol; Freeman et al. 2022]: frame each situation as a test of a feared prediction; success means learning something, not simply tolerating anxiety.
Dropping defence behaviours is the active ingredient [source: Freeman et al. 2022; Lambe et al. 2020]: notice and gently encourage letting go of avoidance-within-the-situation (looking down, keeping distance, hurrying, hiding) so the person discovers they are safe without them.
Learning safety [source: Lambe et al. 2020 abstract]: the aim is an updated belief that everyday places are safe enough, not certainty that nothing bad could happen.
Person-led pacing and choice [source: Freeman et al. 2022]: the person chooses the situation and pace; the supporter does not push levels.
The coach delivers the therapy; the human supports it [source: Freeman et al. 2022; NICE HTG701]: the supporter's job is set-up, encouragement, safety and generalisation, not adding a second therapy on top during the VR.
Lived-experience design and peer support [source: BMJ Open protocol; GMMH news]: the programme was co-designed with people with psychosis, and peer support workers were favoured by the lived-experience panel as supporters; value the supporter's own recovery knowledge.
Real-world transfer is part of the treatment [source: Freeman et al. 2022]: every VR session should lead to an agreed small step outside.
Suitability is individual [source: NICE HTE15]: VR does not suit everyone; assess safety and suitability case by case and discuss options with the person and carers.
An add-on, not a replacement [source: NICE HTG701 ch. 2]: VR is not intended to replace treatment for co-occurring conditions, nor antipsychotic medication; keep the care team involved.
Prepare people for side effects [source: Freeman et al. 2023]: tell people in advance about possible side effects and how to respond, and address worries about what is happening in the room while their vision is covered.
Do not exclude on symptom severity alone [source: Freeman et al. 2023]: in the trial, paranoia, voices, depression and anxiety at baseline did not predict side effects.
Supervised delivery [source: Freeman et al. 2022]: supporters work from a manual with regular supervision; keep to the programme rather than improvising new VR tasks.

4. Intervention inventory

The inventory covers (a) the programme components as published, which the coach delivers and the supporter facilitates, and (b) the supporting clinician's or psychotherapist's own tasks. Example language is illustrative unless quoted.

4.1 Referral and suitability assessment

Purpose: identify people with psychosis whose avoidance of going out is mainly anxiety-driven and who want help with it; confirm the NHS indication (severe avoidance) or a research route (mild to moderate) [source: NICE HTG701; Freeman et al. 2022].
Steps: (1) Confirm the diagnosis with the care team (schizophrenia-spectrum or affective psychosis) [source: Freeman et al. 2022]. (2) Map avoidance: which places, how often the person goes out, whether accompanied, what they fear will happen (e.g. being attacked, stared at, judged, losing control) and what they do to stay safe there [synthesis]. (3) Rate avoidance and distress with the service's chosen measure; the trial used the Oxford Agoraphobic Avoidance Scale (O-AS), which records whether the person can do a set of everyday tasks and how distressing each is; do not reproduce items without permission [source: Freeman et al. 2022]. (4) Check exclusions used in the trial as a guide to caution (section 2) and current risk. (5) Check whether the person is already in intensive psychological therapy and agree sequencing with that therapist [source: Freeman et al. 2022; synthesis]. (6) Decide with the person and team.
Example language (illustrative): "Lots of people with experiences like yours find going out feels unsafe. This programme lets you practise everyday places in a headset first, then try small steps for real. Can we look at which places matter most to you?"
Watch for: avoidance driven mainly by something other than anxiety (e.g. negative symptoms, physical illness, a realistic current threat such as harassment, lack of money) where VR is not the right first step [synthesis]; severity level, since benefit was concentrated in severe avoidance [source: Freeman et al. 2022].

