Is robotic arm-assisted medial UKA more accurate than conventional jig-based UKA?

Listen to Andrew Duckworth, Sam Oussedik and Fares Haddad discuss the paper 'A prospective double-blinded randomized controlled trial comparing conventional jig-based versus robotic arm-assisted medial unicompartmental knee arthroplasty' published in the September 2026 issue of The Bone & Joint Journal.

Click here to read the paper.

Find out as soon as the next episode is live by following us on X, Instagram, LinkedIn, Tik Tok or Facebook!

Transcript

Copy Transcript Copied

[00:00:00] Welcome everyone to our BJJ Podcast Series. I'm Andrew Duckworth, and a warm welcome back to you all from your team here at The Bone & Joint Journal. As always, we'd like to start by thanking you all for your continued comments and support, as well as expressing a big gratitude to our many authors and colleagues who take part in our series

that highlights just some of the great work published by our authors each month. So for today's podcast, we have the pleasure of being joined by two of the authors from an RCT published in the Septem- September edition of the BJJ, reporting on the results of a prospective randomised control trial comparing conventional jig-based versus robotic arm-assisted medial uni- knee arthroplasty, which is from the team at UCL.

So firstly, I have the pleasure of being joined by my editorial board colleague at the Journal, specialty editor for the knee, Mr Sam Oussedik. Sam, thanks so much for joining us. Hi, how are you? Pleasure to be here. Thanks, Sam. So joining Sam, we're delighted to welcome back our awesome editor-in-chief at the BJJ, Professor Fares Haddad.

Prof, great to have you back with us as always. Dux, it's great to be here and thank you for doing this. So Prof, maybe kick off with yourself. Prior to this study, can give us a quick overview of where the literature really currently sits with regards this [00:01:00] topic and where the debate currently stands?

 So I think there's two elements to this. The, the first is the usage and the outcomes of unicompartmental arthroplasty, and the fact that we're probably worldwide underusing it as an operation. So there is the, this whole debate about whether a patient is a uni versus a total knee arthroplasty candidate, and we can bounce around the indications and people's thresholds for going one way or the other.

And that varies country by country, it varies institution by institution, surgeon by surgeon. But at, at, at a very basic level, many of us believe that it's unicompartmental arthroplasty is a lower morbidity, lower mortality operation. It's an easier recovery, a lesser hit for the patient, and by preserving the ACL, that you've got a, a higher functional level.

... But if you look at studies like TOPCAT, which have compared uni and total, the... when you look at PROMs, the outcomes haven't been that dramatically different. So- ... we still think there [00:02:00] should be a focus on UKA. Now robotic- ... UKA has come along, and Mark Blyth started very early on, before really there was widespread interest, doing his studies, and we've published in the journal the ten-year data- ... from his randomised study. So I think there is some evidence that in his hands, and he's a very good surgeon, even his control arm did very well that it goes nicely. So the, the idea now that you know, computer-assisted surgery, particularly CT-based planning with a robotic arm, is getting more and more established, we thought it was worthwhile looking at what is the standard in the UK?

You know- ... what's the majority doing, and if we brought in this technology, will it make a difference? And this is in one centre, but it's in the generality of surgeons in that centre. This is not just Sam or myself. This is fellows, trainees, et cetera, under supervision, but everybody- using this technology to do unis. And at the [00:03:00] time when we set out to do the study, the most popular uni in the UK was the Oxford. ... Very pushed and popularised by its developers and, well known worldwide. Some, some great data. And the robotic uni that we had access to was the Restoris which is the one that's been used in this study with the Mako robotic arm.

So we thought there was time to... It was time to do a study comparing the Oxford, which we've used for a very long time in our institution, over 20 years, with the Mako Restoris, which allowed us to introduce that technology and then compare how we were doing one versus the other. Probably I'd add one more thing, is that compared to the Blyth study, which kind of had done something very similar, our patients were gonna be blinded in the sense that we were creating sham incisions for the pins and therefore patients did not know what they were getting from that [00:04:00] point of view.

