Influence of Virtual Reality on Negative Symptoms and Quality of Life of Patients With Schizophrenia: A Meta-Analysis: Influence de la réalité virtuelle sur les symptômes négatifs et la qualité de vie des patients atteints de schizophrénie : une méta-analyse

Abstract

Background

Schizophrenia brings about diverse challenges to patients, their families and society. It is up to the healthcare system to effectively resolve these concerns and benefit all involved parties. With the emerging development and adaptation of virtual reality (VR), this may offer a new direction and potential for treating people with schizophrenia. Our goal was to employ meta-analysis to evaluate the influence of VR on the clinical outcomes and quality of life in people with schizophrenia.

Methods

We performed an extensive screening of randomized controlled trials (RCTs) examining the effect of VR on the clinical outcomes in people with schizophrenia. Our search included the scientific databases PubMed, Embase, Web of Science and the Cochrane Library, and included RCTs from the date of database establishment till 1 June 2023 and we followed a strict study of inclusion and exclusion criteria. The meta-analysis was conducted in RevMan 5.4.

Results

We selected 963 patients from 10 RCTs. Relative to other forms of interventions, VR therapy considerably alleviated overall clinical (SMD = −4.33, 95% CI = [−6.92, −1.74], P = 0.001) and negative symptomology (SMD = −1.38, 95% CI = [−2.46, −0.30], P = 0.01) among in people with schizophrenia. In contrast, no significant improvements were observed in positive symptoms or quality of life among these patients. Further subgroup analyses of the results indicated that there were differences in the improvement of negative symptoms among patients across the different interventions (P = 0.01).

Conclusions

Based on our meta-analysis, VR-based treatment regimen significantly improves overall and negative symptoms in people with schizophrenia. Further exploration is warranted to elucidate the influence of VR on patient positive symptoms and quality of life.

Introduction

Schizophrenia is a debilitating mental disorder affecting a significant population worldwide.^1,2 It is typically characterized by gross abnormalities in thinking, emotion and behaviour. Schizophrenia symptoms can be classified as follows: positive, negative and cognitive symptoms. ³ Currently, standard care for people with schizophrenia includes medication, cognitive behavioural therapy, cognitive remediation and social skills training. Among them, medication is most recommended due to its efficacy in managing positive symptoms such as delusions, unusual thoughts and hallucinations. ⁴ Unfortunately, among patients with severe negative symptoms, for example, social withdrawal and lack of pleasure, there is reduced drug sensitivity. ⁴ Emerging evidences revealed that people with schizophrenia typically have vastly reduced employment rate, which is intricately linked to the patients’ residual psychiatric symptoms and other factors. ⁵ This significantly impacts the patient, relatives and community at large.^2,6 Cognitive behavioural therapy helps reduce positive and negative symptoms, cognitive remediation helps social-cognitive impairments,^7,8 social skills training is usually conjunction with antipsychotic drugs to improve the clinical symptoms and cognitive abilities of patients with schizophrenia. ⁹ Virtual reality (VR) is a newly developed technology that produces real-world simulations in computer-generated interactive environments. ¹⁰ Thus, it creates a virtual world that mimics reality and facilitates patient interaction in a controlled environment. ¹¹ With recent advancements in VR technology, its application has progressed to multiple areas, such as, entertainment, education, business and so on. The presence of negative symptoms and other contributing factors generally results in patients diagnosed with schizophrenia exhibiting a diminished inclination to engage in social activities. Furthermore, a considerable number of patients are confined to hospitals for extended periods due to factors such as family circumstances, which can have a detrimental impact on their overall functioning and quality of life (QoL). Consequently, the utilization of VR for the rehabilitation of patients diagnosed with schizophrenia is becoming increasingly prevalent within the medical field. ¹² Within controlled and immersive environments, VR simulations facilitate structured social interactions with virtual characters, thereby enhancing social functioning.^11,13 Furthermore, VR environments can be tailored to incorporate goal-oriented tasks and reward-based scenarios that stimulate motivation, enhance sensitivity to positive reinforcement and promote experiential pleasure. ¹⁴ VR-based interventions have been shown to enhance patients’ capacity for pleasure, purposeful activity and social interaction, thereby contributing to their overall functional recovery and QoL. ¹⁵