4.2 Cybersickness and physical-safety screening

Purpose: reduce the risk of side effects and falls, and spot people for whom headset use is unsafe [source: Freeman et al. 2022; Freeman et al. 2023; NICE HTE15].
Steps: (1) Ask about photosensitive epilepsy or seizures triggered by flashing light; this was a trial exclusion [source: Freeman et al. 2022]. (2) Ask about significant visual, hearing or balance problems (trial exclusions) [source: Freeman et al. 2022]. (3) Ask about past motion or travel sickness, migraine, and reactions to video games or 3D films [synthesis]. (4) Ask how the person feels about having their vision covered, including worry about what others in the room are doing [source: Freeman et al. 2023]. (5) Check the space: clear floor, no trip hazards, cables managed, a chair within reach; the person usually stands and may walk a few paces [source: Freeman et al. 2022; synthesis for room checks]. (6) Check the headset manufacturer's health and safety information for its own age and medical cautions [synthesis; uncertain: not verified]. (7) Explain likely side effects: NICE reports motion sickness, dizziness and mild headaches, described by experts as generally mild and transient; in the trial side-effect study the most common items were difficulty concentrating because of thoughts about what was happening in the room (about 14%), a lasting headache (about 8%) and headset-induced panic (about 7%); about 5% felt sick and no one vomited [source: NICE HTE15; Freeman et al. 2023].
Note on measures: the trial team did not use the Simulator Sickness Questionnaire because its items overlap with anxiety symptoms; they used the 33-item yes/no Oxford-VR Side Effects Checklist, completed at the final VR session [source: Freeman et al. 2023]. Use whichever measure the service adopts; do not reproduce items without permission.
Example language (illustrative): "Some people get a headache or feel a bit dizzy or sick in a headset. If that happens, tell me or just take it off. Nothing bad happens if you stop. Has anything like travel sickness or flashing lights been a problem for you?"
Watch for: pallor, sweating, unsteadiness, eye rubbing, the person reaching for the headset; separate cybersickness from anxiety by asking what they notice (nausea and spinning versus fear thoughts) [synthesis].

4.3 Informed consent, psychoeducation and data conversation

Purpose: make sure the person understands what the programme is, who supports it, what it can and cannot do, and how their information will be used [source: NICE information for the public; synthesis].
Steps: (1) Explain the virtual coach and your role. (2) Explain the rationale in plain words: testing fears and letting go of safety behaviours. (3) Explain likely side effects (4.2) and that they can stop any time. (4) Explain that it adds to, not replaces, current care and medication [source: NICE HTG701 ch. 2]. (5) Explain data: NICE's public information notes the NHS is collecting more evidence, so the person may be asked whether their treatment details can be used, and advises them to ask how information is stored and used [source: NICE information for the public]. Find out from your service's information governance lead what the software and headset record, where it is stored, whether Wi-Fi upload occurs (NICE notes some technologies need Wi-Fi) and who can see it [source: NICE HTE15; synthesis]. (6) Invite NICE's suggested questions: how well it works compared with other options, risks, effect on daily life, what happens if it does not work, alternatives [source: NICE information for the public]. (7) Record consent.
Example language (illustrative): "You'll be guided by a coach inside the headset; I'll be here to help with the kit and to plan practice outside. It won't replace your care team or your medication. You can ask me any time how your information is kept."
Watch for: paranoid concerns about recording or being watched through the device; take these seriously, give factual answers about what is and is not recorded, and offer to show the equipment [synthesis; source: Freeman et al. 2023 notes such suspicion is rare in clinical experience].

4.4 Session set-up and headset orientation (session 1)

Purpose: get the person comfortable with the equipment and the coach, and give the rationale [source: Freeman et al. 2022].
Steps: (1) Set up the headset and computer before the person arrives or with them at home. In the trial this was an HTC Vive Pro headset and a laptop [source: Freeman et al. 2022]; use the hardware your service supplies. (2) Introduce the concepts briefly. (3) Help fit the headset and start the programme [source: Freeman et al. 2022]. (4) The coach meets the person in the coach's virtual office, teaches the controls and explains the rationale; the person answers the coach by moving a virtual slider or touching a virtual ball or globe [source: Freeman et al. 2022; BMJ Open protocol]. (5) Tell the person where you are in the room and that you will stay there; speak occasionally so they know you are present [synthesis; based on Freeman et al. 2023 finding on room-related worry].
Example language (illustrative): "I'm sitting just here by the door. I'll stay put. If you want to stop, say so or lift the headset off."
Watch for: difficulty with controls (common in people unused to gaming), headset discomfort, panic when vision is covered [source: Freeman et al. 2023; synthesis].

