Load bearing

Many employers were caught flat-footed, even cornered, by the GLP-1 conversation, and a new drug (Reta) has the potential to make it worse.

GLP-1s have been the most insatiable topic this industry has explored since the ACA. I have written about the investment case for continuing, using Bank of America's own filings, and the cost case for stopping, using Starbucks'. I built the calculator nobody else had built. As far as I am concerned that file is closed, at least for now.

Eli Lilly plans to submit a new drug to the FDA in the first quarter of 2027. Reta is short for retatrutide. It is being filed for three indications: obesity, sleep apnea, and knee osteoarthritis pain. No drug in this class has ever carried an arthritis indication. Whether a weight-management exclusion reaches a knee-pain claim turns on a question a court has already framed, and on clinical evidence that gets tested again in early 2027.

As humans, and especially as modern Americans, we all want to look good, feel good, and live well. All of it feeds the weight loss machine, from GLP-1s to wellness benefits and everything after.

Filed for three doors

Lilly's July 2026 release states that the clinical data package supports submission for obesity, knee osteoarthritis pain, and obstructive sleep apnea. Submission is planned for the first quarter of 2027, which on an ordinary review clock puts a decision somewhere in late 2027 or 2028.

Lilly is not the first to run the knee trial. Novo ran the same study with semaglutide, hit the endpoint, published it in the New England Journal in 2024, and as far as the public record shows has not filed for the indication. Lilly is converting it into a label.

Here is why that matters to a plan sponsor. Most employers who keep this class out do it with a weight-management exclusion. Whether that language reaches a claim for knee pain is the whole question, and it is no longer hypothetical.

It has already been through federal court on the sleep apnea indication. A class action filed in Washington in September 2025 challenged a weight-management exclusion applied to deny Zepbound prescribed for moderate to severe sleep apnea. In June the court dismissed it, and the reasoning is the part worth reading. The judge drew the line on mechanism. A diabetes drug, he noted, is approved to treat diabetes regardless of whether the patient is also obese. Zepbound improves sleep apnea because it reduces body weight. The exclusion reached it.

Hold onto that, because it is the same test the knee indication will meet. Whether your own plan document survives it is a question for your ERISA counsel rather than your broker. What I can tell you is that a court has now told everyone which question it intends to ask, and that this drug arrives with three doors rather than one.

The migration pool is already there, too. An all-payer claims analysis published in June found that 25.5% of current GLP-1 users already carry a sleep apnea diagnosis.

Does it treat the knee, or the person?

The pain results are real. TRIUMPH-4 is the knee arthritis trial inside Lilly's phase 3 program, 445 people with obesity and knee osteoarthritis followed for 68 weeks. Pain fell 4.5 points on a ten-point scale at the 9 mg dose, against 2.4 points on placebo.

But the drug appears to work on the knee by making the person smaller, rather than by acting on the joint. That reading is based on the research as it stands today and on how it has been interpreted, and it matters more than it may sound. A great many employers have decided not to cover drugs that make people smaller. That decision was written against a purpose. This drug arrives with a different purpose printed on the box and generally the same mechanism underneath it.

Here is the short version, and it is short on purpose, because this is a clinical question rather than a benefits one.

Neither trial set out to separate the weight effect from a direct effect on the joint, and neither has published an analysis that does. That goes for TRIUMPH-4 and for STEP 9, Novo's matching knee trial of semaglutide. The one trial that did isolate the question gave liraglutide to knee arthritis patients who had already lost the weight on a diet, and it produced no pain improvement at all. Across the literature, pain relief tracks how much weight comes off, and it takes a great deal of weight to move it.

Two things are worth stating carefully. Weight-mediated is not the same as not causal, since the drug does cause the pain relief and the only question is what carries it. And there is real laboratory evidence of receptors for these drugs in human joint tissue, plus animal data showing a joint effect independent of weight, none of which has been shown in people. A version that goes straight into the knee, producing no weight loss at all, is in trials now with results expected in early 2027. If that reads out positive the picture changes. Whether it does is not mine to settle.

On the evidence available today, a knee osteoarthritis indication for this drug is a weight-loss indication with a different label and a different prior authorization pathway.

Which is, as it happens, the same test the court applied in June. The sleep apnea exclusion held because the drug worked through weight. If the knee benefit works the same way, the same reasoning reaches it. And if that trial of the version injected straight into the knee reads out positive in early 2027, that reasoning gets harder to lean on. Your exclusion and a clinical readout are now tied together, which is not a sentence I expected to write.

How many knees, and attached to whom

Roughly 14 million US adults have symptomatic knee osteoarthritis. That is the published figure and it rests on inputs from around 2012, so the real number today is probably closer to 16 million.

About 57% of them are under 65. The assumption that arthritis is a retiree problem is wrong, and it is wrong in the direction that costs an employer money.

So four in ten Americans with knee arthritis are not obese. Their knees hurt for reasons a weight-loss drug does not address, and prior knee injury raises knee osteoarthritis risk about as much as obesity does.