That, that's great Prof and a, a nice overview of where we were at. And is it... Sam, is there anything you would add to that, not only your own clinical experience, but also your way... what comes across your desk as a specialty editor for knee? Is there anything that you would want to add? So I think across arthroplasty and, we'll focus on knee here, but total knee, compartmental knee, there, there is a, a real interest in what technology can add.

So there are lots of papers coming, lots of studies being carried out to see whether the initial outlay that's involved for investing in technology has a positive outcome for our patients. I think, industry clearly technology. I think surgeons like technology. I think we are struggling a little bit to show patient differences in these groups.

So we know about perhaps differences in precision. We are struggling probably a little bit to show differences in our patient outcomes. And just to add to that, the, the thing with partial joint replacement is yes, there may be a functional advantage, but if we look at joint registries, traditionally there has been a cost to that, and that cost has been a higher revision rate.

And so [00:05:00] that's really the gap that we'd be looking if possible, for technology to help us with, is to bring the revision rate of partial joint replacement down to a level that's similar to total joint replacement. I think if we could fill that gap, then that would be, a, a real advantage.

That's not necessarily what we're looking at in this study but that may be data to come later down the track. No, absolutely, Sam, and I think that you raised some points there which I think are really important in terms of outcomes and things we've mentioned before on previous podcasts, and I w- I'll probably come back to that actually after we've discussed your study if I may.

So Prof, can maybe come back to yourself, about the trial that you guys have done. So this study reports the interim outcomes of an RCT comparing the accuracy of component position, early functional outcomes, patient-reported outcome measures as well as complications w- comparing conventional jig-based uni with navigational control versus robotic arm assisted uni for patients with medial compartment OA.

So for our listeners, Prof, just briefly, what any sort of key inclusion, exclusion criteria you'd want to highlight for them? Yeah, I think we should say, first of all, this is just a medial uni study because- ... actually, lateral [00:06:00] unis, patellofemorals, et cetera, are all part of a separate conversation.

But I think we tried to go with what we would call the standard well popularised in the UK indications of anteromedial osteoarthritis, less than degre- ten degrees of fixed flexion, correctable varus. We di- we didn't really want anything controversial, a range of motion of at least ninety degrees.

So we wanted those- ... standard knees that patients that most surgeons would agree were suitable for a uni. Yeah. We didn't go for anything controversial in this particular dataset. No, and we excluded anybody who had an inflammatory disorder, anybody who had a previous high tibial osteotomy, anybody who had an ACL reconstruction et cetera.

So it was the kind of- Yeah ... typical sort of scenario you would imagine. The patellofemoral joint always causes great anxiety. And as long as basically they weren't bone on bone in the lateral patellofemoral compartment, we were happy to do a uni. But ultimately, it was a surgical decision at the [00:07:00] time for the surgeon doing it.

If- Yeah ... if they thought the patellofemoral joint was so badly damaged that they weren't comfortable with the uni, they could convert using the same CT-based plan, convert to a total knee if they if they so wished. Yeah. Absolutely. And- probably the other thing to, the other thing to add, which we partly hi- hinted at, is the whole premise here is that the reason for the failures and the revisions that Sam was describing could be technical, which- computer assisted would help, and could also be overcorrection that allows to a progression of osteoarthritis, which is why we thought that actually the accuracy and precision of robotics might make a difference here. Yeah. Yeah, absolutely, Prof. And like you say, fairly standard sort of inclusion, exclusion criteria and consistent with what a lot of people would expect.

And before we move on, anything particular about the-- You've obviously men- obviously mentioned the interventions, but anything particular about the interventions or pathways you'd want to highlight for the two arms? Only to say that, th-this was a unit and a group [00:08:00] of surgeons who were very used to using the the control prosthesis, the equipment, the sort of workflows, to that, we just added- Yeah ... some pins, but essentially it didn't change otherwise the workflow that was used for doing a standard Oxford mobile bearing uni, and everybody had been through... we'd managed that early learning curve and were well through it. We... In, in our centre, we'd been using the robotic arm for unis, totals, total hips.