4.5 Belief and confidence ratings

Purpose: track the person's confidence in social situations within and across sessions, which the coach uses to tailor progress [source: Freeman et al. 2022; BMJ Open protocol].
Steps: (1) The coach takes a belief rating about confidence in social situations at the start and end of each session, in VR [source: Freeman et al. 2022]. (2) After the session, the supporter can ask what changed between the two ratings and why [synthesis].
Example language (illustrative): "Your confidence went up a bit by the end. What do you think made the difference today?"
Watch for: ratings that do not shift over several sessions, which may mean safety behaviours are still being used or the level is too easy or too hard [synthesis].

4.6 Choosing a situation and level

Purpose: let the person work at a manageable but meaningful challenge [source: Freeman et al. 2022].
Steps: (1) Each session the person chooses one of the six situations (café, GP waiting room, pub, bus, front door onto the street, small shop) and can repeat earlier ones [source: Freeman et al. 2022]. (2) Each situation has five levels with more and closer people and more interaction at higher levels [source: Freeman et al. 2022]. (3) The coach asks for feedback to tailor progression [source: Freeman et al. 2022]. (4) Before starting, the supporter can help link the choice to the person's real-world goals (e.g. choosing the bus if the goal is a clinic appointment) [synthesis].
Example language (illustrative): "Which of these places would make the biggest difference in your week if it felt easier?"
Watch for: repeatedly choosing the easiest situation without progress (possible avoidance in VR), or jumping to the hardest level and becoming overwhelmed [synthesis].

4.7 VR behavioural experiments: dropping defence behaviours

Purpose: the core change process; test threat beliefs by reducing within-situation defences [source: Freeman et al. 2022; Lambe et al. 2020].
Steps (as the coach delivers them, in general terms): (1) The coach encourages the person to let go of safety behaviours. (2) Coaching encourages moving closer to virtual people and making eye contact [source: Freeman et al. 2022]. (3) The person is not asked to wait for anxiety to fall; the point is to notice what actually happens when defences are dropped [source: BMJ Open protocol]. (4) The supporter's role during VR is to stay calm and quiet, encourage when asked, and not add instructions that conflict with the coach [synthesis].
Example language (illustrative, supporter after VR): "When you looked up at the people in the queue, what did you expect would happen? And what did happen?"
Watch for: subtle safety behaviours that continue (looking at the floor, staying near exits, rushing the task); in-VR improvement that is not reported as belief change [synthesis]. Change in threat beliefs and in within-situation defences were the processes that partly explained benefit in the trial [source: Freeman et al. 2022].

4.8 Game-type tasks and spoken interaction

Purpose: make experiments engaging and require active approach and interaction [source: Freeman et al. 2022; BMJ Open protocol].
Steps: (1) Some levels include game-type tasks, for example bursting bubbles in the café [source: Freeman et al. 2022]. (2) Some levels require spoken replies to on-screen characters, detected by voice recognition [source: Freeman et al. 2022]. (3) If voice recognition struggles, the supporter troubleshoots (volume, microphone) without breaking the person's focus [synthesis].
Example language (illustrative): "Speaking to someone in there counts as a real step. How did it feel to answer them?"
Watch for: frustration with technical glitches being read as personal failure; self-consciousness about speaking aloud with the supporter present [synthesis].

