I'm on Ozempic and I Need a Knee Replacement: What the Evidence Actually Says

Roughly two out of every five patients I evaluate for knee arthritis are also being treated for obesity, type 2 diabetes, or both. A large share of them are now on a GLP-1 medication prescribed by their primary care physician or endocrinologist. That creates two practical questions that arrive in clinic almost every week: Could this drug help me avoid the operation? and Do I have to stop it before surgery?

Both questions now have real evidence behind them. Neither has a clean, one-word answer.

On avoiding the operation: one year of exposure to any GLP-1 agent was associated with a 2.8 percentage point lower cumulative rate of knee replacement at eight years (hazard ratio 0.90, 95% CI 0.83–0.98). Three years of exposure to semaglutide or tirzepatide specifically was associated with a 4.7 percentage point reduction (hazard ratio 0.72, 95% CI 0.67–0.78). Longer exposure and newer agents both tracked with larger effects.¹

Put those percentages in terms you can use. A 4.7 percentage point absolute reduction means roughly 21 patients would need to be treated for three years to prevent one knee replacement over eight years. That is a meaningful population-level effect. It is not a promise to any individual person that the medication will keep them out of the operating room.

On surgical safety: patients who started semaglutide less than a month before surgery had markedly lower odds of minor adverse events (odds ratio 0.16). Reduced odds of severe adverse events reached statistical significance once patients had at least two to three months of exposure before the operation (odds ratio 0.25). The authors' conclusion was that about three months of preoperative semaglutide appears sufficient to reduce both categories.²

And on the aspiration worry that drove a lot of early anxiety about these drugs and anesthesia: a 2025 single-institution series of 83 patients on consistent GLP-1 therapy going into total joint replacement found no cases of intraoperative pulmonary aspiration. One patient — in the dose-escalation phase — was found to have a full stomach requiring decompression.³

• None of this is randomized. Every number above comes from observational data. Patients who fill GLP-1 prescriptions and stay on them for three years differ systematically from those who don't — in insurance coverage, in engagement with medical care, in the ability to tolerate the side effects, in baseline motivation to manage their health. Propensity matching can only balance what the database recorded.
• The authors' most interesting claim is also their least supported. Carter and colleagues suggest the effect may reflect "potential disease modifying activity beyond weight loss alone" — that is, the drug acting on joint inflammation directly rather than just taking load off the knee. That is a genuinely interesting hypothesis. This study design cannot test it. The database does not record how much weight each patient lost, so weight loss cannot be separated out as the mechanism.
• "Fewer knee replacements" is not the same as "less arthritis." A knee replacement is a decision, not a biological event. Patients on GLP-1 drugs may be having fewer operations partly because they feel better, partly because they weigh less and load the joint less — and partly because surgeons and patients defer surgery while a weight-loss plan is underway. Some of the measured effect is likely delay rather than prevention. Eight years of follow-up cannot distinguish those.
• The safety data are strongest where the confounding is worst. An odds ratio of 0.16 for minor complications in patients on the drug for less than one month is implausibly large as a biological effect. A month of semaglutide does not remodel a patient's surgical risk profile that dramatically. The far more likely explanation is that patients who were recently started on semaglutide were healthier, better optimized, and more closely followed to begin with.
• There is a musculoskeletal safety signal that deserves watching. An abstract presented at the 2026 AAOS Annual Meeting, using matched electronic medical record cohorts of about 73,000 patients per group, reported higher five-year rates of osteoporosis (4.1% vs. 3.2%), gout (7.4% vs. 6.6%), and osteomalacia (2% vs. 0.1%) among GLP-1 users with type 2 diabetes and obesity.⁴ That work is a conference abstract, not a peer-reviewed publication, and rapid weight loss from any cause carries bone-density risk. It is not a reason to avoid these medications. It is a reason for anyone on one long-term to have bone health on their physician's list.

When a patient comes in with knee arthritis and is already on a GLP-1 medication, my read is that this is favorable — for the arthritis, for the operation if we get there, and for the recovery afterward. Weight is one of the few genuinely modifiable factors in both the progression of knee osteoarthritis and the risk profile of knee replacement surgery. A patient who has lost meaningful weight before surgery is a patient with lower infection risk, easier exposure in the operating room, and less load on the implant for the life of the reconstruction.

Where the evidence likely fits. A patient in their fifties or sixties with mild-to-moderate radiographic arthritis, elevated BMI, symptoms that are real but not yet dominating daily life, and an appetite for trying non-operative measures first. For that patient, evidence suggests a GLP-1 medication — prescribed and monitored by the appropriate physician — is a reasonable part of a plan that also includes structured physical therapy, activity modification, and an injection trial. Some of those patients will not need me to operate for years, and some may not need it at all.

Where it likely does not fit. A patient with bone-on-bone tricompartmental arthritis, significant deformity, night pain, and a knee that has stopped responding to anything. Starting a weight-loss medication does not regrow cartilage that is gone. I have seen patients delay a needed operation for a year or more waiting for a medication to resolve mechanical, end-stage disease, and arrive with more stiffness and more deconditioning than they started with. If you are already at that stage, the medication is worth pursuing alongside planning the operation — not instead of it.

I do not prescribe these medications. My role is to read the imaging, tell a patient where they actually sit on the arthritis spectrum, and coordinate with the physician managing the medication so that the timing of surgery and the timing of weight loss work together rather than against each other.