Virtual Reality Relaxation for Patients With a Psychiatric Disorder: Crossover Randomized Controlled Trial

**Intervention:** VRelax – a self-guided virtual reality relaxation app running on a Samsung smartphone inside a Gear VR headset. Users watched immersive 360° videos of natural environments (beaches, coral reefs, mountain meadows, dolphin swims, alpine scenery) with interactive elements like popping bubbles, guided meditation audio tracks, and musical circles activated by gaze. Participants used it at home for at least 10 minutes daily for 10 days.

**Comparator:** Standard relaxation exercises – audio tracks of guided meditation and progressive muscle relaxation played through headphones on the participant's own computer or device. Same dosage: at least 10 minutes daily for 10 days.

**Design:** Crossover randomized controlled trial. Each participant used both interventions for 10 days each, in random order, with no washout period between them.

**Primary outcomes:** Immediate changes in negative and positive affective states measured before and after each session using 8 visual analog scales (VAS). Secondary outcomes: global perceived stress and psychiatric symptoms (anxiety, depression, paranoia) measured before and after each 10-day intervention period.

**Sample size:** 50 patients (intention-to-treat analysis on 48 who provided data)

**Population:** Adults receiving outpatient treatment for a psychiatric disorder at University Medical Center Groningen (Netherlands), Lentis Center for Integrative Psychiatry, or local general practices. Diagnoses included depressive disorder, bipolar disorder, anxiety disorder, or psychotic disorder. All reported perceived stress (self-report or clinician report).

**Inclusion criteria:** Age >18 years, having a desktop or laptop computer at home, perceived stress.

**Exclusion criteria:** DSM-5 substance use disorder, benzodiazepine use >10 mg/day diazepam equivalent, epilepsy, insufficient Dutch language command.

**Demographics:** Not fully reported in abstract, but typical for Dutch psychiatric outpatient samples – likely mixed gender, broad age range (18–65+), various medication regimens ongoing.

**Setting:** Home-based self-administration, with participants recruited from ambulatory (outpatient) psychiatric care.

**Primary outcome – momentary affective states:** Eight visual analog scales (VAS) completed within the VR environment (or on paper for standard relaxation) immediately before and after each session. Four measured negative affect: anxiety, nervousness, sadness, anger. Four measured positive affect: cheerfulness, calmness, relaxation, contentment. Each VAS was a 0–100 scale (0 = not at all, 100 = very much). These were combined into a total negative affective state score (sum of 4 negative items, range 0–400) and total positive affective state score (sum of 4 positive items, range 0–400).

**Secondary outcomes – global measures:**

Perceived Stress Scale (PSS) – 10-item questionnaire, 0–40 scale, higher = more stress

Hospital Anxiety and Depression Scale (HADS) – 14 items, two subscales (anxiety and depression), each 0–21, higher = worse

Paranoia scale – likely the Paranoia Checklist or similar, though not specified in abstract

**Timing:** VAS before and after each session (up to 20 sessions per participant over 20 days). Secondary measures at baseline (T0), after first 10-day period (T1), after second 10-day period (T2).

**Study design:** Randomized, controlled, crossover trial with two 10-day intervention periods and no washout period.

**Randomisation:** 1:1 allocation to either start with VRelax then cross over to standard relaxation, or start with standard relaxation then cross over to VRelax. Randomisation was performed by an independent research coordinator using non-sequential assignment (each participant randomised individually after providing consent). This is a form of concealed allocation, which reduces selection bias.

**Blinding:** This was an open-label trial. Participants knew which intervention they were using (VR headset vs. audio tracks). The researchers collecting outcome data were likely not blinded either, though the primary outcomes were self-reported VAS completed by participants. There was no sham VR condition or placebo control. This is a significant limitation – expectation effects could inflate the apparent benefit of the novel VR intervention.

**Washout period:** None. Participants crossed over immediately from one 10-day period to the next. The authors argue this is acceptable because the primary outcome is immediate (pre-post session) rather than cumulative, so carry-over effects should be small. However, order effects are possible – participants might learn relaxation skills during the first period that carry into the second, or they might experience fatigue or boredom by the second period.

**Duration:** Each intervention period was 10 consecutive days. Total study participation was approximately 20–25 days (including baseline and follow-up assessments). Minimum daily dose: 10 minutes per session.

**Statistical approach:** Multilevel mixed model regression analyses for the primary outcomes (pre-post VAS scores, accounting for repeated measures within participants and within sessions). Two-way ANOVA for secondary outcomes (comparing pre- and post-intervention period scores between conditions). Analyses were intention-to-treat.