4.9 Post-session check and debrief

Purpose: settle any side effects, consolidate learning, and spot worries after VR [source: NICE HTE15; Freeman et al. 2023; synthesis for debrief structure].
Steps: (1) Help remove the headset; let the person sit and reorient. NICE notes some VR protocols include grounding exercises to help people reorient [source: NICE HTE15]. (2) Ask briefly about headache, dizziness, nausea and any worries after VR; "worries following VR" emerged as a side-effect factor in the trial [source: Freeman et al. 2023]. (3) Ask what the person predicted, what they did differently, and what they learned [synthesis]. (4) Do not let the person leave unsteady; advise against driving or cycling until any dizziness has passed [synthesis]. (5) Record side effects and pass anything notable to the care coordinator [synthesis].
Example language (illustrative): "Take a minute before you stand. Any headache or wobbliness? What's the one thing you'd want to remember from today?"
Watch for: lingering dizziness or headache; new or increased suspicious thoughts about the device, the coach or the supporter; distress about the content [synthesis; Freeman et al. 2023 reports such suspicion is rare].

4.10 Setting and reviewing real-world practice

Purpose: transfer VR learning to daily life; the supporter sets homework tasks and reviews completion at the next session [source: Freeman et al. 2022; BMJ Open protocol].
Steps: (1) Pick a real situation that matches the VR work (e.g. sitting in a local café for a short time, a single bus stop) [synthesis]. (2) Name the feared prediction and the safety behaviour to drop [synthesis, consistent with the programme's rationale]. (3) Agree when, where and with whom; peers or carers may accompany at first, then step back [synthesis]. (4) Plan what to do if anxiety or voices increase (stay, use a coping step, or leave and try again later) [synthesis]. (5) At the next session, review what happened, what was learned, and set the next step [source: Freeman et al. 2022 for review; synthesis for structure].
Example language (illustrative): "In the headset you managed the bus with people sitting close and you didn't look away. Could you try one real stop this week, and see whether people react the way you fear?"
Watch for: tasks not done because they were too big; practice done with all the usual safety behaviours; genuinely unsafe settings where caution is realistic (adjust the task rather than push) [synthesis].

4.11 Outcome and progress monitoring

Purpose: judge benefit and inform decisions about continuing, repeating or stopping [source: Freeman et al. 2022; NICE HTE15; synthesis].
Steps: (1) Repeat the avoidance and distress measure at the end (the trial used the O-AS at baseline, 6 and 26 weeks; a one-point drop in the avoidance score was taken as clinically meaningful) [source: Freeman et al. 2022]. (2) Record side effects with the service's checklist [source: Freeman et al. 2023]. (3) Track real-world activity: how often the person goes out and where [synthesis]. (4) Contribute data to NICE evidence generation if the service is a participating site and the person consents [source: NICE HTG701 evidence generation plan; NICE information for the public].
Example language (illustrative): "Let's compare where you were at the start with now, on the same questions, so we can both see what's changed."
Watch for: gains at end of treatment that fade by follow-up; in the trial overall differences were not significant at 26 weeks except in people with severe avoidance [source: Freeman et al. 2022].

