Virtual Reality-Based Cognitive Behavior Therapy for Major Depressive Disorder: An Alternative to Pharmacotherapy for Reducing Suicidality

Abstract

Purpose

Cognitive behavioral therapy (CBT) has long been recognized as an effective treatment for depression and suicidality. Virtual reality (VR) technology is widely used for cognitive training for conditions such as anxiety disorder and post-traumatic stress disorder, but little research has considered VR-based CBT for depressive symptoms and suicidality. We tested the effectiveness and safety of a VR-based CBT program for depressive disorders.

Materials and Methods

We recruited 57 participants from May 2022 through February 2023 using online advertisements. This multi-center, assessor-blinded, randomized, controlled exploratory trial used two groups: VR treatment group and treat as usual (TAU) group. VR treatment group received a VR mental health training/education program. TAU group received standard pharmacotherapy. Assessments were conducted at baseline, immediately after the 6-week treatment period, and 4 weeks after the end of the treatment period in each group.

Results

Depression scores decreased significantly over time in both VR treatment and TAU groups, with no differences between the two groups. The suicidality score decreased significantly only in VR group. No group differences were found in the remission or response rate for depression, perceived stress, or clinical severity. No adverse events or motion sickness occurred during the VR treatment program.

Conclusion

VR CBT treatment for major depressive disorder has the potential to be equivalent to the gold-standard pharmacotherapy in reducing depressive symptoms, suicidality, and related clinical symptoms, with no difference in improvement found in this study. Thus, VR-based CBT might be an effective alternative to pharmacotherapy for depressive disorders.

Graphical Abstract

INTRODUCTION

Depression is a widespread mental health condition that affects individuals of all ages, genders, and socioeconomic backgrounds on a global scale. According to the World Health Organization, approximately 3.8% of the world’s population grapples with depression, making it the primary contributor to disability worldwide.¹ The prevalence of depression varies regionally, with certain areas reporting higher incidence rates. In South Korea, for instance, depression is estimated to afflict around 5.7% of the population, and that rate coincides with some of the highest suicide rates in the world.²

Currently, treatment for depression centers on medication worldwide. Antidepressant treatment is cost-effective and efficient, but it has limitations. A study of trends in depression treatment found that the use of antidepressants tripled in the United States between 1987 and 1997, and the use of psychotherapy dropped from 12.6 to 8.7 sessions per year,³ indicating a growing dependence on antidepressant medications for the treatment of depression. However, medications are ineffective in 1/3 of patients, and dropout rates are high due to various side effects.⁴

Meanwhile, cognitive behavioral therapy (CBT) has long been recognized as an effective treatment for depression and suicidality.⁴ The use of CBT in depression has evolved since it was first advocated by Aaron Beck.⁵ Mindfulness-based cognitive therapy (MBCT), dialectical behavior therapy (DBT), and mentalization-based treatment (MBT) are CBT offshoots that have been shown to be effective in improving the symptoms of depression.⁶ However, even though those treatments have proved to be useful psychosocial interventions for depression, the provision of CBT by mental health professionals has been limited,⁷ primarily by the time and money required to receive CBT face-to-face from a therapist.⁸,⁹ Several studies have also shown that minorities at different socioeconomic levels are less likely than others to receive this high-quality psychosocial intervention.¹⁰

In this context, digital mental health interventions (DMHIs) have emerged to widen access to mental health treatment through advances in technology.⁹ Computer-based CBT (cCBT) has been developed, well researched, and commercialized for 20 years. In a meta-analysis, cCBT showed a large superiority effect over the control and high satisfaction rates of 83% to 96%.¹¹ A mobile version of DMHI was developed and has been studied for 10 years, and a meta-analysis of those results showed a small effect.¹² However, the effectiveness, satisfaction, and adherence vary widely among DMHIs, suggesting that successful implementation has certain requirements. Qualitative research has shown that people with depression are attracted to and willing to engage with DMHIs that are relatable and interactive.¹³

Virtual reality (VR) is a computer-generated environment of three-dimensional images that people can feel and interact with in a lifelike way. Although VR has been used in many areas of healthcare, mental health has been a particularly prominent area of VR use since the first study of VR exposure therapy was published in 1995.¹⁴ Verisimilitude and ecological feasibility are benefits of VR that make it a good choice for exposure therapy, CBT, and cognitive training.¹⁵ The use of VR exposure therapy has been highly researched for treating anxiety disorders, post-traumatic stress disorder (PTSD), and addiction disorders, for which it showed particularly compelling evidence. Additionally, significant research has examined the use of VR for cognitive training for dementia and social skills training for autism spectrum disorders.¹⁶ On the other hand, VR has rarely been used to treat depression. Although there are studies of programs that use VR games to increase achievement and cognitive training to correct cognitive decline in depression, few studies have considered VR-based CBT that directly targets depressive symptoms and suicidality.¹⁷,¹⁸

Given that studies have shown that adherence to digital interventions in depressed patients is related to interactivity and immersion, we hypothesized that CBT in VR could be used to improve depressive symptoms and suicidality.¹³ Therefore, we developed VR-based software that includes core components of CBT, MBCT, DBT, and MBT for patients with depressive disorders. A 4-week pilot clinical trial of an earlier version of the program showed that it improved symptoms non-inferior to medication and reduced the risk of suicide in patients with depressive disorders.¹⁹ In this study, we expanded the program into a 6-week treatment protocol and tested its effectiveness and safety for treating depressive disorders.

