Ozempic works for arthritis, and probably other pain (Member Post)

The study basics

Bliddal H, Bays H, Czernichow S, et al. Once-Weekly Semaglutide in Persons with Obesity and Knee Osteoarthritis. N Engl J Med. 2024 Oct;391(17):1573–1583. PubMed 39476339 ❐

Bliddal et al. conducted a randomized, double-blind trial, so neither participants nor researchers knew who was receiving the drug versus a placebo — over 400 adults with both obesity (BMI ≥30) and knee osteoarthritis (radiographically confirmed and quite painful). They got weekly injections of either Ozempic or placebo for 68 weeks, tracking two main outcomes: body weight and pain (WOMAC pain score, a well-established scoring system for knee osteoarthritis).

The Ozempic-takers (271 subjects) lost an average of about 10% more weight than those on placebo (136) for a total of 13.7% — a remarkable result in itself, regardless of any effect on arthritic knees. That is quite a lot! When you lose that much fat, you likely also lose visceral fat, which is usually the problem fat (this point explored in detail in a recent post).

Both groups got counseling on a reduced-calorie diet and exercise, so that semaglutide effect is on top of lifestyle advice, not instead of it. In other words, they lost 10% more than people who were trying — which is much more impressive than losing 10% more weight than people who aren’t trying.

The refreshingly excellent results

There was more pain relief with Ozempic than placebo, a respectable improvement in WOMAC pain scores of -41 versus -27 for placebo, a solid difference of 14 out of 100. The improvement in function scores was similar. This might not sound like a lot, but it easily clears most published thresholds for clinically meaningful improvement, especially in a population starting with high baseline pain.1

“Treatment with semaglutide resulted in greater improvements than placebo across all pain-related end points, a finding that is in line with those from an observational study involving adults with knee osteoarthritis and type 2 diabetes…”

So this trial more rigorously confirmed what had previously been one of those iffy “just a correlation” results, which famously “does not imply causation” … but probably did in this case! Bliddal et al.’s experiment strongly suggests that the reduced arthritis symptoms really are being driven by Ozempic, rather than just being some kind of Ozempic-flattering coincidence.

There are limitations, of course, most notably the narrow scope, short duration, and being paid for by Ozempic’s manufacturer.2 But nothing serious. The design is good enough to bias-proof the study quite well, and we also know that these results are consistent with a lot of other evidence.

How does Ozempic treat arthritis?

Weight loss is already known to be a good investment in managing arthritis, and Ozempic is really good at helping people lose weight, and this study mainly just connects those dots. But how does it work? It’s probably not just the reduced load on the knees. It probably is mainly a metabolic effect. Asaf Weisman:

“Belly fat is active tissue that secretes adipokines and other pro inflammatory agents which contribute to epigenetic changes that accelerate tissue degeneration.”

To be clear, that statement is not directly supported by this study. In fact, Bliddal et al explicitly say that “mechanistic conclusions cannot be drawn.” But they do cite preclinical data suggesting possible anti-inflammatory and anti-degradative effects of this class of drugs.

And we can speculate. We’ve had many reasons to suspect that “it’s the metabolism, stupid” for a long time, but this is a particularly high quality placebo-controlled trial of an effective weight loss drug in humans, and it delivers strong evidence that many people will hurt less if they lose weight, and that result was very predictable based on the metabolic argument.

It’s also quite unlikely that modestly reduced load alone would make so much difference, so quickly. It’s not like there’s suddenly no load on the knees!

This has profound implications! The GLP-1RAs will probably prove helpful for anything painful that is powered by metabolic syndrome and “inflammaging,” the escalation of systemic inflammation and its complications as we age (see Chronic, Subtle, Systemic Inflammation), the collective consequence of poor fitness, obesity, aging, genetics, and hard living. It is the roots of cardiovascular disease and diabetes, which in turn are major risk factors for many kinds of chronic pain … and these drugs clearly treat metabolic syndrome effectively. Cool cool cool! [YouTube].

How clever is that logo? Making the dot on the lowercase ‘i’ look like a head over a skinny body, representing weight loss? That’s some smart design work there.

Does a little bad safety news from Xie et al. nuke the good news from Bliddal et al.?

Xie et al is a truly massive study of almost a quarter million diabetic U.S. veterans taking any drug in this class.3 They were looking for a broad range of effects on biology, good and bad, large or small. It gives us even more evidence of broad health benefits for dozens of conditions — and that harmonizes strongly with the overall good news about the health benefits of these drugs.

But I am contractually obliged to confess that they also found something that seems at odds with Bliddal et al.’s good news, and with the important “it’s the metabolism” point of this post. They actually found small increases in several musculoskeletal problems! Arthritis, joint and bone pain, and tendinitis! Ruh roh! That looks like a direct contradiction: one study says “less pain,” the other says “more!” What are we to make of this?!

I had a bad moment when I first noticed this. Thought I was going to have to re-write the whole damned post! But no, it’s not that bad — just a significant and worthwhile extra section.

So what gives? These studies measured different things, in different ways, in different populations:

• Bliddal asks: “If you already have knee OA and obesity, does high-dose semaglutide help your knee pain?” Answer: yes, very specifically and clearly.
• Xie asks: “Across millions of clinic visits, are people on any similar drug more likely to get non-specific musculoskeletal diagnostic codes?” Answer: only slightly, and the details are not in focus.4

They harmonize much more than they clash. Xie et al. does not undermine the basic idea that losing weight tends to help arthritis, and meanwhile it reinforces the more overall good news about Ozempic and its whole family of drugs.