4.12 Risk management in psychosis

Purpose: keep the person safe while engaging with fear-provoking material and going out more [source: BMJ Open protocol; Freeman et al. 2022; synthesis].
Steps: (1) Before starting, review current risk with the care team, including suicidal thinking, self-neglect, risk from or to others, and the relapse signature; active suicidal intent with plans was a trial exclusion [source: Freeman et al. 2022]. (2) Have the person's crisis plan and local crisis line details to hand [synthesis]. (3) If at home, follow lone-working policy [synthesis]. (4) At each session ask briefly about mood, voices, paranoia and sleep since last time [synthesis]. (5) Stop the session if acute distress, a marked rise in persecutory ideation focused on the device or supporter, or signs of relapse appear; contact the care coordinator or duty team the same day [synthesis]. (6) Record and report adverse events through the service's procedure; in the trial, a standard operating procedure covered adverse events, serious adverse events were defined under ISO 14155:2011 (e.g. deaths, suicide attempts, serious violence, hospital admissions) and adverse device effects included those from software malfunction, inadequate instructions or user error [source: BMJ Open protocol]. Report device-related incidents also through the national medical device reporting route (in the UK, the MHRA Yellow Card scheme) [synthesis; uncertain: not verified]. (7) Any question about medication goes to the prescriber; never suggest changing it [synthesis]. Clinical experts advised NICE that VR would not be offered instead of antipsychotic medication [source: NICE HTG701 ch. 2].
Trial safety data (attributed): serious adverse events were similar between groups and none in the VR group were rated as related to treatment [source: Freeman et al. 2022].
Example language (illustrative): "Before we put the headset on: how have the voices and the worries about people been this week? If anything gets too much today, we stop, and we'll make sure your team knows."
Watch for: early warning signs of relapse; increased suicidal thinking as the person confronts how restricted their life has been; risky real-world tasks (e.g. late-night outings in unsafe areas) [synthesis].

4.13 Supporter training, supervision and fidelity

Purpose: safe, consistent delivery by non-specialist staff [source: Freeman et al. 2022; NICE HTE15].
Steps: (1) Complete the provider's training (half a day in the trial) and read the written manual [source: Freeman et al. 2022]. (2) Attend regular supervision (weekly in the trial) with a qualified clinician [source: Freeman et al. 2022]. (3) Bring to supervision: lack of progress, risk concerns, side effects, ethical issues, the supporter's own reactions (especially for peer workers drawing on lived experience) [synthesis]. (4) If you are a psychotherapist supervising peer workers, focus on boundaries, risk escalation and generalisation tasks rather than adding CBT techniques to the VR session [synthesis].
Example language (illustrative, supervisor): "Talk me through the last session: what level, what safety behaviours you noticed, and what homework you set."
Watch for: drift into extra unscripted interventions; supporters pushing progression; burnout or over-identification in peer workers [synthesis].

4.14 The psychotherapist working alongside the programme

Purpose: integrate gameChange with wider psychological care [synthesis; NICE HTG701 ch. 2 notes it does not replace treatment for co-occurring conditions].
Steps: (1) If the person is in CBT for psychosis or other therapy, agree whether gameChange runs before, alongside or after it; the trial excluded people in current intensive psychological therapy [source: Freeman et al. 2022; synthesis for sequencing]. (2) Use the VR experiments as shared material in therapy (predictions, defences, learning) [synthesis]. (3) Address what VR does not cover: trauma, low mood, hopelessness, voices content, practical barriers [synthesis]. (4) Liaise with the care coordinator and prescriber about progress, side effects and risk, with consent [synthesis]. (5) After the course, help keep up real-world practice and plan for setbacks [synthesis].
Example language (illustrative): "What you learned on the virtual bus is something we can keep building on here, with the places in your real life."
Watch for: duplicate or conflicting tasks between services; the person feeling handed over to a machine [synthesis].

4.15 Ending and relapse planning

Purpose: consolidate change and plan next steps; not detailed in the sources seen [synthesis].
Steps: (1) Review progress with the measures and the person's own goals. (2) Write a short personal summary of learning: feared predictions, what happened, safety behaviours dropped. (3) Agree a real-world practice plan for the next months. (4) Name early signs of returning avoidance and who to contact. (5) Discuss whether further sessions might help; NICE lists the use and effect of extra VR sessions as an evidence gap [source: NICE HTE15]. All steps except (5) are [synthesis].
Example language (illustrative): "If you notice you're staying in more again, that's a signal to restart small steps, not a sign it's all gone. Who would you tell?"
Watch for: abrupt loss of support at course end, especially if the supporter was the main reason the person went out [synthesis].

5. Handling problems

Most guidance below is [synthesis]; source-based points are tagged.