MATERIALS AND METHODS

Participants

We recruited 57 participants from May 2022 through February 2023 using online advertisements. The healthy control and clinical groups were recruited through the same advertisement, and the criteria are described below.

The eligibility criteria were as follows: 1) men and women aged 19 to 50 years who consented to participate in the study; 2) Patient Health Questionnaire-9 (PHQ-9) score of 7 or higher; 3) a score indicating major depressive disorder (F.32) on the Korean version of the Mini-International Neuropsychiatric Interview (MINI) and no psychotic symptoms; and 4) suicidality score of at least 2 and no more than 9 points on the Korean version of the MINI. For the healthy controls, the inclusion criteria were as follows: 1) men and women aged 19 to 50 years who consented to participate in the study; 2) scores below the cutoff score for depression screening on the PHQ-9 (5 or less); 3) no significant psychiatric diagnosis in any module of the Korean version of the MINI; and 4) suicidality score of 1 or less on the Module C of the Korean version of the Mini International Neuropsychiatric Interview (MINI-C).

The exclusion criteria were as follows: 1) difficulty using VR equipment; 2) suicidality score of 10 or more on the Korean version of the MINI; 3) intellectual disability, as shown by a Short Form Intelligence Test (Arithmetic & Common Sense) score of less than 70; 4) increasing suicidality or depression during the study period, as measured by self-report or clinician assessment, to the extent of being unable to participate in the study; 5) psychotic disorders and addiction disorders, as determined during the clinician assessment interview; 6) use of antidepressant, anti-anxiety, or antipsychotic medications in the preceding 6 months; 7) pregnancy or lactation; and 8) inability to read the informed consent form. Other subjects who, in the judgment of the investigator, could not participate in this study were also excluded. The consort flow diagram is presented in Fig. 1.

Fig. 1. Consort flow diagram. PHQ-9, Patient Health Questionnaire-9; VR, virtual reality; TAU, treatment as usual.

Sample size calculation

This study was a comparative clinical study in which VR treatment group was the test group, TAU group was the clinical control group, and healthy individuals were the healthy control group. This study was a non-inferiority trial that showed non-inferiority in the difference between the Korean Self Rating Version of Quick Inventory of Depressive Symptomatology (K-QIDS-SR) scores of VR group and TAU group after 6 weeks of treatment compared to pre-treatment and 4 weeks after the end of treatment. With a one-tailed significance level of 0.025, power of 80%, allocation ratio of 1:1, margin of non-inferiority of 9, and standard deviation of change in K-QIDS-SR of 9.1, the required number of subjects was 16 per group. Considering a dropout rate of 36%, the number of subjects in each arm was calculated to be 25 for a total of 50 subjects. A total of 70 subjects were needed for this study, including 20 healthy controls. However, ultimately not all participants completed the study, and the clinical trial was terminated early, resulting in a final enrollment of 57 participants.

Study design

This multi-center, assessor-blinded, randomized, controlled exploratory trial used two main groups: VR treatment group and treat as usual (TAU) group. The healthy control group of subjects were in the normal ranges in the screening assessments and did not have clinically significant depressive symptoms or suicidality as assessed by a clinician; we used their data to determine if there was a difference in clinical symptoms between VR and TAU groups at baseline. This study was approved by the Institutional Review Board of Yonsei University College of Medicine, Gangnam Severance Hospital (IRB No. 3-2022-0067), and Yongin Severance Hospital (IRB No. 9-2022-0076), for all participating sites. After the recruited subjects provided informed consent, we administered the PHQ-9, Korean version of the MINI, suicidal tendency assessment, basic understanding assessment, and Short Form Intelligence Test to determine their eligibility. After the VR mental health education/training, we conducted a follow-up survey on depression and suicide risk to determine whether they were reduced and verify the effectiveness of the program. Participants with suitable depression and suicidality scores at the screening stage were randomly assigned to VR treatment group and TAU group. VR treatment group received the education/training program, and TAU group received the standard medication treatment. We used different medications for each patient based on the clinician’s judgment, just as in real-world clinical practice, as shown in Supplemental Table 1 (only online). Both groups were assessed immediately after the end of the 6-week VR program and again 4 weeks after the end of VR treatment for follow-up (10 weeks after the baseline assessment). We compared the effects of the VR education and training program with the effects of standard psychopharmacotherapy. The healthy control group received only a baseline assessment. The purpose of the healthy control group was to compare the extent to which the education, training, and medication decreased the Generalized Anxiety Disorder 7-item (GAD-7), Perceived Stress Scale (PSS), Korean version of Quick Inventory of Depressive Symptomatology–Clinician-rated (K-QIDS-C), Hamilton Depression Rating Scale (HAM-D, clinician-rated), MINI-C Suicidality Inventory, and Clinical Global Impression-Severity (CGI-S) scores in VR treatment and TAU groups.

