Virtual reality in autism and physical therapy: a meta-analytical review of clinical outcomes and therapeutic efficacy.
Arar Al Tawil, Siti Hazyanti Mohd Hashim, Bence Király and 1 others
VR added to standard therapy improved social communication and emotion recognition in autistic people, and balance, gait, and arm function in stroke and Parkinson's patients.
It works as an add-on, not a substitute for existing treatment.
Key findings
1Across both autism and physical therapy populations, VR interventions produced a statistically significant positive effect (Hedges' g = 0.66, p < 0.001), with the largest gains in social communication, balance, and gait.
2In autism, social communication showed the strongest effect (g = 0.74, 95% CI 0.61 to 0.87, 7 studies) and emotion recognition g = 0.70; in physical therapy, balance g = 0.72 (6 studies), gait g = 0.69 (5 studies), and upper-limb function g = 0.65 (5 studies).
3Egger's regression test was statistically significant (p = 0.0031), indicating funnel plot asymmetry consistent with small-study effects or publication bias, meaning the true effect is likely smaller than the pooled estimate.
Still to come
How it was doneWhat they foundWhat it means for PTsWhat it means for SLPs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The median study had only 45 participants, and formal power analyses were not uniformly reported, so small effects may be missed and large effects may be inflated. Publication bias is likely: Egger's test was significant (p = 0.0031) and grey literature was excluded, so the true benefit is probably smaller than g = 0.66. Heterogeneity was moderate (I-squared = 44.3%): studies varied in design, population, VR technology, protocol, and outcome measures, which limits how precisely you can apply the pooled number to a specific patient. About 17.4% of included studies had no control group, so some of the observed gains could reflect practice effects or natural recovery rather than VR itself. Few studies assessed whether gains lasted beyond the intervention period, so you cannot yet tell a patient that the effect will persist at 3, 6, or 12 months. The search was limited to three databases and English-language articles, so relevant work in other languages or in databases like CINAHL or Embase may be missing. The quality appraisal used a 4-point scale, which is less detailed than the Cochrane Risk of Bias 2 tool or the Newcastle-Ottawa Scale.
Declared interests
The authors declared that no financial support was received for the work or its publication.
The easy way to misread this
Do not read the pooled effect (g = 0.66) as proof that VR works for every patient in every setting. The meta-analysis detected publication bias (Egger's test p = 0.0031), the median study had only 45 participants, 17.4% of studies had no control group, and the included studies varied widely in design, technology, and protocol. The true benefit for an individual patient is likely smaller than the headline number, and the evidence does not yet show that gains persist after the intervention ends.
Summarised by AI from the full paper, without a clinician reviewing it. Check it against the source before it changes what you do. Read it on PubMed →