Table 1.

Summary of included studies.

Study (country)		Sample	Diagnosis of participants	Study design	Control condition	VRa exercises	Main findings
Cognitive remediation interventions
	Amado et al [44] (France)	Nb=7, kc=1	Schizophrenia or schizoaffective disorder	Single-arm, unblinded, pilot study	N/Ad	Participants acted as pedestrians in a virtual town and completed tasks that were difficult for them in real life (eg, memorizing an itinerary, shopping, or being on time at a meeting point). After the exercise, participants discussed the possible transfer of skills to their real lives and were assigned a task to perform at home.	Pre- and postintervention assessments: the VR group showed improvements in attention, working memory, prospective memory, retrospective memory, and autonomy. There were no improvements in planning. Postintervention qualitative findings: the VR group reported improvements in their amount of energy to develop concrete plans to look for employment or to return to community activities, sparing time, planning, enriched relatedness, and management of their housework.
	Chan et al [45] (Hong Kong)	N=27, k=2; experimental: ne=12; control: n=15	Schizophrenia	Randomized controlled trial, pilot study	Treatment as usual (attended the usual program in the long-term care facility; did not include VR)	The (1) ball and bird and (2) shark bait activities were chosen to train fluid intelligence. Ball and bird activity: balls of different colors fly toward participants, and they must contact the ball using any part of their body, making the ball “burst” or “transform” into doves. Shark bait activity: participants navigate in the sea and chase a yellow star while avoiding distracters.	Compared with the control group, the VR group showed significant improvements in overall cognitive function, repetition, and memory.
	La Paglia et al [46] (Italy)	N=12, k=2; experimental: n=6; control: n=6	Schizophrenia	Nonrandomized controlled trial, pilot study	IPTf and pharmacological therapy	Tasks to train attention and executive function in VR environments such as a park, valley, beach, and supermarket. For example, participants collected and bought products from a shopping list in a supermarket setting to train executive function.	Pre- and postintervention assessments: both groups showed significant benefits in divided attention. The VR group also showed reduced cognitive deficits and improved planning. After the executive function training (VR supermarket), the experimental group showed improvements in decreased errors, reduced time of execution, and increased observance of rules. After the attention training (VR park, valley, and beach), the experimental group showed improvements in reduced time of execution, decreased perseverative errors, and improved sustained attention.
	La Paglia et al [47] (Italy)	N=15, k=2; experimental: n=9; control: n=6	Schizophrenia	Nonrandomized controlled trial, pilot study	IPT and pharmacological therapy	Three different virtual environments (park, valley, and beach) featured hierarchical sequences of tasks designed to train attention (eg, catching footballs that were presented at irregular time intervals, identifying and picking a specific type of flower, or picking up specific types of bottles while being alerted to calls and loudspeaker announcements).	Pre- and postintervention assessments: both groups showed improvements in divided attention. The VR group showed improvements in general cognitive functioning, planning, sustained attention, reduced time of execution, decreased requests for assistance, decreased needs of the therapist’s intervention, and decreased number of omissions.
Social skills interventions
	Adery et al [48] (United States)	N=16, k=1	Schizophrenia	Single-arm, single blind, feasibility study	N/A	Three different VR environments were used (bus stop, shop, and cafeteria) to train participants in both microlevel social skills (ie, eye contact and facial expression) and macrolevel social skills (ie, starting conversations and requesting help). Tasks were administered without time constraints.	Pre- and postintervention assessments: the VR group showed improvements in overall clinical symptoms and negative symptoms.
	Park et al [49] (Republic of Korea)	N=64, k=2; SST-VRg: n=33; SST-TRh: n=31	Schizophrenia	Randomized controlled trial, single blind, efficacy study	Social skills training using traditional role-playing (in-person with the social skills training therapist as the role-play actor instead of using VR)	Role-play with virtual avatars in environments such as a restaurant or walking down a street. Participants were trained in conversation skills, assertive skills, and emotional expression skills. Helper avatars provided positive and corrective feedback as needed.	Compared with the control group, the VR group showed improvements in conversational skills and assertiveness. Compared with the VR group, the control group showed improvements in nonverbal skills.
	Rus-Calafell et al [50] (Spain)	N=12, k=1	Schizophrenia or schizoaffective disorder with a deficit in social skills or social functioning	Single-arm, unblinded, pilot study	N/A	The VR program comprised 7 activities that each targeted different social skills. Participants received positive or negative reinforcement from virtual avatars based on their performance in a bar or supermarket environment.	Pre- and postintervention assessments: the VR group showed improvements in negative symptoms, psychopathology, social anxiety and discomfort, avoidance, social functioning, learning in emotion perception, assertive behaviors, and time spent in a conversation.