Interventions

VR treatment group received six sessions of VR treatment, delivered once a week for 6 weeks. The program for each session is shown in Table 1. TAU group received standard medical care and medication for 6 weekly sessions.

Table 1. Overview of VR Education and Training Program.

Session	Objective	Related VR program
1 (T1)	Evaluate one’s stress level by learning the concept of stress and social readjustment	Stress management, coping with stress (25 mins)
2 (T2)	After cognitive training on mentalization, apply the tools to real-life situations. Explore the development of mentalization ability through attachment relationships	Attachment and mentalization (15 mins)
3 (T3)	Identify the causes and consequences of emotions and the obstacles that make it difficult to recognize emotion	Emotional control training 1/2 (22 mins)
4 (T4)	Recognize emotions and learn how to change to a positive attitude	Emotional control training 3/4 (24 mins)
5 (T5)	Embrace difficult moments and train methods of relaxation for enduring painful situations. Enhance interpersonal communication training	Enduring a difficult moment 1/2/3, communication training (26 mins)
6 (T6)	Be aware of thoughts and feelings through VR and practice how to focus on the here and now	Mindfulness skills training (20 mins)

VR, virtual reality.

Treatment goals and related VR programs for each of the six treatment sessions.

CHEEU.Forest (MindsAI Co. Ltd., Seoul, Korea) VR software was used to screen and treat mood disorders through assessment, education, and training programs. This investigational medical device was built around the concepts of CBT, MBCT, MBT, and DBT, all evidence-based psychotherapies with proven effectiveness in regulating emotion and impulsivity and reducing suicidality (Fig. 2). The program has six sessions that provide training on stress management and coping with stress with a virtual therapist, mindfulness and attachment training, emotional regulation training, interpersonal skills training, and mindfulness training. National Health Service guidelines recommend a minimum of six sessions of CBT, and the program is structured to deliver six sessions, based on evidence from previous antidepressant studies that 50% of patients who reached remission took more than 6 weeks after starting medication. VR treatment group visited the hospital once a week and wore an Head Mounted Display (HMD; Samsung HMD Odyssey+™; Samsung Co. Ltd., Suwon, Korea) to complete the program. After completion of the VR program, a face-to-face evaluation was conducted with a clinician.

Fig. 2. Screenshot samples of the VR software contents. (A) Screenshot of the psychoeducation section. (B) Screenshot of the mindfulness training section. (C) Screenshot of the coping with tough moments section. (D) Overall content of CHEEU.Forest. VR, virtual reality.

Outcome measures

Primary outcome measure

The K-QIDS-C was used as the primary assessment. The K-QIDS-C is a 16-item clinician-rated test that focuses on the nine symptom domains of the Diagnostic and Statistical Manual of Mental Disorders-Forth Edition (DSM-IV) Criterion A for Major Depressive Disorder. Each symptom domain (depressed mood, concentration, self-blame, suicidal ideation, lack of interest, energy/fatigue, sleep disturbance, appetite/weight change, and psychomotor agitation/tardiness) is scored from 0 to 3, for a total score ranging from 0 to 27, with higher scores indicating more severe depressive symptoms.

Secondary outcome measures

The suicidality scale is derived from MINI-C, version 5.0.0, which is based on the major first-axis psychiatric disorders of the DSM-IV and International Statistical Classification of Disease. It is a brief, structured questionnaire. Module C contains six questions about suicide history and is designed to assess suicidal ideation, planning, and behavior. Each question is scored from 1 to 2 points for a total score range of 1 to 12 points, with higher scores indicating a higher risk of suicide.

The PHQ-9 is a brief, self-report test developed in 1999 by Spitzer, et al.²⁰ to screen for depression and assess its severity. It contains nine items that correspond to the diagnostic criteria for major depressive disorder in the DSM-IV and measures how often the subject has experienced those problems in the past 2 weeks. It is rated on a 4-point scale, with scores ranging from 0 to 27.

The GAD-7 was developed in 2006 by Spitzer, et al.²¹ to screen for GAD. Participants rate the extent to which they have been bothered by problems related to anxiety and worry in the past 2 weeks. It contains seven items, each answered on a 0–3 Likert scale. A total score of 21 is possible, with higher scores indicating more intense symptoms of GAD.

