They matched the exercise dose exactly, and the VR advantage fell below the line that matters

What happened

Immersive virtual reality can standardise a rehabilitation exercise and give immediate feedback. Whether that is worth anything, once you control for the extra therapist time that VR groups usually get, has not been clear.

A team at the First Affiliated Hospital of Fujian Medical University, its Binhai campus sports medicine department and the Fujian Orthopaedics Research Institute built a trial specifically to answer it. 86 adults who had undergone arthroscopic rotator cuff repair for small-to-medium full-thickness supraspinatus tears were randomised 1:1 to VR-assisted or conventional rehabilitation during postoperative weeks 4 to 8.

Both groups received matched exercise content, matched dose, matched progression criteria, matched safety limits and matched face-to-face therapist supervision. Only the delivery format differed.

The pain advantage came in at 0.7 cm. The threshold for a difference patients can perceive is 1.4.

The test

Assessor-blinded randomised trial with assessments at baseline, weeks 4 and 6, and months 2, 3 and 6.

The primary outcome was activity-related pain on a 10-centimetre visual analogue scale. Secondary outcomes were rest pain, the 17-item Tampa Scale for Kinesiophobia, the Constant-Murley Score for shoulder function, global shoulder active range of motion, the SF-12 health survey, isometric strength and supraspinatus cross-sectional area. Surface electromyography was exploratory.

That is a wide net – pain, fear, function, movement, strength, muscle structure and quality of life – which makes the pattern of results more informative than a single endpoint would have been.

The numbers

Measure	VR	Conventional
Randomised / analysed	86 randomised, 70 complete cases (36 / 34)
Mean age	58.1 years; 67.1% women
Baseline activity pain (VAS)	7.3 cm	7.2 cm
Activity pain at 6 weeks	4.0 cm	4.7 cm
Activity pain at 2 months	3.1 cm	3.9 cm
Clinical-importance threshold	1.4 cm
Kinesiophobia at 3 months	-3.1 points (95% CI -4.5 to -1.8), Hedges g -1.11
Function, ROM, SF-12, strength, muscle area, sEMG	no between-group difference
All outcomes at 6 months	no clear difference

The result

Mean activity-related pain was 0.7 cm lower with VR at six weeks and 0.8 cm lower at two months. Both differences were smaller than the published 1.4 cm clinical-importance threshold, and the authors state that explicitly rather than reporting the p-value and moving on.

Rest pain was similar between groups. Group-by-time interactions for the Constant-Murley Score, global shoulder active range of motion, SF-12, isometric strength, supraspinatus cross-sectional area and exploratory surface EMG were all non-significant.

One outcome cleared its bar. The kinesiophobia interaction was significant at p = 0.001, with a mean difference at three months of -3.1 points and a Hedges g of -1.11, which is a large effect size.

By six months there was no clear between-group difference in activity-related pain or kinesiophobia either.

How this compares to what came before

The VR rehabilitation literature is large and largely positive. Systematic reviews repeatedly conclude that VR improves outcomes after musculoskeletal surgery and stroke.

Read the methods sections of those trials and a pattern appears: the VR arm frequently receives more supervised exercise time, or a more engaging protocol, or both. The comparison is often against conventional care as normally delivered, which in practice means less of it.

This trial removes that. Matched content, dose, progression, safety limits and supervision. Under those conditions the effect largely disappears, and what remains is psychological rather than physical.

That is not a refutation of the earlier literature so much as an explanation of it. If VR gets patients to do more therapy, VR helps. If the therapy is held constant, the screen itself does comparatively little.

What it means

For rehabilitation services deciding whether to buy VR, the question this trial reframes is: what are you buying it for? If the answer is better physical recovery for the same amount of therapy, the evidence here says no. If the answer is adherence, engagement, or getting patients to complete sessions they would otherwise skip, that is a different question and this trial does not address it.

The kinesiophobia finding is the one to take seriously. Fear of movement is a genuine obstacle after shoulder and knee surgery, it predicts poorer outcomes, and a large effect at three months is worth something even if it faded by six.

For the broader digital therapeutics field, this is the second trial in this batch – alongside an Italian stroke VR trial – showing that matched-control designs give much smaller and much more specific answers than usual-care comparisons. Both are worth more than a dozen favourable usual-care trials.

Business ideas from this paper

Adherence as the pitch. If the physical benefit of VR is nil at matched dose, the honest commercial claim is that patients complete more sessions. That is measurable, defensible and nobody is leading with it.

Kinesiophobia-targeted modules. The one surviving effect was on fear of movement, with a large effect size. A VR product aimed squarely at that – graded exposure, confidence building – has a plausible mechanism and a published signal.

Therapist-time substitution. The commercially interesting version of VR rehabilitation is not better therapy; it is the same therapy with less supervision. A trial that deliberately unmatches therapist time, in the direction of using less of it, would answer the question services actually care about.

Outcome-threshold literacy. Most buyers of digital health do not know that pain scales have minimum clinically important differences. A procurement checklist that forces the question – is this difference above the threshold – would change a lot of purchasing decisions.

What could be wrong with it

86 randomised and 70 analysed means 16 patients dropped out, and the analysis is complete-case rather than intention-to-treat, which can bias results if dropout differed between arms.

The trial tested weeks 4 to 8 after surgery in a specific injury – small-to-medium supraspinatus tears. Larger tears, different joints and different phases of recovery may behave differently.

Patients could not be blinded to whether they were wearing a headset, which matters most for the self-reported outcomes – pain and kinesiophobia – that produced the only signals.

Many outcomes were tested, which raises the chance that one of them reaches significance by accident. Kinesiophobia had a large effect size and a p-value of 0.001, which makes it more credible than a borderline finding, but it is one result among many.

Single institution.

How sure can you be?

We score this 8 out of 10.

It is a peer-reviewed prospective assessor-blinded randomised controlled trial in BMC Musculoskeletal Disorders, registered with the Chinese Clinical Trial Registry before recruitment began, with named institutions in Fujian and – most importantly – a control arm matched on content, dose, progression, safety limits and therapist supervision.

It loses points because it ran at a single institution with no independent site, and the data are not openly deposited.

What raises our confidence is the reporting. A trial that measures a difference, reaches statistical significance, and then tells you the difference is below the clinical threshold is a trial being honest against its own interest.

Do this today

If you are reading any rehabilitation technology claim, find the minimum clinically important difference for the outcome and check the reported effect against it. Statistical significance and clinical significance are different things and the gap is where most digital health marketing lives.

If you run a rehabilitation service using VR, measure session completion rather than pain. That is where the plausible value is, and it is the thing this trial did not test.

If you are designing a trial of a delivery technology, match the dose. It will give you a smaller, truer number, and it is the only design that answers what the technology contributes.

Source: Virtual reality-assisted rehabilitation after arthroscopic rotator cuff repair for small-to-medium supraspinatus tears: a prospective assessor-blinded randomized controlled trial, BMC Musculoskeletal Disorders, August 2026. PMID 42983341 · doi.org ().

Just Out Tech explains new research in plain language. This article was drafted with AI assistance and checked by a human against the original source.