	Vass et al [51] (Hungary)	N=17, k=2; VR-ToMISi: n=9; control: n=8	Schizophrenia	Randomized controlled pilot study	Passive VR control condition (participants used the same VR software as the experimental group but without any intervention)	The VR-based targeted theory of mind (ToM) intervention (VR-ToMIS) used cognitive and behavioral therapeutic techniques. Participants engaged in social interactions with an avatar with prerecorded dialogue that was designed to induce ToM impairment (double meaning sentences, overstatements, and irony). After the interaction, participants visualized the inferred emotions of the avatar. The task was also discussed with a therapist.	Pre- and postintervention assessments: VR-ToMIS was associated with improvements in negative symptoms, in 1 neurocognitive field (immediate memory), ToM, and pragmatic language skills, but no significant change in quality of life was detected. These findings were also significantly greater in the VR-ToMIS group compared with the control group.
Vocational skills interventions
	Smith et al [52] (United States)	N=32, k=2; VR-JITj: n=21; control: n=11	Schizophrenia or schizoaffective disorder	Randomized controlled trial, single blind, efficacy study	Treatment as usual (which did not include a VR component)	The VR-JIT was designed to improve interviewing skills. Participants completed virtual job interview role plays with a virtual human resources representative that were each approximately 20 minutes in duration.	Pre- and postintervention assessments: the VR group showed improvements in role-play job interview scores. Compared with the control group, the VR group showed improvements in the odds of receiving a job offer at a 6-month follow-up. There was also an association between more training and waiting fewer weeks to receive a job offer.
	Sohn et al [53] (Republic of Korea)	N=9, k=1	Schizophrenia	Single-arm, feasibility study	N/A	The virtual reality-based vocational rehabilitation training program included scenarios with a convenience store employee and supermarket clerk. Each scenario included various situations that trained participants on how to manage problems they may encounter in real life. For example, the convenience store situations included training on the arrangement of goods and training for problematic situations.	Pre- and postintervention assessments: the VR group showed improvements in individual and social performance, general symptoms, verbal memory, and immediate and delayed recall on visual memory. Improvements in positive symptoms showed a trend toward significance.
	Tsang and Man [54] (Hong Kong)	N=75, k=3; experimental: n=25; TAGk: n=25; CGl: n=25	Schizophrenia	Randomized controlled trial, single blind, efficacy study	TAG (received therapist-administered vocational training) and conventional treatment group (CG). Neither control group experienced a VR intervention.	The 3D nonimmersive VR training was set in a boutique. The training involved a hierarchical structure divided into levels in which problem-solving competence tests had to be passed to advance levels (pretrainee level, trainee level, and sales level).	Compared with both control groups, the VR group showed improvements in cognitive functioning and executive functioning performance. Compared with the CG, the VR group showed improvements in self-efficacy. Compared with the CG, the VR and TAG groups showed improvements in work performance during an on-site test.
Auditory verbal hallucinations and paranoia interventions
	Dellazizzo et al [55] (Canada)	N=10, k=1	Schizophrenia or schizoaffective disorder	Single-arm study	N/A	Participants who had already undergone CBTm as part of the study by Dellazizzo et al [56] were invited to complete the VR intervention (ie, after finishing CBT). The VR intervention was identical to that used in Dellazizzo et al [56], which is described further in the table.	The VR group showed improvements in auditory verbal hallucinations, beliefs about voices, depressive symptoms, symptoms of schizophrenia, and quality of life.
	Dellazizzo et al [56] (Canada)	N=74, k=2; experimental: n=37; control: n=37	Schizophrenia or schizoaffective disorder	Randomized comparative trial, pilot study	CBT with no VR component	Participants created avatars which they believed most resembled the entity which was the source of their most distressing or dominant auditory verbal hallucination. Participants were encouraged to enter a dialogue with their avatar (which was animated in real time by a therapist). The interaction with the avatar became more supportive and less abusive as the intervention progressed. The conversations were designed to target participants’ emotional regulation, assertiveness, and self-esteem.	Both groups showed improvements in the severity of their auditory verbal hallucinations and depressive symptoms. The VR group also showed improvements in persecutory beliefs and quality of life. Although the results did not show a statistically significant superiority of the VR intervention over CBT in improving auditory verbal hallucinations, the VR intervention did achieve larger effect sizes, particularly on improving overall auditory verbal hallucinations. The VR intervention was superior to CBT at improving affective symptoms.