The PSS was developed by Cohen, et al.²² in 1994, and it focuses on how the subject has perceived and interpreted stress during the past month. It is a 10-item questionnaire rated on a 5-point Likert scale, with total scores ranging from 0–40 and higher scores indicating greater perceived stress.

The HAM-D is the most widely used clinician-administered depression assessment scale. The original version contains 17 items pertaining to symptoms of depression experienced during the past week.

The CGI-S is a measure of symptom severity commonly used when treating patients with mental illness and in research. It is scored on a scale of 1–7, with higher scores indicating greater severity. The Clinical Global Impression-Improvement scale is used to assess whether a patient’s mental disorder has improved or worsened since a therapeutic intervention and is also scored from 1–7, with a score of 1 indicating very much better and a score of 7 indicating very much worse.

Remission was defined as a K-QIDS-C score of 5 or less at follow-up. A response was defined as a 50% decrease in the baseline K-QIDS-C score.

To investigate adverse events that occurred during the course of the study, all participants were asked to report any serious adverse events (death, suicide attempt, psychiatric hospitalization, etc.) to the researchers as soon as they became aware of them. In addition, the Simulator Sickness Questionnaire (SSQ) was used before and after each VR session to determine the level of discomfort that could be caused by the VR-based assessment and training program. The SSQ is the most commonly used measure of VR-related simulator sickness. It is a 16-item self-report assessment. Each question can be rated between 0 and 3 points for total score of 0–48 points computed as a simple raw score. Higher scores indicate greater discomfort. The Frequency, Intensity, and Burden of Side Effects Rating (FIBSER) Scale assesses three domains of antidepressant medication side effects: frequency (frequency of side effects from medications taken within the past week for depression), intensity (intensity of side effects from medications taken within the past week for depression), and burden (degree to which antidepressant medication side effects during the past week interfered with day-to-day functions).

Statistical analyses

We analyzed the data using SPSS Statistics software (version 29, IBM Corp., Armonk, NY, USA). To investigate differences in the outcomes between groups, we conducted independent sample t-tests. The t-value represents the difference between the sample mean and the population mean, standardized by the standard deviation of the sample. To investigate within-group differences from baseline to post-intervention, a paired t-test was used. A chi-square analysis was used to understand and test the relationships among categorical variables. Repeated measures analysis of variance (ANOVA) was conducted to examine within-group changes over time, and analysis of covariance was used to test differences between groups while adjusting for baseline differences.

RESULTS

Baseline characteristics

A total of 57 participants took part in the study, with 19 participants in each of the three groups. The demographic and clinical characteristics of the intention-to-treat population are summarized in Table 2. A chi-square test was performed to determine whether the groups differed in gender ratio, and no significant differences were found. There was no significant difference in the proportion of participants with a history of diagnosed depression between VR and TAU groups (47% in VR group, 37% in TAU group, chi-square analysis resulted in a p-value of 0.51). Additionally, there was no significant difference in the average duration of illness between the two groups (2.26 years in VR group, 2.06 years in TAU group, p=0.857). One-way ANOVA was performed to compare the baseline assessment scores among the three groups, which did not differ significantly in age, IQ, or years of education. The GAD-7, PSS, K-QIDS-C, MINI-C, CGI-S, and HAM-D scores differed significantly among the groups, but least significance difference post-hoc tests showed no difference between VR and TAU groups and the lowest scores in the healthy control group. All the scales have higher scores as symptoms worsen, so the lowest scores would indicate the healthiest members of the control group.

Table 2. Baseline Characteristics.

Characteristics	VR (n=19)	TAU (n=19)	Healthy control (n=19)	F	p value
Gender, women/men	14/5	18/1	14/5	-	0.165
Age, mean (SD, range)	28.00 (7.11, 20–42)	27.05 (6.92, 19–44)	28.58 (8.25, 19–47)	0.203	0.817
Mean education years (SD)	14.21 (2.49)	13.79 (2.20)	14.95 (1.93)	1.329	0.273
Depression diagnosis history (Y/N)	9/10	12/7	0/19	-	0.515
Duration of illness (depression)	2.26 (3.21)	2.06 (3.48)	0 (0)	0.033	0.857
IQ	100.32 (11.08)	96.63 (7.69)	102.63 (11.36)	1.679	<0.001*
PHQ-9	18.42 (4.66)	17.47 (5.34)	3.68 (3.71)	60.623	<0.001*
MINI-C	6.16 (1.92)	5.68 (2.11)	0.42 (1.61)	53.828	<0.001*
GAD-7	13.68 (4.67)	12.11 (4.63)	1.83 (1.58)	49.039	<0.001*
PSS	27.89 (5.17)	28.58 (4.94)	13.89 (6.10)	44.314	<0.001*
K-QIDS-C	17.89 (4.12)	18.00 (4.23)	3.74 (5.31)	60.826	<0.001*
HAM-D	15.95 (3.29)	16.11 (2.90)	3.68 (3.89)	84.193	<0.001*
CGI-S	3.74 (0.65)	3.84 (0.69)	1.37 (0.76)	75.379	<0.001*

