Robotic assistance in knee arthroplasty is expanding faster than the evidence for it, and the honest question is whether better component positioning translates into better patient outcomes. This double-blind randomised trial provides an interim answer.
One hundred and seven patients with symptomatic medial compartment osteoarthritis were randomised to conventional jig-based unicompartmental knee arthroplasty with navigational control or to robotic arm-assisted surgery. Both groups had CT-based planning, the same medial parapatellar approach and identical rehabilitation, with outcomes recorded at intervals to two years.
The robot did what it claims. Accuracy in achieving planned femoral and tibial component positions was better (both p<0.001). Inpatient pain scores were lower (p<0.001), inpatient opioid consumption was lower (p=0.008), and length of stay was shorter (p=0.004).
But at two years the Oxford Knee Score, KOOS and WOMAC showed no difference (p=0.299, 0.261 and 0.281). The one outcome that did favour the robot at two years was the Forgotten Joint Score (p=0.021), which measures how often the patient is aware of the joint — a different construct from pain or function.
One observation in the paper is worth noting for what it says about surgical trials generally. Recruitment declined during the study because more eligible patients preferred robotic surgery and the operating surgeons lost equipoise. That is how a technology becomes standard without ever being shown to improve the outcomes that matter, and it is the reason this interim report is worth reading carefully rather than as an endorsement.
- Robotic assistance improved component positioning accuracy and early recovery in unicompartmental knee arthroplasty
- Standard functional scores — Oxford, KOOS and WOMAC — showed no difference at two years
- The Forgotten Joint Score favoured the robot, which measures joint awareness rather than pain or function
- Weigh the shorter stay and lower opioid use against capital and per-case cost in your own setting
- Note the loss of equipoise during recruitment, which limits how far these interim results can be generalised
The statistics, in plain English
This is an interim analysis of a trial whose recruitment was compromised by loss of equipoise, so the sample is the cohort that happened to be recruited rather than a prespecified target — that weakens the null findings in particular, because an underpowered trial finds no difference easily. The p values for the functional scores, around 0.26 to 0.30, are far from significance, but without knowing the intended sample size it is hard to say whether that is a true null or insufficient power. Note also that positive results are reported for process measures the robot directly controls, and null results for the patient-reported outcomes that matter — a pattern worth recognising in device trials generally.
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.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for orthopaedics, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free