It's kind of part of the practice, been part of the furniture- Yeah ... for a while. Yeah. No, absolutely. I think that's important for them, for our listeners to know. And Sam, if I can come to yourself, in terms of, how the patients were followed up and what the outcome measures were, what's the key thing, things you'd want to highlight from that?

Yep. The, the primary outcome in this case was accuracy measured on a CT. And so looking at previous data we were able to power this to detect a one degree between group difference. And when we feed those numbers in, we end up with a group size of 63. So we tried to recruit 70 to each group to a-account for a little bit of attrition, as you might.

Secondary [00:09:00] outcomes are probably important to mention at this stage as well. So we can group those into the inpatient and the sort of follow-up outcomes. We had- ... a group of inpatient outcomes which were pain, analgesia use, physio sessions, stair use and length of stay. And then on top of that six-, 12-, 24-month outcomes, which were a standard array of PROMs, Oxford Knee, KOOS, Short Form 12s Forgotten Joint Score and WOMAC.

Yeah we... Do you want me to go on to the patient patient recruitment? Yeah. We'll maybe come onto that in a minute, Sam. Okay. I think you've just covered the outcomes that you looked at pretty well there. But just to highlight what you sugge- said there in terms of the, the plan was to recruit 70 patients per arm and 140 in total, and I think we'll come onto that maybe in just a second in terms of what you actually recruited.

So if we move on to the key findings, there were... In the actual study, there was 107 p-patients recruited in the final study cohort, 52 in the conventional group and 55 in the robotic group. So Sam, could you give us those key results and findings from those analyses you reme- mentioned?

Also [00:10:00] what we just talked about there, about, about recruitment and how that sort of progressed as the trial went on. So I think it's fair to say that recruitment became more problematic as we went along, and that was for two reasons. I think o-one is that we ap- actually approached 249 patients in, in the study period which was what, four and a half years.

And 86 declared they would only have a robotic procedure, which makes it then very difficult to randomise them, as you'll appreciate. But the other thing that happened over the period was that the, the surgeons lost equipoise, and as a result, it was very difficult to recruit to the non-robotic arm which is why we had to finish up a little bit early before too much time passed.

To get onto the results whi-which you asked about there. Better accuracy we saw across all parameters for the implant placement. I think one might expect that. This is a, a, a, a very precise tool we're using to achieve a CT-based plan. Clearly that is usually more achievable with less transformation error when we use a robotic arm rather than manual instruments.

I think one might [00:11:00] expect that. The, the surgery took a little bit longer in the robotic group which probably reflects differences in the surgical workflow. There are a few extra steps in the robotic procedure, setting it all up. That kind of thing can take a little bit longer. What was interesting was perhaps echoing some of the findings from previous studies is the robotic group had less pain during the early post-operative period both on, on the visual analog scores and opioid usage.

That's a finding that we've seen previously, and we can maybe guess as to what the reasons might be there. But also less physio required and a slightly shorter length of stay. So for a procedure which is already a little bit less invasive than what might be an alternative for many that are total knee replacement, we're seeing small incremental improvements with the robotic arm on that inpatient experience.

In terms of the outcome measures we looked at then in the follow-up period the, the KOOS was a little bit better at six months in the robotic group, but otherwise the other measures were pretty much even, except the Forgotten Joint Score, which was better at all time points [00:12:00] actually and better by a mean of around six points at the six, 12, and 24 follow-up point marks.

It doesn't seem- ...to be tailing off either. It seems to have maintained that difference between the groups. So- Yeah I guess in summary, a slightly faster and easier recovery in the robotic group and the, the f-main functional outcomes look a little bit similar but the Forgotten Joint Score are better in the robotic group.

No, that's a really nice summary, Simon. I think those are the things like you said to me is that persistence of that difference for the FJS that you highlight. But also those-- I know it's a relatively small numbers, but in that Table 3 there, the differences in that immediate sort of postoperative pain, opioid consumption, just getting them up and about and out is quite consistent across the board there, isn't it, in terms of favour of robotics.

So Prof, maybe if I could come to yourself, back to yourself now, in terms of, how you feel we bring this all together. The take... What's your you feel the take home message of the trial is? And I suppose caveated by any sort of strengths and limitations you'd want to highlight.