VR, virtual reality; TAU, treatment as usual; IQ, intelligence quotient; PHQ-9, Patient Health Questionnaire-9; MINI-C, Module C of the Korean version of the Mini International Neuropsychiatric Interview; GAD-7, Generalized Anxiety Disorder-7; PSS, Perceived Stress Scale; K-QIDS-C, Korean version of Quick Inventory of Depressive Symptomatology–Clinician-rated; HAM-D, Hamilton Depression Rating Scale; CGI-S, Clinical Global Impression-Severity; ANOVA, analysis of variance.

Data are expressed as the mean (SD, range) or mean (SD). F and p values are from one-way ANOVA.

^*p<0.001.

The antidepressants used for TAU group were fluoxetine, escitalopram, venlafaxine, vortioxetine, bupropion, mirtazapine, and amitriptyline. Benzodiazepines, such as etizolam, alprazolam, and clonazepam, as well as doxepin, were used to reduce anxiety and insomnia symptoms. The average dosage of the antidepressant was 31.3 mg in fluoxetine-equivalent, which is generally within the commonly recommended range,²³ and there was no difference between the per-protocol and dropout groups.

Primary outcome

Analysis of the primary outcome was performed using a per-protocol approach that included all randomized participants regardless of the intervention received, controlling for baseline differences when appropriate.

The primary efficacy endpoint, K-QIDS-C scores on the third treatment day (T3), the sixth treatment day 6 (T6), and follow-up (FU, 4 weeks post-treatment) from baseline, was compared between VR treatment and TAU groups using a two-sample t-test, and no statistically significant differences were found. In other words, the K-QIDS-C scores did not differ between VR and TAU groups at any point in treatment (Fig. 3). When the change in scores from baseline to T3, T6, and FU were compared by group using a paired t-test, both VR and TAU groups showed a statistically significant decrease in K-QIDS-C scores at T3 [t(16)=4.744, p<0.001 and t(11)=4.675, p<0.001, respectively]. TAU group also showed a significant decrease in K-QIDS-C score at T6 from T3 [t(11)=2.157, p=0.027].

Fig. 3. Comparison of depression scores over time by treatment group. Changes in depression scores for VR group (solid line with circle points) and TAU group (dotted line with square points) by time with repeated measures ANOVA. ^*p<0.05; ^**p<0.001. K-QIDS-C, Korean version of Quick Inventory of Depressive Symptomatology–Clinician-rated; VR, virtual reality; TAU, treatment as usual; T3, third treatment day; T6, sixth treatment day; FU, follow-up 4 weeks after the end of treatment; ANOVA, analysis of variance.

Repeated measures ANOVA was used to analyze differences in K-QIDS-C scores between the groups at T3, T6, and FU. The main effect of time was significant [F(3,25)=33.375, p<0.001]. No significant time and group interaction was found. The difference in K-QIDS-C scores between VR and TAU groups had a mean of 1.59, a lower confidence limit of 5.38, and an upper confidence limit of 2.59, which was less than the non-inferiority margin of 9, confirming that VR group was non-inferior to TAU group (Supplementary Table 2, only online).

Secondary outcomes

A two-sample t-test was conducted to test for differences in MINI-C suicidality scores between VR and TAU groups at baseline, T6, and FU. The scores did not differ significantly between the groups at any time. A paired t-test was used to compare the change in scores from baseline to T6 and FU in each group. The suicidality scores decreased significantly from baseline (6.18±2.01) to FU (2.82±3.49) in VR treatment group [t(16)=4.320, p<0.001] and from T6 (5.12±4.91) to FU in VR treatment group [t(16)=2.358 p=0.031]. The scores did not change significantly in TAU group. A repeated measures ANOVA was performed to determine whether the suicidality scores at baseline, T6, and FU differed between the groups. There was no main effect of time, and the interaction of time and group was not significant either. Repeated measure ANOVAs were conducted to determine whether the PHQ-9, GAD-7, PSS, and HAM-D scores at baseline and each measurement time differed between the groups. The main effect of time was significant for PHQ-9 [F(3,25)=13.994, p<0.001], GAD-7 [F(3,25)=12.533, p<0.001], PSS [F(3,25)=9.909, p<0.001], and HAM-D [F(2,26)=43.798, p<0.001]. However, the interaction of time and group was not significant in any of the measurements. All results are presented in Table 3 and Fig. 4.

Table 3. Repeated Measure ANOVA Results.

