RACER-Knee: robotic assistance cost more and changed nothing at a year

Design

Pragmatic, participant-masked and assessor-masked, superiority randomised controlled trial at ten British hospitals

Population

339 patients with advanced knee osteoarthritis (168 robotic, 171 conventional), median age 68.8 years

Primary outcome

Forgotten Joint Score at 12 months, prespecified target difference 12 points

Effect

Adjusted mean difference -1.5 (95 per cent CI -7.5 to 4.5, p = 0.62) favouring conventional; 16 serious adverse events in each arm

Three hundred and thirty-nine patients with advanced knee osteoarthritis were randomised at ten hospitals across Great Britain, with 33 surgeons taking part, to total knee replacement using the Mako robotic-arm system or conventional instruments. The masking is what makes this trial worth reading: participants were blinded with sham incisions, additional draping and masked operation notes, and assessors were masked too. Complex implants, inflammatory arthropathy and previous fracture were excluded, so this is the ordinary primary knee replacement population.

The primary outcome was the Forgotten Joint Score at 12 months, with a prespecified target difference of 12 points. The robotic group scored a mean 49.2 (SD 28.4) and the conventional group 50.2 (29.9). The adjusted mean difference was -1.5 (95 per cent CI -7.5 to 4.5, p = 0.62) - numerically favouring conventional instruments, and nowhere near the target. Sixteen participants in each arm had a serious adverse event. Robotic surgery cost more.

This is the outcome the technology's own advocates said would settle the question, delivered pragmatically in routine practice rather than in a single enthusiast's series. It does not prove the robot can never help - the confidence interval still admits a 4.5-point advantage, well below what a patient would notice, and long-term follow-up continues for implant survival. What it does close off is the argument that robotic assistance improves how a knee feels a year later. For an Indian unit weighing a capital purchase, that is the number that matters, and it is zero.

• Do not quote a patient-perceived benefit from robotic assistance at one year; this trial was designed to find one and did not.
• Judge any future robotic claim against a 12-point Forgotten Joint Score difference, the target used here.
• Note the masking - sham incisions and masked operation notes - before comparing this with unmasked series.
• Longer-term implant survival remains open; the trial's follow-up is ongoing.
• Weigh capital and per-case cost against a measured benefit of zero when the business case is made.

The statistics, in plain English

This is a properly negative superiority result, not an underpowered one: the confidence interval runs from -7.5 to 4.5 and the prespecified meaningful difference was 12, so the trial can exclude the benefit it set out to find. A p value of 0.62 with the point estimate on the conventional side means there is no hint of an effect to chase. It cannot, however, speak to revision rates or implant survival, which need years rather than months.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.