Robotic knee replacement, masked with sham incisions, gives nothing back

Design

pragmatic, multicentre, participant-masked and assessor-masked superiority randomised controlled trial at 10 hospitals in Great Britain (RACER-Knee)

Population

339 patients with advanced knee osteoarthritis (168 robotic, 171 conventional); median age 68.8 years, 49% female; 33 surgeons

Primary outcome

Forgotten Joint Score at 12 months, intention to treat, prespecified target difference 12 points

Effect

49.2 vs 50.2; adjusted mean difference -1.5 (95% CI -7.5 to 4.5), P=0.62, favouring conventional surgery; 16 serious adverse events in each group

Robotic assistance for total knee replacement has spread on the strength of alignment accuracy and marketing rather than patient outcomes, and it is expensive. RACER-Knee tested it properly. Ten hospitals in Great Britain and 33 surgeons randomised 339 patients with advanced knee osteoarthritis, 1:1, to conventional instruments or the Mako robotic-arm system, excluding inflammatory arthropathy, previous fracture and cases needing complex implants.

The masking is what makes this trial unusual. Participants were blinded using sham incisions and additional draping, operation notes were masked, and assessors were masked too, which removes the enthusiasm effect that has coloured most previous comparisons. The primary outcome was the Forgotten Joint Score at 12 months, a measure of how completely a patient stops noticing the joint, with a prespecified target difference of 12 points.

At 12 months the mean score was 49.2 in the robotic group and 50.2 in the conventional group, an adjusted difference of -1.5 (95% CI -7.5 to 4.5, P=0.62), numerically favouring conventional surgery. Serious adverse events were identical, 16 in each group. The robotic approach cost more.

The result is worth more than its subject, because it is a demonstration of what happens when a device is tested with the rigour applied to a drug. A technology can improve an intermediate measure, here component alignment, without improving how the patient's knee feels a year later. Where a robotic system is already installed, nothing here suggests harm; where the question is whether to buy one, this trial answers it for patient-reported outcomes at one year, with longer follow-up still running.

• Do not present robotic assistance to patients as offering a better functional result at one year
• Ask what a new surgical technology improves for the patient, not what it improves on imaging or in theatre
• Note the confidence interval excludes the 12-point target difference the trial was designed to detect
• Recognise sham incisions and masked operation notes as the design feature that makes this trial credible
• Long-term follow-up continues; implant survival is a different question this trial has not yet answered

The statistics, in plain English

The confidence interval, -7.5 to 4.5 points, excludes the 12-point difference the trial was designed to detect, so this is not an underpowered trial that failed to find an effect: it rules out the benefit that would have justified the technology. Blinding matters more here than in most surgical trials, because the Forgotten Joint Score is entirely patient-reported and a patient who knows they had the newer operation tends to report better outcomes. Sham incisions and masked notes remove that, which is why a null result from this design carries more weight than a positive result from an unmasked one.

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.