Outcome		Group	Baseline	T3			T6		FU		Time			Time × Group
											F	p value	ηp2	F	p value	ηp2
Primary outcomes
	K-QIDS-C	VR	18.18 (3.97)	11.41 (4.74)			9.94 (5.60)		9.47 (6.40)		33.375	<0.001	0.553	0.107	0.746	0.004
		TAU	18.50 (3.50)	11.50 (4.50)			8.67 (4.74)		8.50 (6.13)
Secondary outcomes
	MINI-C	VR	6.18 (2.01)	-			5.12 (4.91)		2.82 (3.49)		0.931	0.407	0.067	0.688	0.414	0.025
		TAU	6.08 (2.15)	-			5.50 (5.20)		6.17 (9.75)
	PHQ-9	VR	17.94 (4.42)	13.24 (5.57)			11.65 (6.83)		10.35 (7.22)		13.994	<0.001	0.627	0.082	0.777	0.003
		TAU	17.33 (5.38)	13.00 (5.19)			10.50 (6.67)		10.25 (6.92)
	GAD-7	VR	13.53 (4.45)	10.24 (4.71)			8.53 (5.57)		7.94 (5.23)		12.533	<0.001	0.601	0.568	0.457	0.021
		TAU	11.83 (4.41)	8.42 (4.94)			6.50 (5.45)		8.67 (5.91)
	PSS	VR	28.12 (4.14)	25.06 (5.54)			23.47 (6.40)		21.29 (7.87)		9.909	<0.001	0.543	0.043	0.838	0.002
		TAU	28.75 (5.07)	25.17 (5.17)			20.08 (9.93)		22.25 (8.24)
	HAM-D	VR	15.94 (3.19)	-			9.24 (4.82)		8.29 (5.51)		43.798	<0.001	0.771	0.006	0.941	0.000
		TAU	16.67 (1.67)	-			8.42 (5.14)		8.08 (4.34)
			Baseline	T1	T2	T3	T4	T5	T6	FU
	CGI-S	VR	3.76 (0.66)	3.24 (0.44)	2.94 (0.75)	2.94 (0.66)	2.65 (0.61)	2.65 (0.97)	2.71 (0.92)	2.35 (1.17)	14.804	<0.001	0.354	0.028	0.868	0.001
		TAU	3.83 (0.58)	3.42 (0.67)	3.08 (0.52)	2.83 (0.39)	2.92 (0.97)	2.58 (0.90)	2.50 (1.09)	2.33 (0.78)
	CGI-I	VR	-	3.00 (0.67)	3.82 (0.64)	3.94 (0.66)	3.65 (0.70)	4.18 (1.33)	3.94 (0.97)	3.59 (1.12)	1.432	0.205	0.050	0.124	0.727	0.005
		TAU	-	3.58 (0.79)	3.08 (0.97)	3.50 (0.91)	4.08 (1.08)	3.58 (1.51)	3.92 (1.38)	4.08 (1.51)

VR, virtual reality; TAU, treatment as usual; T3, third treatment day; T6, sixth treatment day; FU, follow-up 4 weeks after end of treatment; K-QIDS-C, Korean version of Quick Inventory of Depressive Symptomatology–Clinician-rated; PHQ-9, Patient Health Questionnaire-9; MINI-C, Module C of the Korean version of the Mini International Neuropsychiatric Interview; GAD-7, Generalized Anxiety Disorder-7; PSS, Perceived Stress Scale; HAM-D, Hamilton Depression Rating Scale; CGI-S; Clinical Global Impression-Severity; CGI-I, Clinical Global Impression-Improvement; ANOVA, analysis of variance.

Data are expressed as the mean (SD). F, p, and ηp² values are from repeated measures ANOVA by time and by time and group.

Fig. 4. Changes in the secondary outcome measures for VR group (solid lines with circle points) and TAU group (dotted lines with square points) by time with repeated measures ANOVA. (A) MINI-C, Module C of the Korean version of the Mini International Neuropsychiatric Interview. (B) PHQ-9, Patient Health Questionnaire-9. (C) GAD-7, Generalized Anxiety Disorder 7-item. (D) PSS, Perceived Stress Scale. (E) HAM-D, Hamilton Depression Rating Scale. (F) CGI-S, Clinical Global Impression-Severity. (G) CGI-I, Clinical Global Impression-Improvement. ^*p<0.05; ^**p<0.001. VR, virtual reality; TAU, treatment as usual; ANOVA, analysis of variance.

The remission rate and response rate results are shown in Fig. 5. The remission rates were 29.4% in VR group and 33.3% in TAU group. The response rates were 41.2% in VR group and 66.7% in TAU group. The chi-square test showed no significant difference in the remission and response rates between the groups.

Fig. 5. Remission and response rates. Between-group differences in the remission (A) and response (B) rates. Remission was defined as a K-QIDS-C score of 5 or less at follow-up. A response was defined as a 50% decrease in K-QIDS-C score from baseline. K-QIDS-C, Korean version of Quick Inventory of Depressive Symptomatology–Clinician-rated; VR, virtual reality; TAU, treatment as usual.

