Yes, and the placebo arm is the part to read first. STEP 9 randomized 407 people with obesity and moderate knee osteoarthritis to semaglutide 2.4 mg or placebo for 68 weeks, both arms receiving counseling. Pain improved in both. The knee-osteoarthritis phenotype analysis examines whether the response varies between patient groups rather than treating every painful knee as the same trial participant.
Knee replacement, the harder outcome, has three database studies behind it, and they do not agree. One followed people who already had knee osteoarthritis out to eight years. Three years of semaglutide or tirzepatide gave an absolute risk difference of −4.71 percentage points [2], which is 4.71 fewer operations per hundred people.
The other two found no gap in knee replacement. A five-year cohort started with people who had no hip or knee osteoarthritis [3]. Among diabetic patients with obesity, GLP-1 users were diagnosed with knee osteoarthritis more often, a hazard ratio of 1.52. They also had more joint injections, and knee replacement rates did not differ. A second cohort matched 452,445 adults with type 2 diabetes per arm against SGLT2 inhibitor users [4]. Knee osteoarthritis was less common on the SGLT2 drugs, a hazard ratio of 0.831. Knee replacement again did not differ.
None of the three was randomized, and each compared a different group over a different window. The eight-year figure is one reading of the record, not a settled effect on surgery.
If the operation happens anyway, the two molecules look alike, though not on evidence strong enough to say so. Across 415 matched pairs no outcome differed significantly[5]. The surgical complication interval ran from 0.845 to 3.152, consistent with one drug being slightly better and with it tripling the risk. That is no difference found rather than no difference.
Knee osteoarthritis is one of the reasons people want to lose weight rather than a side issue, and STEP 9 is the trial that measured it directly. It randomized 407 people with obesity and moderate knee osteoarthritis two to one, semaglutide 2.4 mg against placebo for 68 weeks. Both arms received counseling on physical activity and a reduced-calorie diet[1]. The headline number is large. The number that describes the drug is smaller, and it is the one worth carrying into any decision about what to pay for it.
Both arms improved
Mean WOMAC pain score fell by 41.7 points on semaglutide and by 27.5 points on placebo (P < 0.001). The scale runs 0 to 100 with higher scores worse, and mean baseline pain was 70.9. So the placebo arm — which was not an inert arm, since it included diet and activity counseling — accounted for roughly two thirds of the improvement seen on the drug.
The drug’s own contribution is the 14.2-point difference. That is a real and meaningful effect on a 100-point pain scale. It is not 41.7, and a page that quotes 41.7 without its comparator is quoting the trial rather than reporting it.
The weight figures
Body weight fell 13.7 percent on semaglutide against 3.2 percent on placebo. Physical function on the SF-36 improved by 12.0 points against 6.5. Serious adverse events were similar between groups. Discontinuation for adverse events ran 6.7 percent on the drug against 3.0 on placebo, gastrointestinal problems being the most common reason — which matches what the titration literature reports elsewhere.
What it means for a purchase
Two things. The trial used 2.4 mg semaglutide. That is the weight-management dose most compounded sellers titrate toward, so the dose here does transfer in a way the kidney trial’s 1.0 mg does not. The comparator matters too. The trial is evidence for the drug plus diet and activity counseling, not for the drug alone, and almost no telehealth seller includes structured counseling in what it sells.
If pain relief is the reason for starting, that is worth asking a seller about directly. Most publish nothing about what support comes with the prescription, which is the gap we count in the questions they leave open.