Robotic knee replacement was no better for patients at one year than conventional surgery in a blinded trial

Hospitals around the world are buying robots to help surgeons replace knees, on the promise that a more precise operation makes a better joint. The largest blinded trial of the idea, by its organizers’ description, found that the benefit to patients, a year on, was not there. The organizers say the precision was.

In the trial, published in The Lancet in August, 339 people in Great Britain had a total knee replacement either with a robotic arm guiding the surgeon’s cuts or with conventional hand-held instruments. Twelve months later the two groups scored much the same on how much they noticed their new knee, how much it hurt and how well they walked. The robotic operations took longer and cost more.

The investigators do not call it a verdict against the technology. They will follow the patients for ten years to see whether better-placed implants last longer.

What the robotic knee replacement trial tested

A total knee replacement swaps the worn surfaces of an arthritic knee for metal and plastic parts. In the conventional version, surgeons rely on their experience and a standard set of instruments to perform the procedure. In the robotic version a mechanical arm, controlled by the surgeon, holds the cutting tools to a plan drawn up for that patient.

Use of the machines has spread ahead of the evidence. Robotic systems are being increasingly used to assist with total knee replacement, the authors note. The share of knee replacements done with a robot is about 6% in Britain, 16% in the United States and 42% in Australia, according to the University of Warwick, which coordinated the trial.

The belief behind that growth was written into earlier research. The protocol for a smaller British trial, published in 2022, stated that robotic-arm-assisted surgery is associated with better patient-reported outcomes compared to manually performed jig-based knee arthroplasty. Arthroplasty means joint replacement, jigs are the hand-held cutting guides of conventional surgery, and patient-reported outcomes are results that patients rate themselves. The authors of the new trial were more cautious, describing it as uncertain whether robotic systems improve outcomes.

Earlier comparisons had a weakness: patients knew which operation they had been given, and a patient who knows a robot was used may rate the result differently. The new trial, known as RACER-Knee, kept them unaware. It ran at ten hospitals in Great Britain involving 33 surgeons. In all, 807 patients were screened and 339 were randomly assigned, 168 to robotic surgery and 171 to conventional. Their median age was 69 and about half were women. The robot was the Mako system made by Stryker, which the university describes as the most widely used in the world.

Blinding a surgical trial takes some stagecraft, because the two operations leave different marks. Participants and assessors were masked through methods including using sham incisions, additional draping, and masked operation notes. Masked means kept unaware of which operation had been done.

What the trial found a year after knee surgery

The main measure asked patients about their knee in everyday life. The score measures the extent to which patients are aware of their artificial joints during daily activities. It runs from 0 to 100. A higher score means the joint is noticed less, and the ideal replacement is one its owner forgets.

	Robotic	Conventional
Patients randomized	168	171
Joint awareness score at 12 months	49.2	50.2
Patients with a serious adverse event	16	16

The difference, after adjustment, was 1.5 points in favor of conventional surgery, with a plausible range from 7.5 points in favor of conventional to 4.5 in favor of the robot. Before the trial began, the investigators had decided what would count as worthwhile. The prespecified target difference was 12 points. The whole of the plausible range falls short of it, so the trial does more than fail to find a benefit. It rules out one of the size the researchers were looking for.

Other measures agreed. One year after surgery, patients in both groups had similar recovery, mobility and pain levels. Pain in hospital and in the first three months was also comparable, as was the likelihood of a further operation.

The authors’ conclusion is that robotic surgery, as delivered in routine practice, did not provide a clinically meaningful patient benefit at 12 months.

By the university’s account the robot did what it was built to do, placing implants more precisely. The announcement gives no figure for how much more. Whatever the gain in precision, it did not reach the patient.

Safety, time and cost of robotic knee replacement

Safety was a draw. In the trial, 16 participants in each group had one serious adverse event, an adverse event being any medical problem recorded during a study.