Safety results

The safety analysis was conducted for the 19 patients who received the investigational medical device at least once after providing written informed consent. None of the 19 subjects experienced more than one adverse event following the treatment. The SSQ scores provided before and after the first and sixth treatments (T1 and T6, respectively) are shown in Fig. 6. Since the post-treatment values did not meet normality, the non-parametric Wilcoxon’s signed rank test was performed to test between time points. The T1 pre-treatment and post-treatment scores did not differ with statistical significance. The difference between pre- and post-treatment scores at T6 was statistically different, with a p-value of 0.02. In the post-treatment FIBSER scores for T1–T6, one FIBSER score of 3 or higher was reported at T2, T4, and T6.

Fig. 6. Safety evaluation pre- and post-treatment. Comparison of pre- and post-treatment SSQ scores at T1 and T6 with paired t-test. ^*p<0.05. SSQ, Simulator Sickness Questionnaire; VR, virtual reality; T1, first treatment day; T6, sixth treatment day.

DISCUSSION

Our aim in this study was to evaluate the feasibility of our VR CBT program for depression in terms of its effectiveness and safety, and to compare its potential beneficial effects with those of TAU (medication). The primary efficacy endpoint, the K-QIDS-C score, decreased from baseline to T3, T6, and FU in both VR and TAU groups, and it did not differ between the groups at any treatment time. The secondary efficacy endpoint, suicidality score, decreased significantly post-treatment, but only in VR group. The perceived stress, anxiety, and depression scores also tended to decrease overall post-treatment, with no difference between VR and TAU groups. The post-treatment remission rates and response rates shown by the depression scores also did not differ between the groups. In terms of safety, the pre- to post-treatment simulator sickness scores were significantly lower at session 6 than at session 1. Overall, this study demonstrated that VR CBT treatment for major depressive disorder has the potential to be equivalent to the gold standard of pharmacotherapy in reducing depressive symptoms, suicidality, and clinical severity, as well as reducing secondary symptoms such as anxiety and perceived stress.

Both medication and CBT have demonstrated efficacy in treating depressive disorders, albeit through different mechanisms. Antidepressant medications, such as selective serotonin reuptake inhibitors and serotonin-norepinephrine reuptake inhibitors, work by altering neurotransmitter levels in the brain, specifically serotonin, norepinephrine, or dopamine.²⁴,²⁵ These medications have shown effectiveness in managing symptoms for many individuals, often leading to improvements in mood, sleep, and overall functioning. However, they can take several weeks to show significant effects, and finding the right medication and dosage can be a trial-and-error process.²⁴,²⁵ Moreover, some patients experience side effects such as nausea, weight changes, or sexual dysfunction, which can affect treatment adherence.²⁶ On the other hand, CBT has consistently demonstrated effectiveness in treating depression by targeting negative thought patterns and behaviors.²⁷ CBT helps individuals recognize and challenge distorted thinking, replacing it with more realistic and adaptive thought processes.⁴,²⁷ By teaching coping skills and problem-solving techniques, CBT equips individuals with the tools they need to manage depressive symptoms both during and after therapy. Studies have shown that CBT not only reduces symptoms but also helps prevent relapse, making it a valuable long-lasting therapeutic option. Moreover, CBT tends to have enduring effects even after therapy concludes, as individuals continue to apply the skills they have learned in their daily lives.²⁸

A meta-study²⁹ comparing the efficacy of medication and CBT for depression did not confirm an advantage between them, though some researchers suggest that in the short term, medication might show a faster reduction in symptoms.³⁰ However, in the long term, CBT can be just as effective as medication, especially in preventing relapse.³¹

In-person CBT traditionally involves face-to-face sessions with a therapist who offers personalized interaction, immediate feedback, and a deep emotional connection. This format allows for a tailored approach in which therapists can adapt techniques based on individual responses and foster a strong therapeutic alliance.³² Conversely, digital CBT, often delivered through online platforms or apps, provides flexibility and convenience, enabling individuals to access therapy remotely at their own pace and time.³³,³⁴ Although it lacks the immediacy of an in-person interaction, digital CBT offers a structured format, self-guided modules, and interactive tools that empower individuals to actively engage in their treatment.³⁴ Studies suggest that both in-person and digital CBT can be effective in reducing the symptoms of various mental health conditions, with digital interventions demonstrating promise in increasing access to care and addressing barriers such as geographical limitations and scheduling constraints.⁹,³⁴ However, challenges remain in ensuring the quality and depth of the therapeutic interaction in digital CBT, as well as addressing the needs of individuals who might require more intensive or personalized interventions. The CBT in this study used VR technology with a virtual therapist. Therapist intervention has been shown to be useful in making DMHIs more engaging and effective.³³ A therapist in VR can create the feeling of real human involvement, which could make VR CBT more effective and increase treatment adherence. VR is often used as an exposure therapy for addiction, anxiety, and PTSD in mental health.¹⁶ VR has high ecological validity and allows participants to interact and immerse themselves in the environment. It is said to be useful for awareness, decision-making, mindfulness, and real-life applications. There have been various attempts to use VR in psychiatry. Since the first attempts at exposure therapy for phobias, various versions of VR exposure therapy for panic disorder, social anxiety disorder, and specific phobias have been tested and proved effective.¹⁴,³⁵ In addition, cue-exposure therapy for addiction and eating disorders through VR has recently been tested and shown to be helpful, and virtual therapeutic immersion environments for treating attention-deficit hyperactivity disorder and autism are also beginning to be tested.¹⁶ In depressive disorders, however, VR interventions are limited; and to our knowledge, no previous randomized controlled trials have been conducted on VR programs for depression that provide CBT similar to that used in clinical practice.