I am one of them. I am six foot two and two hundred ten pounds, which puts my BMI at 27.0, almost exactly the floor Lilly used to enroll its knee trial. I have also had four knee surgeries in the space of three years, and not one of them had anything to do with my weight. On paper I would qualify for this drug on the knee indication. Losing weight would not have prevented a single one of those operations.

The trial floor was a BMI of 27. That captures roughly 78% of knee osteoarthritis patients on paper and leaves three to four million below the line.

Per 1,000 covered adults, run this against your own census:

• about 40 with symptomatic knee osteoarthritis
• about 24 obese, 16 not
• about 31 above a BMI 27 line
• about 20 who will ever appear in your claims data with a knee osteoarthritis diagnosis

That last line is the one to sit with. Your claims file sees about half the people who actually have this. Anyone sizing the exposure off claims is looking at half the population.

Two caveats belong here, and both cut against the argument I am making. About half of US knee osteoarthritis is attributable to excess weight, so even a perfectly effective obesity intervention leaves the other half exactly where it is. And 84% of TRIUMPH-4 participants had a BMI of 35 or higher, with a mean starting weight around 248 pounds, which is far heavier than the knee osteoarthritis population generally. What the drug does for a 29-BMI knee is an extrapolation rather than a finding.

Why the dose logic breaks

There is a direct relationship between body weight and knee pain, and that much is not in dispute. What gets skipped is the shape of the curve. Most of the benefit arrives in the first 5 to 10 percent of weight lost, and it flattens hard after that. Quadrupling the weight loss buys roughly two and a half times the symptom improvement, not four times. In the landmark diet-and-exercise trial in this population, diet alone at 9.5 percent weight loss produced no more pain relief than exercise alone at 2.0 percent. Reta is a drug built to deliver close to four times the weight loss at which most of the joint benefit has already been collected.

Which sets up the plan design question. If most of the joint benefit is bought in the first ten percent, what is a plan paying a premium price for when it pays for twenty-eight?

One thing has to be conceded here, and it is the strongest argument on the other side. An injection is less invasive than a knee replacement, and knee replacement is where a meaningful share of this population ends up. In a commercially insured book, about 8.9% of diagnosed knee osteoarthritis patients had an arthroplasty in a single year, and another 12.2% over the following three, so roughly one in five reaches surgery within four years, at an average commercial cost around $35,600.

Surgery, or a subscription

As for what Reta will cost, there is no number yet. Lilly has said nothing about pricing on any earnings call this year, and every figure circulating online traces back to somebody selling the grey market version.

My expectation is that it lands close to where tirzepatide already sits, with room above it. Lilly has never priced by indication, and there is a structural reason it would not start now. A drug carries one price per NDC, and there is no lawful, workable way to charge a knee patient differently from a weight-loss patient. When Lilly added the sleep apnea indication to Zepbound in December 2024, displacing CPAP and in some cases airway surgery, it did not move the price at all.

The room above it is real, though, and somebody else already documented it. ICER put the value-justified price for tirzepatide at $11,700 to $16,100 a year. It actually nets about $7,973. The class is already selling well below what its own clinical value would support, which means the knee and the sleep apnea indications do not have to invent a new price. They only have to defend a higher one inside a range that is already published.

The wrong pocket

Which is where this stops being a question about price and becomes a question about the calendar.

The median age at which symptomatic knee osteoarthritis gets diagnosed is 55. You cover a member until 65. That is ten years of therapy, and at today's net price it runs a little under $80,000 per member before anybody has avoided anything.

The mean age at knee replacement is 67.6.

Read those two sentences together and the shape of the problem shows up. The drug spend sits inside the employment window. The surgery it is supposed to displace sits, on average, just outside it. Only about a third of knee replacement volume is billed to commercial insurance at all, on the most recent national payer split anyone has published, which dates to 2004. The plan bears the load. It does not, usually, get the knee.

And you do not buy one knee. Take it back to the census.

Per 1,000 covered adults, about 16 carry a knee osteoarthritis diagnosis and sit above the trial's BMI floor. At today's net price their therapy runs about $128,000 a year, and that number holds year over year, because as some age out at 65 others age into the diagnosis behind them. Play it forward a decade and it is roughly $1.3 million per 1,000 covered adults.

If every one of those 16 went on to have a knee replaced, that is about $570,000.

Two things cut against this and both belong in the piece. Somewhere between four and four and a half in ten knee replacements are performed before 65, so the handoff is not clean. And about one in five diagnosed knee osteoarthritis patients reaches surgery within four to five years, which for a 55-year-old is comfortably inside the window.

The other direction is worse, though. The employer paying for year one is usually not the employer paying for year ten. Average enrollment in a commercial plan runs about 48 months, and only a quarter of members are still there after five years. So the decade of drug spend gets split across two or three plans, none of which will be holding the member at 67.6. I wrote about the age side of this in the greening rate, and this belongs on that list.

