The Effectiveness of Immersive Virtual Reality-Based Treatment for Mental Disorders: A Systematic Review With Meta-Analysis

Background: The increasing number of studies of immersive virtual reality (VR) interventions for mental disorders call for an examination of the current level of evidence on their effectiveness. The findings may guide scalability and contribute to the advancement and optimization of immersive VR-based interventions for mental disorders.

Methods: A systematic literature search across four databases screened 2443 studies. Outcomes were disorder-specific symptoms, cognition, function, and quality of life. The study is registered on PROSPERO (CRD42023465845) and follows the reporting standards outlined in the PRISMA guidelines.

Results: Fifty-five studies involving a total of 3031 participants covering 10 mental disorders were included in the analysis. VR interventions demonstrated statistically significant effects of post-treatment compared to active control conditions for alcohol use disorder (reduced state anxiety, g = 0.89, 95% CI[0.24, 1.55]) and schizophrenia spectrum disorders (reduced psychotic symptoms, g = 0.37, 95% CI[0.04, 0.70]). Compared to passive control conditions, statistically significant effects of VR interventions were observed for panic and agoraphobia (g = 1.28, 95% CI [0.47, 2.10]), social anxiety disorder (g = 0.83, 95% CI [0.49, 1.17]), specific phobias (g = 1.07, 95% CI[0.22, 1.92]), depression symptoms in PTSD (g = 0.67, 95% CI [0.22;1.13]). In contrast, no significant differences were found between VR interventions and active control conditions for functioning and quality of life in schizophrenia spectrum disorder and panic or agoraphobia. No meta-analyses were conducted on cognition due to insufficient data. Over 50% of the included studies were assessed as having a high risk of bias. According to the GRADE assessment, evidence for VR-based interventions across various mental disorders was generally of low to very low certainty, with a few exceptions rated as moderate certainty.

Conclusion: VR interventions may potentially have benefits, particularly when compared to passive control conditions, however, the evidence remains uncertain necessitating more large-scale, methodologically robust studies. Current findings can thus only be considered indicative. Recommendations on future directions of the VR field are discussed.