Infection is one of the complications surgeons watch for after any joint replacement. An earlier pooled analysis looked at that question alone. Eight studies were included, comprising 758,453 knees. Deep infection rates were 0.96% and 0.66% in conventional and robotic procedures, respectively, the robotic figure being the lower one, though the authors could not tell the difference apart from chance.

Time and money went against the robot. Robotic operations took longer to perform (on average 10.5 minutes) and cost about 950 pounds more each, a cost that the robot’s maker covered within the trial. Measured against the limits the National Health Service uses to decide what is worth paying for, the robotic system was not cost-effective over the first year after surgery.

What national registry data show about robotic joint replacement

A trial of 339 people cannot say much about rare events such as an implant failing. For that, two studies published in the BMJ on September 30 used the National Joint Registry, which covers England, Wales, Northern Ireland and nearby islands. They were led from Queen Mary University of London.

The knee study covered 697,145 operations between 2018 and 2024, of which 22,111, or about 3% across the whole period, were robotic. After matching similar patients, meaning each robotic case was compared with conventional cases alike in age, sex and other features, the share of total knee replacements still in place at five years was 98.5% for conventional surgery and 98.6% for robotic, with no between group difference in revision risk. Revision is the term for a repeat operation to replace a failed implant.

The hip study covered 666,283 operations. Five-year implant survival was 98.8% either way. One difference did appear. Robotic hip replacement was associated with a lower revision risk from causes related to implant malpositioning, such as dislocation, at about half the rate. Such revisions are rare with either method.

Both papers carry the standard warning for studies that observe and do not assign treatment. The possibility of unmeasured and residual confounding cannot be ruled out, confounding meaning that the groups differed in ways the analysis could not fully correct. Hospitals and surgeons that buy robots are not a random sample. The registry authors raise the question of money too, calling for careful evaluation given the substantially higher capital and procedural costs involved, meaning the price of the machine and of each operation.

Why the knee replacement result matters

Knee replacement surgery is one of the most commonly done and cost-effective musculoskeletal surgical procedures, a 2018 review in The Lancet noted. Musculoskeletal means of the bones, joints and muscles. It also leaves a stubborn minority disappointed. Most patients do well, the same review said, but multiple studies have reported that 20% or more of patients do not.

Robots were one of the hopes for shrinking that minority. On the evidence so far, more precise placement has not been shown to reduce it.

The investigators frame the result as a stage in development. “Technology such as this could really help the care we give for patients, but for robotic-assisted knee replacements, there is still work to do before we see benefits such as less pain or better movement,” said Andrew Metcalfe of Warwick Medical School, one of the two chief investigators.

The other, Edward Davis of the Royal Orthopaedic Hospital in Birmingham, argued that surgeons do not yet know what position is ideal for each patient. “These results are by no means a condemnation of the robotic systems,” Davis said.

Limits of the robotic knee replacement trial

One year. Implants are meant to last decades. The authors note that long-term follow-up is ongoing, with patients to be tracked for ten years. If precision reduces wear or loosening, the benefit would show up late.

One robot, one country. The trial tested a single system in British hospitals. Other machines and other ways of planning the operation were not tested.

Routine practice. The trial was pragmatic, meaning it studied the operation as it is ordinarily done. It does not show what the robot might achieve with different targets for positioning the joint.

The maker’s role. The trial was funded by the UK’s National Institute for Health and Care Research (NIHR), a public body. Stryker, which makes the robot, met the additional costs of doing robotic surgery within this study. According to the university, the company’s staff were not involved in the analysis or interpretation.

The registry studies. They are observational and include far fewer robotic operations than conventional ones. Patients were followed for two and a half years on average, so the five-year figures rest on the minority followed that long.

Which operation is right for a particular person is assessed by their surgeon.

In a blinded trial of 339 patients, robotic-arm-assisted knee replacement was more precise, slower and costlier than conventional surgery and left patients no better off at one year, with the question of longer-lasting implants still to be answered.