QoL is a multidimensional construct encompassing physical, psychological, social and other functional domains. ¹⁶ The term is most commonly defined as “an individual's perception of their position in life, in the context of the culture and value systems in which they live, and in relation to their goals, expectations, standards and concerns.”^{a, 17} The Quality of Life Enjoyment and Satisfaction Questionnaire-Short Form (QLES-Q-SF) primarily focuses on subjective enjoyment and satisfaction across general life domains. ¹⁸ Recovering Quality of Life (ReQoL) emphasizes recovery-oriented aspects of QoL, such as autonomy, hope and social connectedness, thereby capturing experiential and value-based dimensions of mental health recovery. ¹⁹ Manchester Short Assessment of Quality of Life (MANSA) is a widely utilized tool for assessing both objective and subjective QoL across a range of life domains, including financial status, interpersonal relationships and living conditions. ²⁰ Informed by previous meta-analyses,^21,22 we included all three instruments to comprehensively capture both general life satisfaction and recovery-specific experiences among individuals with schizophrenia.

VR can be fully immersive, semi-immersive or non-immersive. Fully immersive VR completely immerses the patient into the virtual world, with minimal interference from the real world. This allows the patient to achieve significant involvement and engagement with the virtual environment.^23,24 Alternately, non-immersive or semi-immersive VR is easier to operate, however, it permits more interference from the real world, which may influence patient engagement.

In recent years, VR is increasingly used for neuropsychiatric disorder management. ¹² Some examples include the VR-mediated rehabilitation of mild cognitive impairment,^25,26 cognition of dementia patients, ²⁷ cognitive functioning and life skills following stroke, ²⁸ cognitive aspects of Attention Deficit Hyperactivity Disorder individuals, ²⁹ depression, ³⁰ anxiety and related disorders,^31,32 as well as mental disorders. ³³ In addition, there are multiple reports on the enhancement of cognitive abilities among patients with psychiatric disorders. ¹² Previous meta-analyses and systematic reviews have primarily focused on the effects of VR interventions on symptom improvement in various psychiatric disorders, ^34-36 or on the effects of VR on clinical symptoms and cognitive function in people with schizophrenia.^37,38 Some studies have also compared VR interventions with other treatment modalities. ³⁹ In contrast, the studies included in the present analysis are all randomized controlled trials (RCTs), which comprehensively assess the overall impact of VR on clinical symptoms and QoL, emphasizing its impact on negative symptoms—an area that has often been underrepresented in previous research. This focus is particularly important, as the alleviation of clinical symptoms is closely linked to improvements in patient prognosis, QoL and functional abilities. Furthermore, we conducted subgroup analyses to evaluate whether different intervention approaches yielded varying effects. This provides new evidence supporting the use of VR-based treatments to enhance clinical outcomes for individuals with schizophrenia. Therefore, in this report, we conducted a meta-analysis to explore the quality of relevant RCTs investigating VR significance in schizophrenia management. We also evaluated the overall efficacy of VR usage in treating schizophrenia, with particular emphasis on the alleviation of clinical symptoms and enhancement of patient QoL. Our goals for this study were as follows: (a) to elucidate the efficacy of VR-based therapies in improving clinical symptoms and QoL of people with schizophrenia, (b) to explore feasibility of VR therapy and identify differences in the impact of various forms of VR interventions on patient prognoses.

Methods

Screening of Relevant RCT

The study protocol strictly followed PRISMA guidelines. ⁴⁰ We conducted a systematic screening of relevant articles in scientific databases, including PubMed, EMBASE, Web of Science and Cochrane Library databases from the date of establishment till 1 June 2023. The search terminologies were as follows: “schizophrenia”, “virtual reality” (please refer to Supplemental Material for more details).

Two investigators independently selected eligible RCTs by reviewing titles and abstracts. Any disagreements were resolved via mutual discussion. The EndNote20 software was employed for duplicate removal. Duplicates were only removed from analysis if both investigators agree to its elimination. If at least one investigator selected a study or if the study title and abstract did not include appropriate details to achieve an immediate decision, then both investigators reviewed the entire manuscript to determine eligibility in meta-analysis. Once selection was complete, data extraction was carried out based on a pre-designed table. The extracted data included first author, year of publication, country, sample size, age, treatment modality, frequency of intervention (and control measures), outcome indicator measures and types of intervention (and control measures). Both investigators independently retrieved data, and any discrepancy was resolved via mutual discussion.