What could possibly go wrong? There are some concerns

In terms of side effects, taking Ozempic mostly went fine: 10% of the experimental group had some kind of medical trouble, compared to 8% on placebo, but those incidents were probably mostly unrelated to the drug.5 There were some predictable side effects, mainly nausea and diarrhea, which caused 6.7% to stop taking Ozempic, versus 3% for placebo — not nothing, but not bad.

Immediate side effects are not the only concern. I can easily imagine bigger picture problems! Drugs do not exist in a vacuum, and GLP-1RAs may work in theory (“efficacy”) and still fail in the real world over time (“effectiveness”). Ironically, they may be “ineffective” precisely because they are so efficacious, because rapid weight loss has risks, like lost muscle mass and bone density, and eventually an increase in falls and fractures. And that would only be exacerbated by rapidly regaining weight after stopping the drug.

And so good medical supervision is critical, but unfortunately we cannot count on this. Indeed, many people are taking these drugs without any oversight at all, or incompetent oversight. Abuse and neglect may already be widespread. I have seen this happen to a family member, so this is not a hypothetical scenario to me.

This kind of evidence can and probably will inspire excessive and misplaced blaming and shaming. As my last post on this topic strongly emphasized, it is definitely possible to be heavy and healthy. People who do not need to lose weight will be hectored as if they do. And many of those who do will be pushed too hard, with too much moralizing. It’s now clear that taking Ozempic is somewhat stigmatized. The attitude that it’s “cheating” and “lazy” appears to be widespread on social media, and some of that is coming from healthcare professionals who should know better.

So studies like Bliddal et al. are bound to inflame all that.

Losing weight is one of the most easier-said-than-done problems in life. No one should take on a challenge like that without clear cause, or without plenty of compassionate and competent support.

But these issues are all about abuses of the drug. The evidence still strongly suggests that weight loss is a valuable investment in fighting at least some major kinds of future pain for many people, and we have strong reasons to believe it’s a metabolic effect rather than a mechanical one, and probably isn’t limited to arthritis.

If that is actually how things work, then we must just forge ahead as best we can, and start to think of weight loss as an important tonic for many kinds of chronic pain … and Ozempic as a way to make it possible for many millions of people, and much easier for many more.

Notes

1. Many osteoarthritis trials struggle to achieve between-group differences of even 8 points. A 14-point difference is not just clinically significant, but by a healthy and easily felt margin. In osteoarthritis research, the minimal clinically important difference (MCID) for WOMAC pain is 10–12 points on the 0–100 normalized scale. Some analyses, especially in more symptomatic populations, place the threshold a bit higher, closer to 15–20 points for a change that patients clearly perceive as important. The semaglutide–placebo difference sits squarely within this range.

   People in this trial had fairly intense knee pain, 71/100, and MCID estimates tend to skew slightly upward in cohorts with high baseline severity — making a 14-point gap even more impressive.

   Finally, this wasn’t just a mean difference: the whole distribution shifted substantially, nearly all semaglutide participants improved, and many improved a lot. That graph shape strongly supports that the effect is not only statistically significant but meaningful in lived experience.

2. All the limitations, for the record:

   • Narrow scope. This is a study only of knee osteoarthritis in people with obesity.
   • Mechanism unknown. The study doesn’t show how this works (e.g. metabolic vs. load-reduction).
   • No body-composition data. Total weight changed, but nice-to-have data on visceral fat, biomarkers, and inflammatory measures aren’t there.
   • Relatively short-term and symptom-focused. It ran 68 weeks and didn’t evaluate structural progression or durability beyond that period.
   • Industry involvement. Ozempic-maker Novo Nordisk funded this — which is not a deal-breaker in itself, but it is important context.
   • Imperfect blinding. Blinding may have been compromised in the semaglutide group, because people tend to notice when they lose ~14% of their body weight. Ironically, the greater the objective weight-reduction, the higher the risk of biasing the pain-reduction results! The more obvious the weight loss caused by semaglutide, the higher the risk that the reported pain relief reflects patient expectations rather than a purely biological benefit. But this is far from a fatal flaw. There might not have been any inflation, or not much, and the results are impressive enough that some uncertainty is hardly a deal-breaker. A legitimate effect on pain is highly plausible and even predictable given substantial weight loss (however achieved), and the weight-loss result is not in doubt here at all.
3. Xie Y, Choi T, Al-Aly Z. Mapping the effectiveness and risks of GLP-1 receptor agonists. Nat Med. 2025 Mar;31(3):951–962. PubMed 39833406 ❐
4. The hazard ratios were tiny — just ~1.1! — and based on notoriously imprecise diagnostic code data, not joint-specific symptom measurements. Small increases in those MSK codes are definitely not necessarily drug side effects: they could come from more clinic contact, detection bias, miscoding, activity-related aches during weight loss, or unmasking old pains as people move more. Semaglutide clearly improves knee OA symptoms in the specific group studied by Bliddal, while Xie’s broad database signals reflect tiny, noisy shifts that are much harder to interpret.
5. That’s not wishful thinking. Remember, we have Xie et al to confirm that there was distinct lack of negative health effects in a huge sample of patients.