 And I think this backs up previous data, both data from Mark Blyth's work and long-term [00:13:00] data from Andy Pearle and others showing that robotic-assisted unis do very well. So I think this supports the continued use and evaluation of this procedure and for us probably it's become the way to do it because you can plan three-dimensionally and execute as a resurfacing.

It works very well in the lateral compartment as well as the medial compartment. I think I, I'd probably add a few things just to show the limitations. I think the first one is we recruited over a long period because of COVID partly, so I think everyone needs to be aware of that. And s- and Sam's highlighted the fact we didn't hit our target numbers, but that, that really was loss of equipoise.

And this was, i- interestingly, this was illustrated by NIHR when the RACER study started. It was suggested that RACER shouldn't run in centres that offer robotics for anyone except patients in the study, and that's- that's great if you're recruiting. It's not great in terms of experience and training the team, and so on.

So for us, because you could access it that was a problem. A couple of things to add. [00:14:00] O- one is that this was a mobile-bearing prosthesis we were comparing against that- ... p- people will be concerned by that because fixed-bearing prostheses seem to generally be doing better long term in terms of revisions.

But the data for the Oxford is great. It's out there. It was the most popular in the UK at the time. I don't think that takes away from the study significantly, but there'll be people who make some noise about that. And then the, the other one is something I've tried to express as clearly as I can over the last few years, which is that MCID as a measure in this kind of study to measure a difference is unrealistic.

We're dealing with interventions that have a huge effect size, and to hit more than a third of that effect size again to show a difference is almost impossible. You'd have to do something really transformative- Yeah ... to do that. So the fact actually we're seeing a difference of six points in Forgotten Joint Score for me is a signal.

It's a, it's a signal we take seriously. Yes, it does not reach so-called MCID, but I think beware the use of [00:15:00] MCID in this setting. That's something we need to caveat. So yes, we haven't got massive numbers, but I think the data speaks for itself. That certainly if I were going to have my unis tomorrow, I'd want someone to plan it on a CT scan and, Yeah

do it do it using this kind of technology. Yeah, thanks, Fares. And, Sam, anything you add to that, and I suppose something I want you to comment as well is maybe just, this- it's the thing that comes up an- over and over again is the sort of cost aspects of this in terms of, people say, "Oh, the robot's expensive, so what is it?

Can it really be cost effective in the end?" Yeah. It's a good question 'cause it, it does involve quite an investment if you're gonna move over to this sort of precision technology a- and wh- whichever type you use, quite frankly. W- what are the advantages that might outweigh that?

I think we're seeing across the piece a slightly quicker recovery. Now, clearly, that's multifactorial. It depends a little bit on your unit. We know of many units across the country or some anyway who are simply doing every joint replacement as a day case so how much incremental gain can you get from introducing technology?

I think the message that comes across from those studies as a [00:16:00] whole is that it makes the patient experience a little bit easier, and I value that as a clinician. I value the fact that my patients recover a little bit faster, find the process of having a joint replacement easier because that's often the thing that scares people off, let's face it.

And the ultimate results from joint replacement are so transformative to people's lives that not going through with it because you're worried about the process is a significant barrier to accessing care, in my view. We've seen better accuracy apro- across the piece. Now I think what we still need to do is see the signal that comes through to long-term outcomes and revision because ultimately it's reducing the revision burden, particularly in uni compartmental knee replacements, we said at the top, that's likely to make this more cost-effective and obviously prevent additional procedures for our patients and protect them from the harm associated with that.

There, there are small signals from the NJR, for example. If we look at the two prostheses we compared here, there are small differences in the revision rate at five years already s- visible on, on the NJR. How that pulls through to ten years and whether this is [00:17:00] the best way of achieving those improvements in revision rate to a large cohort I think, probably is a little bit beyond the remit of what we were looking at here.

But, a- as Fares said I certainly, my patients feel the benefit of it, and again if I were to have a procedure, I would have it using something similar to this. I think that's very well said, Sam. Sorry, sorry, Fares. It's worth saying also that there are some kind of knock-on discussions that were taking place in the UK around you have to do a minimum number of unis before you can do a uni.

