Design: systematic review with meta-analyses, stratified by immersion depth and mechanism, GRADE certainty
Population: 12 trials in knee osteoarthritis and 9 after total knee arthroplasty; samples 36 to 306, follow-up mostly 3 months or less
Primary outcome: pain-related functioning and pain intensity
Effect: osteoarthritis pain-related functioning SMD −0.59 (95% CI −1.11 to −0.06); post-arthroplasty pain intensity SMD −0.12 (−0.75 to 0.52)

A systematic review searched to December 2025 for trials of extended reality in knee osteoarthritis or recovery after total knee arthroplasty, and — unusually — stratified them by depth of immersion and clinical mechanism rather than lumping every headset and app together. Twelve trials addressed osteoarthritis and nine addressed post-arthroplasty recovery, with samples from 36 to 306 and follow-up mostly three months or less.

For knee osteoarthritis, interactive digital rehabilitation — guided exercise with feedback, not full immersion — improved pain-related functioning (pooled SMD −0.59, 95% CI −1.11 to −0.06, k = 5) and reduced pain intensity at six to eight weeks (SMD −0.46, −0.92 to 0.00, k = 4), both at low certainty. After arthroplasty the same approach improved pain-related functioning but made little difference to pain intensity (SMD −0.12, −0.75 to 0.52, k = 3). Virtual reality digitally augmented exercise was inconsistent throughout. No adverse event signal was found for interactive digital rehabilitation, though only five studies reported adverse events at all.

The authors' conclusion is appropriately unequal: interactive digital rehabilitation may be ready to integrate into osteoarthritis care, while its use after arthroplasty needs more evidence. That distinction matters commercially as well as clinically — these products are sold as one category.

Consider a guided digital rehabilitation programme as an adjunct in knee osteoarthritis, not a replacement for physiotherapy
Do not extend the osteoarthritis evidence to post-arthroplasty rehabilitation
Ask what a product actually delivers — guided exercise with feedback, or immersive virtual reality
Check follow-up duration before believing a durability claim; most trials stopped at three months
Note that adverse events were barely reported, so safety is assumed rather than established

Why it matters

It separates a category that is marketed as one thing into an intervention that works and several that have not been shown to.

The statistics, in plain English

An SMD of −0.59 is a moderate effect, but the prediction interval runs from −1.72 to 0.55, meaning a future trial could plausibly find no benefit at all. The pain intensity confidence interval touching exactly 0.00 is a result that is statistically borderline, not a clear effect.