Selection Criteria

The following RCTs were included in meta-analysis: those that (a) examined individuals meeting the diagnostic criteria for schizophrenia; (b) explored VR-based intervention with control receiving VR alone without intervention, conventional treatment and conventional cognitive training intervention; (c) explored outcome indicators, such as, overall schizophrenia symptoms (PANSS), positive symptoms (PANSS-P, SAPS), negative symptoms (PANSS-N, SANS), and QoL (Q-LES-Q-SF, ReQoL, MANSA); (d) employed a randomized controlled design, and examined subjects from all age groups. Among the articles excluded from meta-analysis were non-English language articles, those involving subjects with other mental illnesses, those lacking the specified outcome indicators and non-RCTs. This work was prospectively registered with PROSPERO (CRD42024554618).

Assessing Bias Risk

RCT bias risk was evaluated using tools recommended by the Cochrane Handbook version 5.1.0, ⁴¹ and focused on seven aspects as follows: random sequence generation, allocation concealment, participants and personnel blinding, outcome assessment blinding, incomplete outcome data, selective reporting and other biases. Individual items were classified as either “low,” “high,” or “unclear” bias risk. The risk of bias was assessed independently by two authors, both with experience in systematic reviews and meta-analysis. An RCT was classified as having “low” bias risk when all 7 aforementioned items were categorized as “low” bias risk. If ≥1 items were categorized as “high” bias risk, then the RCT was classified as having “high” bias risk. RCTs that did not clearly fit into either “low” or “high” category were assigned as having an “unclear” bias risk. Disagreements between the two reviewers were resolved by discussion and, if necessary, a third senior author was consulted to reach consensus.

Data Collection and Analyses

Meta-analysis was conducted using RevMan version 5.4, with a significance threshold of P < 0.05. Result heterogeneity was assessed using both Dixon's Q test and the I-squared (I²) statistical test, with I² values of 25%, 50% and 75% representing low, moderate and high heterogeneity, respectively. A random effects model was employed for all analyses to account for potential heterogeneity across studies. Standardized mean differences (SMDs) with 95% confidence intervals (CIs) were reported to ensure comparability of results across studies with varying scales. Sensitivity analyses were performed to evaluate the robustness of the pooled estimates by excluding studies with high risk of bias or studies with extreme effect sizes.

Results

Article Screening and Selection

Figure 1 summarizes the PRISMA flowchart. An initial literature screening of 4 scientific databases, along with duplicates removal, provided 783 eligible articles. Following title and abstract review, only 45 articles were selected for full text review, among which, 15 were found to have irrelevant content, e.g., intervention was not VR-related and missing outcome data was found in 16 articles, 4 articles were non-RCT articles. In total, 35 articles were excluded at this step. Ultimately, we selected 10 articles that fit our strict inclusion and exclusion criteria.

Figure 1.

PRISMA flow diagram depicting study selection.

Characteristics of Selected Articles

Table 1 details the major profiles of selected investigations. In all, 10 studies were deemed eligible for our meta-analysis, among them, 2 studies were from Hungary, 2 from UK, 2 from China, 2 from Canada, and 2 from the Netherlands.

Table 1.

The Essential Characteristics of all Included Studies.