	du Sert et al [57] (Canada)	N=15, k=1	Schizophrenia or schizoaffective disorder	Randomized, partial crossover trial, pilot study	Treatment as usual (antipsychotic treatment and meetings with clinicians; no VR component).n	Participants created avatars which they believed most resembled the entity which was the source of their most distressing or dominant auditory verbal hallucination. Participants were encouraged to enter a dialogue with their avatar (which was animated in real time by a therapist). The interaction with the avatar became more supportive and less abusive as the intervention progressed. The conversations were designed to target participants’ emotional regulation, assertiveness, and self-esteem. n	Pre- and postintervention assessments: the VR group showed improvements in auditory verbal hallucinations, beliefs about voices, general symptoms (however, positive and negative symptoms did not significantly improve), and quality of life.
	Geraets et al [58] (Netherlands)	N=91, k=2; experimental: n=43; control: n=48	Schizophrenia, schizoaffective disorder, or not-otherwise specified psychotic disorder	Randomized controlled trial, single blind	Treatment as usual with no VR component	VR-based CBT for reducing paranoia and improving social participation was used in this study. Evidence-based CBT elements were used by trained psychologists and exercises and behavioral experiments were completed in VR. Participants interacted with human avatars in social environments (a street, bus, café, and supermarket). Characteristics of the social environments (number of avatars and avatars’ responses to the participant) could be edited by the therapist, and they communicated with the participant during the VR sessions.	Pre- and postintervention assessments (baseline vs 3-month follow-up): compared with the control group, the VR group showed improvements in average levels of paranoia (feeling suspicious, disliked, and hurt) and negative affect (feeling anxious). Pre- and postintervention assessments (baseline vs 6-month follow-up): compared with the control group, the VR group showed improvements in average levels of paranoia (feeling disliked and hurt) and negative affect (feeling down and insecure). Positive affect did not improve more in the VR group than in the control group. The VR intervention did not change the interplay between affective states and paranoia.
	Moritz et al [59] (Germany)	N=33, k=1	Schizophrenia	Single-arm, proof of concept study	N/A	Participants met 6 different pedestrians while navigating through a virtual street on 2 occasions (in addition to 1 practice trial) in either a noise or no noise condition. Then, participants participated in a recognition task graded for confidence where they were asked to recollect the pedestrians and their corresponding facial affect. Participants also received feedback on the accuracy of their recall.	Pre- and postintervention assessments: the VR group showed improvements in paranoia. Improvement was associated with lower confidence ratings (both during the experiment, particularly for incorrect responses, and according to retrospective assessment).
	Pot-Kolder et al [60] (Netherlands)	N=116, k=2; experimental: n=58; control: n=58	Schizophrenia, schizoaffective disorder, delusional disorder, or not-otherwise specified psychotic disorder	Randomized controlled trial, single blind	Treatment as usual with no VR component	VR-based CBT for reducing paranoia and improving social participation was used in this study. Evidence-based CBT elements were used by trained psychologists and exercises and behavioral experiments were completed in VR. Participants interacted with human avatars in social environments (a street, bus, café, and supermarket). Characteristics of the social environments (number of avatars and avatars’ responses to the participant) could be edited by the therapist and, they communicated with the participant during the VR sessions.	Pre- and postintervention assessments (baseline vs 3-month follow-up): compared with the control group, the VR group showed improvements in momentary paranoid ideation when in the presence of others and momentary anxiety when in the presence of others. The VR group did not show a significant improvement in the amount of time spent with others. Pre- and postintervention assessments (baseline vs 6-month follow-up): the improvements shown at the 3-month follow-up were maintained at the 6-month follow-up. At the postintervention and follow-up time points, quality of life did not differ significantly among groups.
	Pot-Kolder et al [61] (Netherlands). o	See Pot-Kolder et al [60]	See Pot-Kolder et al [60]	See Pot-Kolder et al [60]	See Pot-Kolder et al [60]	See Pot-Kolder et al [60] for the specific VR exercises used in this study. The main outcomes of this study related to the feasibility of the VR intervention.	Feasibility of the VR intervention: the average incremental cost per quality-adjusted life year was €48,868 (US $55,220.84). When relevant baseline differences were included, the average cost per quality-adjusted life year gained was €42,030 (US $47,493.9).

^aVR: virtual reality.

^bN: total sample size.

^ck: number of groups.

^dN/A: not applicable.

^en: group sample size.

^fITP: integrated psychological treatment.

^gSST-VR: social skills training using VR role-playing.

^hSST-TR: social skills training using traditional role-playing.

ⁱVR-ToMIS: virtual reality-based targeted theory of mind intervention.

^jVR-JIT: virtual reality job interview training.

^kTAG: therapy administered group.

^lCG: conventional group.

^mCBT: cognitive behavioral therapy.

ⁿThe group comprised participants who received an “immediate” VR intervention as well as participants who received “delayed” VR intervention (that is, after they participated within a control, treatment as usual group). Both VR interventions were identical.

^oThis study used the same sample as Pot-Kolder et al [60]. This study reported novel outcomes.