Person declines the headset or fears it. Offer a demonstration with the headset held, not worn; show the coach's office first; agree a stop signal. Respect refusal and offer other routes (e.g. in-person CBT for psychosis or graded real-world work) [synthesis]. NICE notes VR may not suit everyone [source: NICE HTE15].
Paranoid ideas about the device, coach or supporter. Give plain facts about what is recorded; let the person inspect the kit; keep the supporter visible or audible. If suspicion grows, pause and involve the care team [synthesis]. The trial team report such suspicion is rare in clinical experience [source: Freeman et al. 2023].
Worry about what is happening in the room. This was the most common side effect in the trial [source: Freeman et al. 2023]. Say where you are, stay still, speak now and then, and consider a familiar person present [synthesis].
Cybersickness, headache or dizziness. Stop, remove the headset, sit, offer water and time to reorient; grounding may help [source: NICE HTE15 for removing headset and grounding]. Next time, shorten time in VR, keep movements slow, ensure good headset fit; if symptoms persist across sessions, stop the programme and discuss alternatives [synthesis].
Panic in the headset. In the trial about 7% reported headset-induced panic [source: Freeman et al. 2023]. Offer to pause or remove the headset; afterwards, discuss what the person feared; return to an easier level when ready [synthesis].
Stuck at a level or ratings not shifting. Check for continuing safety behaviours; repeat the situation at the level where learning occurred; link to a real-world goal to raise meaning; review in supervision [synthesis].
Racing ahead without learning. Ask what the person expected and found at each level; suggest repeating a level while dropping one more safety behaviour [synthesis].
Homework not done. Shrink the task; plan it in detail with time and route; consider accompanied practice first; look for practical barriers (money, transport, harassment) [synthesis].
Missed sessions. In the trial three sessions was the minimum adherent dose [source: Freeman et al. 2022]. Offer home sessions (the trial allowed home or clinic [source: Freeman et al. 2022]), flexible times, reminders and peer support [synthesis].
Rupture with the supporter. Name it, ask what went wrong, apologise for your part, and agree changes; offer a different supporter if needed [synthesis].
Overwhelm, relapse signs or increased risk. Stop VR, assess risk, follow the crisis plan, and contact the care coordinator or duty team; resume only after team review [synthesis]. Report adverse events through the service's procedure [source: BMJ Open protocol].
Technical failure (voice recognition, tracking, software error). Pause, reassure that it is the system, not the person; log device issues, since the trial recorded adverse device effects including software malfunction and user error [source: BMJ Open protocol].
Gains fade after the course. Overall group differences were not significant at 26 weeks in the trial [source: Freeman et al. 2022]. Keep up real-world practice, link with peer or community activities, and consider whether further sessions are available; NICE flags relapse and extra sessions as evidence gaps [source: NICE HTE15; synthesis].
Avoidance not mainly anxiety-based. If avoidance is driven by negative symptoms, depression, physical health or real threats, VR alone is unlikely to fit; address the main driver with the team [synthesis].
Language, sensory or learning needs. The trial excluded insufficient English, substantial sensory or balance impairment and clinically significant learning disability [source: Freeman et al. 2022]; for such people, suitability is untested; discuss adaptations with the provider and team or use other approaches [synthesis].
Under-18s. NICE's indication starts at 16; NICE notes there is no NICE guideline on treating agoraphobia in people under 18 [source: NICE HTG701 ch. 2]. For 16–17-year-olds involve parents or carers as appropriate, follow local consent rules for young people, and check headset manufacturer age guidance [synthesis].
Data or privacy worries. Answer from the service's information governance documents; the person can decline sharing data for evidence generation without losing care [synthesis; NICE information for the public advises asking how information is used].
Service out of scope. Mild to moderate avoidance is research-only in NICE's guidance [source: NICE HTG701]; explain that the person may be offered research participation or other treatment instead.

Sources

NICE status statements are dated as of October 2026 per the NICE pages listed above; product and regulatory status may change.

← Previous · IX-092Virtual-reality exposure therapy Next · IX-094 →AVATAR therapy for voices (Leff, Craig)