Opinion article | Thomas V. King MD | 24 September 2026
Patients increasingly ask whether they need a robot for their joint replacement. After 15 years of promotion, that is a reasonable question. The answer should depend on better pain relief, better function, fewer complications, or longer lasting implants—not on whether a hospital owns an expensive machine.
A new randomized trial in The Lancet puts that question to a direct test. The RACER-Knee investigators assigned 339 patients at 10 UK hospitals to robotic-arm assisted or conventional total knee replacement. At 12 months, robotic surgery had not produced a clinically meaningful benefit for patients and cost more.
What repeated studies have found
The Lancet result is part of a longer pattern. A systematic review of 12 randomized trials involving 2,200 patients found that robotic surgery probably had little or no effect on patient-reported outcomes or range of motion. The review did not find a difference in revision rates or major adverse events. [2]
In a large randomized study with a mean of 13 years of follow-up, patients reported similar function after robotic and conventional knee replacement. Implant survival at 15 years was 98% in both groups. The authors challenge the assumption that greater technical precision inevitably translates into longer implant life. [3]
Registry data tell a similar story about early revisions. An American Joint Replacement Registry analysis found no lower adjusted odds of revision at two years with robotic assistance. A recent Dutch registry study covering 79,115 knee replacements also found no significant difference in adjusted two-year revision risk.
Where the evidence is mixed
There are findings that deserve attention on the other side. A randomized trial of 180 patients using an imageless handheld system reported better pain, function, satisfaction, and other patient-reported scores at one year. It also recorded fewer ligament releases in the robotic group, while finding no difference in serious adverse events.
The trial was funded by Smith & Nephew, and four authors disclosed paid consulting relationships with the company outside this work. These disclosures do not invalidate the findings, but they should inform how we weigh them. The reported patient benefits rely primarily on subjective questionnaires, and we could not verify whether patients and outcome assessors remained blinded. [6]
An alternate conclusion might be that the non-robotic surgery was performed by surgeons with inadequate experience in non-robotic surgery.
The standard should be benefit to patients
I have performed joint replacement with and without a robot and believe the surgeon’s judgment, soft-tissue management, pain control, rehabilitation, and continuity of care matter greatly. That is my clinical perspective, not a claim that my approach has been compared head-to-head with every robotic program.
I think of a surgical robot much as I think of the GPS in my car. I do not use it to drive home from work every day. I modify my driving depending upon road conditions, traffic, weather, and experience. The GPS may even misdirect me. From my perspective, a robotic arm can obstruct my view and get in the way of the fine adjustments I need to make. The technology must help me respond to what I see and feel during surgery, not make that work harder.
The evidence so far does not justify telling patients that a robotic knee replacement is inherently better, or that conventional surgery is obsolete.
In the past, orthopedics has embraced technologies before their promised benefits were firmly established, leading to disastrous results for some patients. Metal-on-metal hips and thermal capsulorrhaphy in the shoulder are cautionary examples.
Innovation deserves an open mind. It also deserves the same standard we apply to other interventions: show patients a worthwhile improvement, account for the cost, and keep testing as the technology changes.
For now, patients should ask a surgeon practical questions: What are your outcomes? What does the robot improve for me? Is the claimed benefit demonstrated in patients? Until there are clear answers, marketing should not outrun the evidence.
References
[1] Parsons H et al. RACER-Knee randomized trial. The Lancet. 2026. doi:10.1016/S0140-6736(26)00986-4. https://pubmed.ncbi.nlm.nih.gov/42624811/
[2] Ruangsomboon P et al. Randomized-trial systematic review. Acta Orthopaedica. 2023. https://pubmed.ncbi.nlm.nih.gov/36805771/
[3] Kim YH et al. Long-term randomized trial. Clinical Orthopaedics and Related Research. 2020. https://pubmed.ncbi.nlm.nih.gov/31389889/
[4] American Joint Replacement Registry analysis of early revision. Clinical Orthopaedics and Related Research. 2024. https://pubmed.ncbi.nlm.nih.gov/37962943/
[5] Vink MC et al. Dutch arthroplasty register analysis. Knee. 2026. doi:10.1016/j.knee.2026.104632. https://pubmed.ncbi.nlm.nih.gov/42759156/
[6] Bollars P et al. Imageless handheld robotic knee randomized trial. 2025. https://pubmed.ncbi.nlm.nih.gov/40479849/