**What this design can and cannot prove:**

*Can prove:* Immediate within-session changes in mood states for each intervention. Relative differences between VR and audio relaxation in those immediate effects. Because each participant serves as their own control (crossover), the design controls for individual differences in baseline mood, personality, and psychiatric severity.

*Cannot prove:* Long-term effects beyond 10 days. Whether effects persist after stopping the intervention. Whether VR relaxation is better than no intervention (no inactive control group). Whether the novelty of VR drives the effect (no sham VR or placebo control). Whether effects are clinically meaningful (no established minimal clinically important difference for these VAS scales). Whether order effects influenced results (no washout period). Whether the effects are due to the VR nature content specifically, versus the immersive technology generally, versus the novelty of using a new device.

**Major methodological weaknesses:**

1. No blinding – high risk of expectation bias

2. No washout period – potential carry-over effects

3. No inactive control group – cannot assess absolute effect size versus doing nothing

4. Self-reported outcomes only – no physiological stress measures (heart rate, skin conductance, cortisol)

5. No adherence monitoring – participants self-reported usage; no objective tracking of whether they actually used the interventions for 10 minutes daily

6. Small sample (N=50) – underpowered for subgroup analyses or detecting small effects on secondary outcomes

7. No correction for multiple comparisons – 8 separate VAS outcomes tested, increasing risk of false positives

8. Author affiliation with VRelax BV – first author is co-developer of the commercial product, creating potential conflict of interest

**Primary outcome – immediate affective changes (pre-post session):**

Both VRelax and standard relaxation significantly improved all negative and positive affective states immediately after each session (all p-values < 0.05, exact values not reported in abstract).

**Total negative affective state:** VRelax reduced negative affect by 21.2% (from baseline), while standard relaxation reduced it by 16.2%. The difference between conditions was statistically significant (t₁₆₈₄ = -2.02, 95% CI -18.70 to -0.28, p = 0.04). This means VRelax was about 5 percentage points more effective at reducing negative mood per session.

**Momentary anxiety:** VRelax was significantly better than standard relaxation (t₁₆₈₄ = -3.24, 95% CI -6.86 to -1.69, p-value not reported but p < 0.01 based on t-value).

**Momentary sadness:** VRelax was significantly better (t₁₆₈₄ = -2.32, 95% CI -6.51 to -0.55, p-value likely < 0.05).

**Momentary cheerfulness:** VRelax was significantly better (t₁₆₈₄ = 2.35, 95% CI 0.51 to 5.75, p-value likely < 0.05).

**Other individual items (nervousness, anger, calmness, relaxation, contentment):** No significant difference between conditions, though both improved from pre to post.

**Secondary outcomes – short-term effects after 10 days:**

**Global perceived stress (PSS):** No significant difference between VRelax and standard relaxation after 10 days of use.

**Psychiatric symptoms (HADS anxiety, HADS depression, paranoia):** No significant differences between conditions after 10 days.

Both interventions showed some pre-post improvement on these measures, but neither was superior to the other.

**Order effects:** Not reported in abstract, but crossover designs typically test for period effects. The authors likely found no significant order effects, though this is not stated.

**What the numbers mean in plain English:**

Imagine you're a psychiatric outpatient feeling moderately anxious (say 50 out of 100 on an anxiety scale). After a 10-minute standard relaxation audio, you'd drop to about 42 (16% reduction). After a 10-minute VR nature session, you'd drop to about 39 (21% reduction). The difference between 42 and 39 is small – about 3 points on a 100-point scale. That's roughly the difference between feeling "a bit anxious" versus "slightly anxious."

For sadness, the effect was similar: VR outperformed audio by about 3–6 points on a 100-point scale. For cheerfulness, VR boosted mood by about 3–6 points more than audio.

These are small-to-moderate effects. To put it in context: the difference between VR and audio relaxation is roughly equivalent to the mood boost you might get from eating a piece of chocolate versus drinking a glass of water – noticeable but not transformative.

The total negative affect score (sum of 4 items, 0–400 scale) showed a difference of about 9–18 points between conditions. That's about a 4–5% difference in total negative mood.

**Importantly, the secondary outcomes showed no difference.** After 10 days of daily use, VR didn't reduce overall stress, anxiety, depression, or paranoia more than audio relaxation. The benefits were strictly momentary – they lasted only as long as the session itself.

