“Patellar resurfacing during TKR associated with long-term cost savings.”
That was the headline. Here is what the data showed: over 20 years, total healthcare costs per patient differed by six pounds.
That is not a cost savings. That is a tie, and it is being marketed as a mandate.
I have performed more than 10,000 joint replacements. I have been publishing on this exact question since 2008, when I wrote an evidence-based analysis in the Journal of Arthroplasty arguing that routinely resurfacing every patella was not supported by the data.¹
When Orthopedics Today asked me to write the invited perspective on this trial, I had about 250 words. This is the longer version.
Credit First
David Murray and his colleagues at Oxford ran the largest and longest randomized trial ever conducted on this question. It enrolled 1,715 patients across 34 centers with 116 surgeons, followed them for 20 years, and was published in The Lancet.²
That is heroic work. Very few research teams have the patience to wait two decades for an answer.
But a trial can be rigorous and still answer a question nobody is asking anymore.
The Experiment Expired Before It Was Published
Patients were enrolled between 1999 and 2003. That is before Facebook existed.
The femoral components in this trial came from an era when trochlear geometry was not designed with the native patella in mind. Modern cruciate-retaining implants are fundamentally different. They have patella-friendly trochlear grooves, better tracking, and enhanced conformity.
Any engineer sees the flaw immediately. The patella does not function in isolation. It rides in the trochlea. If you redesign the groove, you have changed the experiment.
Testing a mating surface and then applying the result to a redesigned mating surface is not evidence. It is extrapolation.
Read the Paper, Not the Press Release
The primary outcome was the Oxford Knee Score. The difference was 0.76 points on a 48-point scale, and it was not statistically significant. No other clinical endpoint reached significance either.
The headline rests on a secondary measure: quality-adjusted life-years. The resurfaced group gained 0.38 more QALYs over 20 years.
But when the researchers reran the analysis assuming equal survival in both groups, the QALY advantage fell by more than half and lost statistical significance.
Let that sink in. A meaningful share of the “quality of life” benefit came from differences in how long patients lived. I have never met a polyethylene button that extends life expectancy.
Then there is follow-up. Of 1,715 patients, only 242 provided 20-year outcome data. With a mean age of 70 at surgery, most of the rest died. That is reality, not a design failure, but it means the 20-year conclusions rest on a small surviving fraction of the original cohort.
The Variable Nobody Measured
With 116 surgeons involved, technique on the unresurfaced patella must have varied enormously. Yet the paper says almost nothing about it, and nothing about lateral facetectomy.
That omission matters. Leaving a patella unresurfaced is not “doing nothing.” In my hands it is a specific procedure: release of the lateral retinacular tissue from the lateral facet, an aggressive lateral facetectomy, and removal of peripheral osteophytes.
We studied 1,053 consecutive knee replacements with unresurfaced patellae and modern patella-friendly components. Only 4 patellae were revised (0.4%), with 98.9% patella-related survivorship out to 8.5 years. Three of those four revisions occurred in knees without a lateral facetectomy.³
Our follow-up work found that facetectomy produced significantly more neutral patellar tilt after surgery.⁴
Comparing resurfacing to a poorly defined alternative tells you nothing about a well-executed alternative.
Now Hold Me to the Same Standard
If I am going to name the limits of Oxford’s study, I owe you mine.
Our data is retrospective, Level IV evidence. It is not a randomized trial, and 8.5 years is not 20. The registry literature also cuts both ways: some AJRR analyses show higher revision rates without resurfacing, while a 2025 analysis of more than 390,000 knees found no increase in short-term revision.⁵ And I consult for and receive royalties from implant companies, which is disclosed with my commentary.⁶
So my conclusion is not “never resurface.” I resurface in patients with inflammatory arthritis like rheumatoid disease, and in laterally maltracking patellae with severe bone-on-bone wear on the lateral facet, provided there is enough bone thickness to do it safely.
What I reject is one word: standard.
Why “Standard of Care” Is a Dangerous Phrase
The authors now recommend resurfacing as standard care for most patients. A co-investigator praised the result because it could help reduce variation in surgical practice.
Reducing variation sounds great in a boardroom. But from an engineer’s perspective, variation is only bad when it is noise. When it is signal, meaning a surgeon responding to the specific patella, bone stock, and tracking in front of them, it is the entire point of having a surgeon.
The United States already resurfaces about 90% of kneecaps, down from nearly 96% a decade ago, as surgeons gained confidence with modern implants.⁷ Resurfacing carries its own risks, including fracture, avascular necrosis, loosening, and extensor mechanism disruption.
A blanket standard built on 1999-era implants would freeze a field that has been moving forward.
The Vintage Data Test
Before any study changes how you operate, run it through three questions:
Were the implants and techniques tested the ones I use today? If not, the result is history, not guidance.
Is the primary outcome significant, or is the headline built on a secondary one? Press releases almost always lead with the second.
Was the comparison group treated well, or just treated differently? A poorly defined control arm will lose to anything.
What You Do Now
To surgeons and residents: read the paper, not the headline. Then run the test. A 20-year randomized trial deserves respect, but it does not deserve automatic obedience.
To patients: ask your surgeon one question before knee replacement. “Do you resurface every kneecap, or do you decide case by case, and why?” A good surgeon will have a clear answer.
To everyone generating evidence, including me: publish modern, prospective data with modern implants. That is the only way this debate ends.
Evidence is only as current as the implant it was tested on.
What do YOU think?
Footnotes
Meneghini RM. Should the patella be resurfaced in primary total knee arthroplasty? An evidence-based analysis. J Arthroplasty. 2008;23(7 Suppl):11-14.
Murray DW, Hudson J, Dakin H, et al. Patellar resurfacing in total knee replacement: 20-year clinical and economic results of a large multicentre, randomised controlled trial in the UK. Lancet. 2026;408(10549):52-61.
Gunderson ZJ, Luster TG, Deckard ER, Meneghini RM. The fate of unresurfaced patellae in contemporary total knee arthroplasty: early to midterm results. J Arthroplasty. 2024;39(8):S65-S69.
Gunderson ZJ, et al. Patellofemoral OA severity is not correlated to PROMs in TKA with selective patella resurfacing. AAOS Annual Meeting, 2025.
Leaving the patella unresurfaced does not increase the risk of short-term revision following total knee arthroplasty: an analysis from the American Joint Replacement Registry. J Knee Surg. 2025;38(3):122-129.
Orthopedics Today / Healio. Patellar resurfacing during TKR associated with long-term cost savings. October 23, 2026.
Sloan M, Donnelly P, Chen AF. Trends and outcomes comparing resurfaced to un-resurfaced TKA patients in the AJRR, 2012-2021. AAOS Annual Meeting, 2023.