Study(Years/ Country/Type of study)	Sample size(E/C)	Duration of illness (years) :M(SD)	Age:M(SD)	Intervention	Control	Types of VR measures	Outcome	Length of sessions	Treatment modalities/methods and therapist/trainer facilitation	Interventions
Roos M C A Pot-Kolder (2018/Netherlands/RCT)	58/58	E:13.3 (10.6) C:14.9 (9.5)	E:36.5 (10) C:39.5 (10)	VRC	UC	Immersive	MANSA	16 sessions over 8–12weeks (60minutes/session)	Virtual-Reality Based Cognitive Behaviour Therapy: Four virtual social environments (a street, bus, café and supermarket) were created with Vizard software. Participants used head-mounted display. Conducted by psychologists with CBT training.	Individual
Laura Dellazizzo (2021/Canada/RCT)	28/34	E:18.0 ± 10.6 C:14.6 ± 10.2	E:43.6 ± 12.0 C:41.4 ± 13.4	AT	CBT	Immersive	PANSS、Q-LES-Q-SF	9weeks(60minutes/ session/week)	Patients were immersed in VR through the Samsung Gear VR head-mounted display. Patients sat in an adjacent separate room from the therapist, who would converse with patients either through the voice of the avatar or as themselves. The immersive virtual environment consisted of an avatar seen from a first-person perspective standing in a dark room. An inventory of facial expressions was integrated into the platform to use at the therapist's discretion to enable the avatar to express emotions that patients would easily recognize such as joy, sadness, anger and fear based on the Facial Action Coding System.	Individual
Daniel Freeman (2022/UK/RCT)	174/172	-	E:36.6 (12.8) C:37.8 (12.2)	VRC	UC	Immersive	ReQoL	6weeks(30minutes/ session/week)	HTC Vive Pro headset, Dell G5 155590 laptop. A virtual coach, within the VR environments, guided the participant through the therapy. At the beginning of the first session, the virtual coach explained the rationale behind the therapy, and the participant selected one of six VR social scenarios (café, general practice waiting room, pub, bus, opening the front door of the home onto the street and small local shop). Each scenario comprised five levels of difficulty (based on the number and proximity of people in the social situation and the degree of social interaction) and participants worked their way through each level. Getting closer to other people and making eye contact are encouraged by the coach in many of the scenarios, and sometimes participants were the centre of attention within a situation (e.g., being asked to ring the bell at the front of the bus). Therapeutic game-type tasks are included within several levels. These tasks are designed to help the participant let go of defence behaviours and thereby make new learning.	Individual
Olivier Percie du Sert(2018/Canada/ partial cross-over RCT)	15/7	E:17.8 (4.7) C:-	E:42.9 (12.4) C:-	AT	UC	Immersive	PANSS	7 weekly sessions	Immersive virtual reality in which participants created an avatar best resembling the most distressing entity believed to be the source of the malevolent voice and engaged in a dialogue with it. Idiosyncratic avatars were created using Unity 3D game engine and voice was simulated with a voice transformer. Participants were immersed through Samsung GearVR head mounted display and smartphone. Facilitated by a psychiatrist. Participants engaged with the therapist via the personalized avatars.	-
Tom KJ Craig (2018/UK/RCT)	75/75	E:20.5 (10.1) C:19.8(10.8)	E:42.5(10.1) C:42.9(11.2)	AT	SC	Nonimmersive	SAPS、SANS、 MANSA	6weeks(50minutes/ session/week)	Participants sat in one room facing their avatar on a computer monitor. The therapist sat in a second room with a control panel that allowed them to speak in his or her own voice, or as the avatar. A video link allowed the therapist to see and hear the participant's responses, enabling them to adjust therapeutic interventions and modify the avatar interaction according to the unfolding dialogue. The progress of sessions was determined by a discussion in each session concerning any change in severity, malevolence, or frequency of the voices. All sessions were audio recorded and a copy of the avatar dialogue was provided on an MP3 player to the participant with instructions to listen to the recording at home.	Individual
Edit Vass (2021/Hungary/RCT)	9/8	E:20.8(12.65) C:21.5(7.19)	E:38.6(13.49) C:48.8(8.87)	VRC	P-VR	Immersive	PANSS、 LQOLP	6weeks(50minutes/ session/week)	Simulated social interactions with an avatar in immersive VR environments, with virtual conversations dialogue to train Theory of Mind. Head-mounted display (HMD), a Samsung S7 smartphone and a Samsung Simple Controller were used. Conducted by a psychotherapist.	Individual
Edit Vass (2022/Hungary/RCT)	21/21	E:17.31(11.19) C:19.76(10.29)	E:36.71(11.73) C:42.47(8.74)	VRC	P-VR	Nonimmersive	PANSS、 LQOLP	9weeks(60minutes/ session/week)	Samsung Gear Head Mounted Displays were used with a Samsung S7 or S8 smartphone and a Samsung Simple Controller. A Theory of mind intervention through immersive virtual reality led to diverse effects in comparison to passive VR environment exposure in patients suffering from schizophrenia.	Individual
Shangda Li (2022/China/RCT)	30/32	E:195.10 ± 107.86 C:249.94 ± 97.55（months）	-	VRC	UC	Immersive	PANSS	2weeks(5 sessions/week)	A virtual reality (VR) supermarket scenario was developed. Unity 5.3.5f1 and Visual Studio 2015 (Microsoft) were utilized for the design and construction of the VR program. Patients were instructed to don the helmet to undertake various shopping tasks with distinct lists. The shopping tasks comprised tasks A and B, with each task comprising four levels.	-
Nana Liang (2022/China/RCT)	32/33	E:4.19 ± 3.27 C:4.61 ± 3.22	E:25.3 ± 5.3 C:26.5 ± 6.8	AT	CBT	Nonimmersive	PANSS、Q-LES-Q-SF	7–9 weeks(60minutes/ session/week)	Each session included three parts: Predialogue: the therapist and the patient would review the preceding week, discuss the objective of this session, and agree the focus of the dialogue between the patient and the avatar. Trialogue: The therapist and the patient sat in separate rooms but engaged in direct dialogue for approximately 10–15 min. Postdialogue: The rest of the time was occupied with the patient's feedback following the trialogue. The dialogue between the patient and the avatar during each session was recorded and provided to the patient on an MP3 (portable digital audio) player for continued use at home.	Individual
S. A. Nijman (2022/Netherlands/RCT)	41/40	E:11.0（8.8） C:14.3（12.0）	E:35.9（10.4） C:39.7（12.4）	VRC	P-VR	Immersive	PANSS	8weeks(45–60-min/twice sessions/week)	The VR environments were created by CleVR BV and were displayed using an Oculus Rift head-mounted-display. Therapists controlled the environment and avatars through a tablet interface. Participants practiced individually with social-cognitive strategies throughout the intervention.	Individual