Attempts have been made to provide a sense of accomplishment for people with depressive disorders by offering rewards through VR games, and VR cognitive training has been tested to prevent depression-related cognitive decline.¹⁷,¹⁸ One VR program attempted to reduce self-criticism by having a participant comfort a person in distress and then change their perspective to listen to the comfort they had just offered.³⁶ Various VR mindfulness programs have been developed to increase positive emotions.³⁷ CBT for depressive disorders has also been translated into VR in the following ways. Immersive psychoeducation, physical activity, and social gatherings can be enhanced through VR.³⁸ Immersive cognitive restructuring can be experienced by providing example situations with interactive visualization, and positive imagery training by re-scripting VR imagery could also be considered.³⁹ Although no clinical trials of those programs have been conducted in patients with depressive disorders, the demonstrated effectiveness of our VR program could encourage their therapeutic use.

In both this study and the previous 4-week pilot trial, suicide risk was reduced only in VR treatment group. A systematic review found that CBT and DBT interventions had a modest benefit in reducing the suicide risk, compared with TAU.⁴⁰ The risk of suicide can be explained by the stress-diathesis model using a combination of heightened perception of emotional distress, distorted social cognition, impaired problem-solving, and hopelessness.⁴¹ As mentioned above, CBT addresses negative thoughts and teaches coping skills and problem-solving techniques; and DBT, MBT, and MBCT can be used to train emotional regulation, impulse control, effective interpersonal skills, and distress tolerance. All of these techniques are included in our VR treatment program, and performing those treatments in an immersive environment could be a mediator of suicide risk reduction. A previous study showed that mindfulness therapy through VR is more effective than in the usual environment, as it offers a more immersive experience.⁴²

The major strength of our study is that it is the first to combine CBT with immersive VR technology and compare its effectiveness with medication in patients with depression. Notably, existing VR CBT has mostly been used to treat anxiety disorders, with only one study associated with depression.⁴³ In addition, to the best of our knowledge, our findings are the first to show that only six sessions of VR treatment with CBT content based on mindfulness and DBT is non-inferior to medication. This suggests that VR-based CBT could be an effective alternative to pharmacotherapy for depressive disorders.

Nonetheless, our study still has many limitations. First, our study is an exploratory clinical study to examine the effectiveness and safety of a VR-based mental health education and training program for treating depressive disorders, and the number of subjects is relatively small. This means that the statistical power of our results is small. It is necessary to conduct a randomized controlled trial with a larger number of subjects in the future to demonstrate the effectiveness of the treatment content. Second, although the advantage of digital CBT is its accessibility, our study was conducted in a hospital setting with in-person visits. One meta-study has shown that the moderators of CBT effectiveness are the number of sessions and duration of intervention.³⁴ We suggest a follow-up study to maximize the benefits of digital CBT by increasing the number of sessions in a home-care setting. Next, according to the sample size calculation, the study required a total of 70 subjects, including 20 healthy controls. However, in the end, not all participants were able to complete the study, and the trial was terminated early, resulting in a final enrollment of 57 participants, which reduces the statistical power. Lastly, in this study, VR group also had a brief interview with a clinician after treatment for symptom and side effect evaluation, and we acknowledge that this face-to-face interaction itself may have had an impact on depression. In the future, we expect to conduct a study design that minimizes clinician intervention through a home-based VR therapy application study to measure the pure effect of VR therapy.

In conclusion, the integration of VR technology enhances CBT by offering immersive experiences that potentially increase engagement and treatment adherence. This study’s innovative approach highlights the potential of VR-based interventions as a valuable therapeutic option for depressive disorders, suggesting that VR CBT could serve as an effective alternative or adjunct to traditional pharmacotherapy. Further research and development in this area are warranted to fully explore the potential of VR technology in mental health treatment.

SUPPLEMENTARY MATERIALS

Supplementary Table 1

The Average Medication Dosage for Each Group within TAU Group

Supplementary Table 2

ANCOVA Results