Examining the effectiveness of virtual, augmented, and mixed reality (VAMR) therapy for upper limb recovery and activities of daily living in stroke patients: a systematic review and meta-analysis.

The trials disagreed with each other sharply on every pooled outcome, worst of all on the Wolf Motor Function Test. The authors split the trials by recovery stage to try to explain it and say substantial disagreement remained in the Fugl-Meyer subgroups. Immersive and non-immersive systems were pooled together, and so were virtual, augmented and mixed reality. The authors state they did not separate them, so this review cannot tell you which kind of technology to choose. Group sizes ran from 3 to 263, so a handful of large trials carry much of the pooled estimate while several arms had fewer than ten patients. Not every included trial was an arm study. The table lists robot-assisted gait training, Lokomat walking, vestibular rehabilitation and balance testing among the 50 trials, and at least one balance trial reported a Box and Block score, so the upper limb evidence base is smaller than 50 trials suggests. The subgroup results rest on very few trials. The independence finding for chronic patients came from 4 trials and the subacute comparison from 2. The review states the control groups were treated at the same frequency as the intervention groups, but its own table contradicts this in places. In one trial the virtual reality group had 60 min sessions 5 times a week against 30 min sessions 3 times a week in the control group, so extra practice time rather than the technology could account for part of a difference. Thirty-eight of the studies measured only before and after the intervention. Only the remainder looked again a month later, so whether any gain lasts is largely unknown. The database searches did not use MeSH terms, which the authors say may have limited what they found, and only English language articles were eligible. No single outcome is named as the review's main one, so the split verdict across the four scales has to be read as a whole rather than as one headline result. Almost all the pooled outcomes cover motor, daily and hand function. Social functioning and cognition were barely measured, so the review says nothing useful about them.

The work was funded by the Hong Kong Government. No company that makes or sells virtual, augmented or mixed reality equipment is named as a funder, and the text supplied carries no individual author disclosures.

Do not read this as proof that adding virtual reality makes a patient's hand work better. Pooled across all the trials, the Box and Block Test and the Wolf Motor Function Test put the technology groups slightly ahead with no clear advantage, and a dexterity gain surfaced only when the chronic-stage trials were analysed on their own. The two results that did reach significance were averages of 3.91 and 4.25 points, and the paper never states the range of either scale, so it does not tell you how much that is worth to a patient.