Note. E = experimental group; C = control group; M = mean; SD = standard deviation; a = age in years (range); VRC = VR CBT based intervention; AT = avatar therapy-based intervention; UC = usual care; CBT = cognitive behavioural therapy; SC = supportive counselling; P-VR = passive-VR condition, patients could freely explore the virtual destinations without any intervention; PANSS = Positive and Negative Syndrome Scale; SANS = Scale for Assessment of Negative Symptoms; SAPS = Scale for Assessment of Positive Symptoms; MANSA = Manchester Short Assessment of Quality of Life; Q-LES-Q-SF = the short form version of the quality of life enjoyment and satis-faction questionnaire; ReQoL = Recovering Quality of Life total; LQOLP = Lancashire Quality of Life Profile; - = Not mention.

Sample Characteristics

The selected studies were published between the years 2018–2023, and included 963 participants. The largest sample was 296 and the smallest sample was 17. Table 1 details all sample demographics.

Intervention Characteristics

We referred to a recent systematic review and categorized the VR interventions included in the studies into two main types based on the modalities used: cognitive behavioural therapy-based interventions (VRC), avatar therapy-based interventions (AT). ³⁹ The experimental groups utilizing VR-based cognitive and behavioural therapy were classified into the VRC group, while those incorporating therapist-guided role-playing based on VR to simulate auditory and visual hallucinations for practising different responses were categorized into the AT group. A total of six studies were classified under the VRC group, while the remaining four studies were assigned to the AT group. The intervention duration ranged between 2 and 12 weeks. The training frequency was between 1 time/week-5 times/week, with majority adopting 1 time/week, ^42-47 and 1 study administering 5 times/week. ⁴⁸

Outcome Measures

Table 1 lists all reported outcome measures. All overall clinical symptoms were assessed using the PANSS scale. Moreover, 2 studies evaluated patient QoL using the Q-LES-Q-SF scale,^42,49 the remaining 2 adopted the MANSA^45,50, and 1 study employed the ReQoL total. ⁴³

Bias Risk Evaluation

Figure 2 presents our bias risk evaluation results. Using random sequence generation analysis, we identified 5 RCTs as having “low” bias risk. In terms of participant and researcher blinding, 8 RCTs were regarded as being “unclear.” In addition, 7 RCTs showed “low” bias risk for evaluator blinding. In terms of incomplete outcome data, 9 RCTs exhibited “low” bias risk. In case of selective reporting, all trials exhibited “low” bias risk. Lastly, in case of other biases, 6 RCTs showed “low” bias risk whereas 1 RCT revealed “high” bias risk.

Figure 2.

Bias risk assessment for included studies.