You have to convert X proportion of your practice to uni before you do uni. And actually, the fact that this tool, this technology allows you to deliver a resurfacing, to have a plan and execute that plan without necessarily doing that number may actually then afford a bigger group of patients the opportunity to have this smaller operation.

As Sam says, the real proof down the line will be if we get fewer revisions. But I think what technology does, it changes the [00:18:00] philosophy slightly in terms of what you can and cannot deliver for patients, rather than defaulting to what you're used to doing day in and day out, which is a standard total knee.

It used to be, a lot of people just doing a standard PS total knee, and suddenly we're cruciate preserving, we're doing unis, we're going cementless. That's... I'm talking very much in generalisations. But I think that's that, that could be a big difference. And then the other facet is, I think, surgeon comfort.

I know we're primarily based around patient outcomes, but actually the surgeon knowing that this operation's gonna be less stressful, less demanding, and not worrying about what that x-ray's gonna look like, because those uni x-rays all look the same in the intervention arm. I can tell you in the control arm, they don't all look the same.

So- yeah ... our NPT's always interesting. Yeah. No, sorry, Sam, yeah. No, guys just to add to that it's quite easy also to have a sort of UK-centric view of this and particularly with uni usage. We're a mature market, inverted commas for uni usage. As, as I go round the world a- and meet surgeons from different parts of the world, [00:19:00] you realize that unis are still quite experimental to some people a- and trying to convince them of some of the benefits often you come up against this yes, but the revision rate is greater.

A- and so I think this goes some way perhaps of spreading the word that actually there are a group, there is a group of patients who may benefit from a, a less invasive procedure and still get very good outcomes. And I was talking to a surgeon last week who was very proud of moving his uni percentage from 2% to 4% over the last three years.

And and I've been around 30% forever, quite frankly and amongst my colleagues, that's probably, about average, maybe a bit low for some. So it's just a different mindset, as Fares says and it opens up that technology and this procedure to a group of patients who might not otherwise benefit from it for a whole bunch of reasons.

No, that's very well said, Sam. And maybe just, maybe, Fares, if I can just finish up with your co-thought, thoughts on this. W- as we're all at the Journal, we all believe in, high quality research. But if we... Is it gonna be a case in the line of robotics, unless we find something that [00:20:00] we can show a difference with in terms of the patient board outcome, as you talked about at the beginning, about our issue with trying to find that difference, and also we've, we know that RACER has come out and what it's shown.

Our ultimate though, people go, surgeons and patients are gonna vote with their feet and say, "Actually, we..." Because all I ever hear from you guys, as somebody who doesn't do n- replacement surgery is, it's definitely better. And from what your trial shows is the patients felt it was definitely better because they voted with their feet as well.

Is that where we're gonna end up with this in the end, do you think? I think it's where it's gone, and confirmation bias is always going to affect how you look at things. Yeah. Look the, the reality is these tools, these technologies deliver precision. Yeah. It's difficult to believe that precision is bad.

So it comes down to how you use the technology, and I think that may be the really critical thing if you're gonna look at big pragmatic randomized studies. And you could, RACER is done by a great group, you know- ... re- really good structure- Absolutely ... around their trials [00:21:00] and so on. But actually, the minute you make it pragmatic and let anyone choose any, CRPS, patella resurfacing, no patella resurfacing, you put it in this way, that way, or any other way, that creates a whole load of noise around what you're looking at.

It's great to have people studying it. It's great to have big randomised studies. But at the end of the day, it may just come down to making us think a little bit harder about how it is that we're going to use this technology rather than whether we are going to use it. Absolutely. No, I think that's a very good point to finish on.

So well, both of you, I'm afraid that's all we have time for today, and thank you so much to both of you taking time to join us. A great discussion and really a great study that's obviously clearly added to the literature in the area. And to our listeners, we do hope you've enjoyed joining us, and we do encourage you to all share your thoughts and comments on the various platforms or alike.

Feel free to post about anything we have discussed here today, and thanks again for joining us. Take care, everyone.