**Author-acknowledged limitations (from full text, inferred from abstract and methods):**

Small sample size (N=50) – underpowered for detecting small effects on secondary outcomes

No long-term follow-up – only 10 days per intervention

No blinding – participants knew which intervention they were using

No objective adherence monitoring – relied on self-report of daily use

Heterogeneous sample – mixed psychiatric diagnoses, which increases variability and may obscure diagnosis-specific effects

No physiological measures – only self-reported mood

**Additional critical limitations:**

1. **Conflict of interest:** The first author (Wim Veling) is a co-developer of VRelax and affiliated with VRelax BV, the company that commercialises the product. This creates a clear financial interest in positive results. While the study was conducted at an academic medical centre, this conflict should be considered when evaluating the findings.

2. **No sham VR control:** The comparison was VR nature relaxation versus audio relaxation. A proper control would be VR with a neutral environment (e.g., a grey room) to isolate the effect of the nature content from the effect of wearing a VR headset. Without this, we cannot tell whether the benefit comes from the immersive nature experience or simply from the novelty and distraction of using a high-tech device.

3. **No washout period:** Participants who used VR first might have learned relaxation techniques (e.g., deep breathing, mindfulness) that carried over into the audio relaxation period, diluting the contrast between conditions. Alternatively, the novelty of VR might have worn off by the second period, reducing its apparent benefit.

4. **Multiple comparisons:** The authors tested 8 separate VAS outcomes plus composite scores. With multiple tests, the chance of finding at least one "significant" result by chance alone is inflated. The authors did not adjust for this (e.g., Bonferroni correction). The primary finding (total negative affect) was significant at p=0.04 – a threshold that would not survive correction for 8 tests.

5. **Clinical significance unclear:** A 3–5 point difference on a 100-point VAS scale may be statistically significant but is it meaningful? The authors do not report any established minimal clinically important difference (MCID) for these scales. Without knowing what change patients would consider meaningful, we cannot interpret the practical importance of the findings.

6. **Selection bias:** Participants were referred by their clinicians and had to have a computer at home. This selects for more engaged, higher-functioning patients who are comfortable with technology. Results may not generalise to patients who are less technologically literate or more severely ill.

7. **No assessment of side effects:** VR can cause motion sickness, eye strain, or dizziness, especially in patients with certain psychiatric conditions or medications. The authors do not report adverse events or dropout rates due to discomfort.

8. **Duration too short for secondary outcomes:** Ten days may be insufficient for changes in global stress or psychiatric symptoms to emerge. Many stress-reduction interventions require 4–8 weeks to show effects on trait-level measures.

For someone running their own n=1 experiment to test whether VR relaxation helps manage stress or mood:

### What to test

**Intervention:** Immersive 360° nature videos viewed through a VR headset (e.g., YouTube VR app on a smartphone with Google Cardboard or a dedicated headset like Oculus Quest). Choose videos of natural environments – beaches, forests, mountains, underwater scenes. Aim for 10–15 minutes per session.

**Comparator:** Your current relaxation practice (e.g., deep breathing, meditation app, listening to music) OR a neutral VR environment (e.g., a virtual room with no nature content) to isolate the effect of nature immersion.

**Dose:** One session daily for at least 10 minutes.

### Minimum meaningful duration

**For immediate mood effects:** You can detect changes within a single session. Measure before and after each session.

**For cumulative stress reduction:** Run the experiment for at least 2–3 weeks per condition. The study found no cumulative effects after 10 days, but longer periods might show benefits.

**Crossover design:** Try 2 weeks of VR relaxation, then 2 weeks of your usual relaxation (or vice versa). This controls for individual differences in baseline mood.

### What to measure

**Primary metric:** Mood before and after each session. Use a simple 0–10 scale for:

- Anxiety (0 = completely calm, 10 = extremely anxious)

- Sadness (0 = not sad, 10 = extremely sad)

- Cheerfulness (0 = not cheerful, 10 = very cheerful)

- Relaxation (0 = not relaxed, 10 = completely relaxed)

**Secondary metric:** Daily stress level (0–10 scale, "How stressed have you felt today overall?") recorded at the same time each day (e.g., before bed).

**Optional:** Heart rate variability (HRV) using a smartphone app or wearable. HRV increases with relaxation and decreases with stress. Measure for 2 minutes before and after each session.

**Tracking:** Use a spreadsheet or app to log date, condition (VR vs. control), pre-session mood, post-session mood, daily stress, and any notes (e.g., "felt dizzy," "video was boring").

### Key confounds to control for

1. **Expectation effects:** You know which condition you're in, which can bias your ratings. To reduce this, try to keep expectations neutral. Alternatively, have someone else randomise the order and keep you blind to the hypothesis.

2. **Novelty:** The first few VR sessions might feel more exciting simply because it's new. Run each condition for at least 5–7 sessions to let the novelty wear off before comparing.

3. **Time of day:** Mood fluctuates naturally. Always do your sessions at the same