Outcomes

The Influence of VR on Overall Symptoms in People With Schizophrenia

In total, 6 RCTs examined the impact of VR on overall clinical symptoms in people with schizophrenia (PANSS). Measurements were completed both before and after intervention, and included 263 subjects. Our random effect model-based meta-analysis revealed that VR substantially diminished patient total PANSS scores, relative to controls (SMD = −4.33, 95% CI = [−6.92, −1.74], P < 0.05, I² = 53%) (Figure 3).

Figure 3.

Meta-analysis of the impact of VR on overall symptoms among people with schizophrenia.

The Influence of VR on Positive Symptoms in People With Schizophrenia

In all, 454 subjects were tested in 8 RCTs for the impact of VR on positive symptoms (PANSS-P). Using random effect model-based meta-analysis, it was revealed that VR did not significantly alter patient PANSS-P scores, relative to controls (SMD = −0.49, 95% CI = [−1.03, 0.06], P = 0.08, I² = 19%) (Supplemental Figure 1).

The Influence of VR on Negative Symptoms in People With Schizophrenia

Overall, 8 RCTs examining 455 subjects evaluated the impact of VR on negative symptoms (PANSS-N). Random effect model-based meta-analysis revealed that VR considerably reduced patient PANSS-N scores, relative to controls (SMD = −1.38, 95% CI = [−2.46, −0.30], P < 0.05, I² = 34%) (Figure 4).

Figure 4.

Meta-analysis of the effect of VR on negative symptoms among people with schizophrenia.

The Influence of VR on the Quality of Life of People With Schizophrenia

Two of the seven studies investigating quality of life were excluded due to a lack of quantifiable data. A total of 639 subjects from five RCTs were therefore included in the meta-analysis. Based on our random effect model-based meta-analysis, VR did not markedly enhance patient QoL scores, relative to controls (SMD = 0.00, 95% CI = [−0.34, 0.35], P = 1.00, I² = 52%) (Supplemental Figure 2).

Subgroup Analysis

To better understand potential differences in treatment outcomes, a subgroup analysis was conducted based on the type of VR treatment received (immersive vs. non-immersive). This classification was chosen because immersive VR systems, which provide a fully simulated environment, are hypothesized to have a stronger therapeutic impact compared to non-immersive VR, which primarily relies on external screen displays. Clinical outcomes, including overall symptoms and negative symptoms, were evaluated within these subgroups. Our meta-analysis found no statistically significant difference between immersive and non-immersive VR interventions in terms of overall symptom alleviation (P = 0.06; Supplemental Figure 3) or negative symptom improvement (P = 0.38; Supplemental Figure 4). Furthermore, a subgroup analysis of the overall clinical symptoms and negative symptoms improvement was performed based on the VR intervention types (VRC group and AT group). The results demonstrated that there was no statistically significant difference between the two intervention types in terms of overall clinical symptom alleviation (P = 0.77; Supplemental Figure 5). However, the present study found that the VRC group exhibited superior improvement in negative symptoms in comparison with the AT group (P = 0.01; Supplemental Figure 6). Finally, we conducted a subgroup analysis comparing the two intervention types in terms of QoL improvement. The results indicated that non-immersive interventions were superior to immersive VR interventions in enhancing QoL (P = 0.03; Supplemental Figure 7), whereas no significant difference was observed between the VRC and AT groups (P = 0.53; Supplemental Figure 8).

Publication Bias

We used funnel plots to assess publication bias for the primary outcome, and the data are shown in Supplemental Figures 9 and 10. The funnel plots were largely visually symmetrical, and the data from the Egger regression test indicated that there was no significant publication bias for either the overall symptoms (P = 0.21; Supplemental Figure 9) or the negative symptom outcomes (P = 0.81; Supplemental Figure 10).

Discussion

Summary of Major Findings

Prior reports strongly suggested a positive outcome of VR-based interventions on the cognitive abilities of people with schizophrenia.^12,48 Our findings, particularly the ability of VR to alleviate negative symptoms in patients, align with the conclusions of systematic reviews conducted by Novo et al., ³⁶ Miranda et al. ³⁷ and Schroder et al. ³⁸ Consistent with the findings of the recent meta-analysis by Zeka et al., ³⁴ our study also found no significant improvement in patients’ QoL as a result of VR use. However, Zeka F et al. also concluded that VR led to a significant improvement in positive symptoms in patients, which contrasts with the findings of the present study. Potential explanations for this discrepancy may include methodological differences between studies. Although our analysis overlapped with Zeka et al. in terms of four studies, we exclusively included RCTs of immersive and non-immersive VR interventions in schizophrenia patients, VR studies that focused exclusively on improvements in cognitive domains (e.g., working memory, sustained attention) and did not assess core clinical symptom outcomes were excluded from our study. These variations in inclusion criteria and scope may account for the divergent conclusions. Future studies should prioritize standardized protocols and expanded cohorts to validate these findings.

In the context of the present meta-analysis, seven of the RCTs^42-46,49,50 included were found to overlap with several major previous meta-analyses and systematic reviews.^34,38,39 Despite the overlapping literature, we also included several RCTs that were not included in other meta-analyses.^47,48,51 These studies serve to extend current understanding of the matter and provide new evidence for the use of VR interventions in the treatment of schizophrenia.

Finally, we performed subgroup analyses based on whether the VR intervention was immersive and the different intervention modalities used in the experimental groups. The results indicated that neither immersive nor non-immersive VR had a significant statistical effect on the outcomes. However, significant statistical differences were observed in the intervention for negative symptoms between the VRC and AT groups (P = 0.01; Supplemental Figure 6). The structured nature of the VRC group, based on cognitive behavioural therapy, is reflected in clearly defined intervention phases and standardized operational manuals. Conversely, the AT group employs VR-based role-play, placing significant emphasis on real-time interactions between patients and virtual avatars. This approach may place greater demands on the initiative of patients with negative symptoms. The VRC intervention has been shown to help mitigate withdrawal behaviours resulting from impaired executive functioning by using standardized scenarios and explicit instructions. A plethora of studies have previously indicated a robust correlation between executive dysfunction and negative symptoms,^52,53 thereby suggesting that structured interventions may prove efficacious in supporting patients in the more effective management of these challenges. Ultimately, repeated training may transform complex social interactions into automated responses, thereby reducing avoidance behaviours caused by skill deficits. On the other hand, the overall clinical symptoms measure includes multidimensional symptoms (e.g., positive, negative and general symptoms), which may have also influenced subgroup specificity. These findings suggest that future research should select appropriate intervention techniques based on the target symptoms that patients need to improve in order to achieve better therapeutic outcomes. However, due to differences in study design, intervention goals and VR technology platforms, these approaches are characterized by greater heterogeneity in intervention intensity, frequency and content, which affects the assessment of the overall effects of VR interventions and increases the complexity of interpreting study results.

Despite the primary focus of this report being on the impact of VR therapy on negative symptoms, analyses on positive symptoms and QoL were included for the following reasons: Firstly, both positive symptoms and QoL are significant components of schizophrenia pathology and patient outcomes. Exploring their response to VR provides a more comprehensive understanding of the therapy's broader effects. Secondly, while negative symptoms represent a critical target for VR interventions, analyzing other symptom dimensions (e.g., positive symptoms) allows for a comparison of VR efficacy across distinct schizophrenia symptom domains. This comparison not only contextualizes the observed benefits for negative symptoms but also highlights the limitations of VR therapy in other areas, such as positive symptoms and QoL.

We hypothesize that the differential effects of VR on the positive and negative symptoms of schizophrenia may be attributed to their distinct pathophysiological mechanisms. Positive symptoms, such as hallucinations, delusions and thought disorders, are commonly linked to functional and biochemical imbalances in the brain, particularly involving hyperactivity of dopaminergic pathways. ⁵⁴ On the other hand, negative symptoms, including emotional apathy, social withdrawal and lack of motivation, are primarily associated with deficits in normal brain functioning, often related to hypoactivity in specific regions, such as the prefrontal cortex. ⁵⁵ This difference in underlying mechanisms may explain why VR therapy demonstrates a stronger effect on negative symptoms than positive symptoms in patients with schizophrenia. VR-based interventions, particularly serious games, are thought to stimulate cognitive and social engagement, which aligns closely with the needs of patients experiencing negative symptoms. Previous studies have supported this hypothesis by showing that VR interventions effectively alleviate negative symptoms. ³⁷ Our findings corroborate these results and suggest that VR-mediated improvement in negative symptoms may be due to enhanced activation of brain regions involved in motivation and emotional processing, as well as improved opportunities for patients to engage in structured social interactions.

VR produces real-life simulations in a safe, controlled and relaxing virtual world, thereby stimulating participant emotional and behavioural motivation.^56,57 Given that negative symptoms often originate from a lack of social interaction and environmental stimulation, regular VR can robustly enhance patient social interactions via diverse scenario simulations, behavioural activation and social skills training. In addition, researchers can design personalized and dynamic treatment plans based on the specific conditions of patients. This can allow negative symptom patients to better integrate their abilities and adapt to the real environment. ⁵⁸ In case of positive symptoms, despite provision of behavioural and cognitive training, VR has limited influence on patient positive symptoms and therefore produces inferior outcomes, relative to the current medication-based interventions. ⁵⁹ Moreover, certain VR-based stimuli may be misconstrued by patients as real threats. This may aggravate existing delusions or hallucinations. ⁶⁰ It is, however, reassuring to note that no studies reported severe adverse effects of VR in people with schizophrenia. Therefore, VR-based interventions are practical for individuals with schizophrenia.

In this report, we revealed no significant improvement of patient QoL with VR treatment. Subgroup analysis further demonstrated that non-immersive VR exhibited a more pronounced effect on enhancing QoL in comparison to immersive VR. However, due to the limited number of studies included and heterogeneity (I²=73%), this conclusion should be interpreted with caution. Additionally, no significant differences were observed between the VRC group and the AT group in this study. This may be due to the following reasons: Firstly, the total intervention duration of the examined RCTs may have been insufficient to induce a significant change in patient QoL. Thus, exploration of alterations in patient QoL following prolonged VR treatment is warranted. Secondly, patient QoL is not only determined by symptoms of disease, ⁶¹ but also by support from society, ⁶² economic situation of patient and families, physical health, and other factors. ⁶³ Thus, proper management of a singular factor may not effectively enhance patient QoL.

Study Limitations and Suggestions for Future Research

This study has some notable strengths: Firstly, prior systematic reviews and meta-analyses primarily focused on the impact of VR on cognitive functioning of neuropsychiatric disorder patients. In contrast, this report examined the influence of VR therapy on the symptoms and QoL of people with schizophrenia. In addition, we also explored the efficacies of distinct VR therapies using subgroup analysis, however, our result was not significant.

Among the limitations of our investigation are the following: Firstly, our meta-analysis included English-only articles, which may have introduced selection bias. Secondly, intervention controls varied substantially between test and control groups, which resulted in heterogeneity in results. Owing to this challenge, we were unable to conduct further subgroup analyses based on the type of control condition. Thirdly, the quantity of analyzed RCTs was relatively low, i.e., only 10, therefore, our results require further validation with a large sample size, i.e., more RCTs. It should be noted that although our search strategy exclusively targeted schizophrenia, some of the included studies involved mixed populations comprising both people with schizophrenia and people with schizoaffective disorder (e.g., S. A. Nijman, 2022 ⁵¹ ; Edit Vass, 2022 ⁴⁷ ). Although these two conditions share similar symptom profiles, this may affect the generalizability of our findings specifically to people with schizophrenia. Future studies should conduct diagnostic stratification analyses to determine the differential effects of VR therapy across related disorders. We observed certain statistical heterogeneity in our meta-analysis, which was addressed as follows: Firstly, potential heterogeneity was considered in results between various RCTs likely due to variations in the study protocols. Secondly, one study presented data in the form of the 1st and 3rd quartiles, ⁴⁸ which we converted to mean and standard deviation values, which may have introduced heterogeneity. Moreover, adoption of various scales of measurement for patient QoL may have incorporated diversity in results. Finally, variations in sample characteristics, sample size and VR treatment modality (such as, duration, frequency and total treatment duration) across studies may have contributed to statistical heterogeneity.

Conclusion

Using meta-analysis, we demonstrated that VR significantly improved overall clinical symptoms and negative symptoms of people with schizophrenia. Further subgroup analyses of the results indicated that there were differences in the improvement of negative symptoms among patients across the different interventions. However, the patient QoL remained unaffected, which may be due to its regulation by other factors. This study not only provided new perspectives on schizophrenia management, but also provided strong evidence for VR usage in treating psychiatric patients. VR possesses certain unique benefits over traditional interventions. Firstly, it can be personalized based on individual patient needs and preferences, which can significantly enhance patient outcome. Secondly, relative to other interventions, VR is more flexible, particularly, in terms of treatment location and timing, as long as the appropriate equipment is available, and without worry regarding scheduling conflicts.