Health economists have a name for this. It is the wrong pocket problem, where one party funds an investment and a different party collects the return, and employment-based coverage is close to the textbook case of it. The finding is not new either. A 2011 paper in the American Economic Review concluded that turnover-driven underinvestment during working years leaves workers with higher medical spending once they reach Medicare. This is that mechanism, arriving with a price tag attached.

The problem is what each one buys. A replaced knee does not need replacing again next year. A medication is a different story. After somebody stops, about three-quarters of the weight comes back, half of it inside 23 weeks, and real-world persistence at one year is only about a third. Surgery is a one-time cost carrying a durable result. The drug is a recurring cost carrying a result that depends on continuing to pay for it.

What to do

Three moves, and none of them require knowing what FDA does in 2027.

1. Pull your plan document and your PBM's utilization management criteria and find out whether coverage is defined by drug, by drug class, or by indication. Most exclusions were drafted against two molecules and a weight-loss purpose. Ask before renewal, not after.
2. Ask your PBM in writing how it will adjudicate a claim in this class written against a non-weight diagnosis, and what documentation it will require. Sleep apnea has an expensive objective gate, a sleep study with a number attached to it. Knee osteoarthritis's gate is clinical criteria plus a plain X-ray.
3. Size the population off how many people actually have this rather than off your claims file, and know that the two differ by roughly half.

The Tail

A tail worth noting is that Reta is already available on the grey market. Influencers on TikTok and Instagram tout its ability to shed weight while holding onto muscle mass, which is a claim no published trial supports. Lilly knows. In August the company filed a round of suits over illicit retatrutide, and says it has referred more than two hundred individuals and entities to regulators and flagged more than fourteen thousand websites, ads and posts.

The reason this matters is that some of your employees may have a head start on your strategy, your plan, and your budget.

Reta is not approved, so it cannot be prescribed. It also cannot be legally compounded, and this is the part most coverage gets wrong. Compounded semaglutide and tirzepatide were lawful for a period under the shortage carve-out, and that window closed in 2025. Reta never had a window. It is not in a pharmacopeia monograph, it is not a component of any approved product, and it was never on a shortage list. It fails every route.

So it moves as a research chemical, labeled not for human consumption, sold online. Gallup put 11% of US adults on something in this class in June. Of those, 19% were on a compounded or custom-mixed product, and 12% did not know which they were taking.

None of this is settled until FDA decides, and there is time to prepare. What is already settled is that some share of that 19% works for somebody, and none of it is in anyone's claims data.

Sources

1. Eli Lilly and Company, "Lilly's triple agonist, retatrutide, successful in two additional Phase 3 obesity trials, delivering significant improvements in weight and A1C," July 23, 2026.
2. Eli Lilly and Company, "Lilly's triple agonist, retatrutide, delivered weight loss of up to an average of 71.2 lbs along with substantial relief from osteoarthritis pain in first successful Phase 3 trial," December 11, 2025 (TRIUMPH-4). Topline release only; no peer-reviewed publication exists as of August 2026.
3. Bliddal H, et al., "Once-Weekly Semaglutide in Persons with Obesity and Knee Osteoarthritis," New England Journal of Medicine, October 31, 2024 (STEP 9), and the trial record at ClinicalTrials.gov NCT05064735. Neither the published outcome measures nor the registration includes an analysis separating the weight effect from a direct joint effect.
4. Gudbergsen H, et al., liraglutide following diet-induced weight loss in knee osteoarthritis, randomized placebo-controlled trial, 2021.
5. Zhu H, et al., "Glucagon-like peptide-1 receptor agonists as a disease-modifying therapy for knee osteoarthritis mediated by weight loss," Annals of the Rheumatic Diseases, volume 82, issue 9, September 2023.
6. Meurot C, et al., "Liraglutide, a glucagon-like peptide 1 receptor agonist, exerts analgesic, anti-inflammatory and anti-degradative actions in osteoarthritis," Scientific Reports, 2022.
7. 4Moving Biotech, Phase 2a trial of 4P004, an intra-articular GLP-1 analog for knee osteoarthritis; FDA Fast Track designation, May 2026.
8. Hamburger v. CVS Caremark and Group Hospitalization and Medical Services, Inc., No. 1:25-cv-03000 (D.D.C.), complaint filed September 4, 2025, dismissed June 10, 2026. The court held that a "prescription drugs for weight loss" exclusion reached Zepbound prescribed for obstructive sleep apnea, distinguishing drugs approved to treat a condition regardless of obesity from a drug that improves the condition because it reduces body weight.
9. Trilliant Health, analysis of emerging and newly approved indications among GLP-1 patients, reported June 11, 2026.
10. Deshpande BR, Katz JN, Losina E, et al., "Number of Persons With Symptomatic Knee Osteoarthritis in the US," Arthritis Care and Research, volume 68, issue 12, November 2016.
11. Dillon CF, et al., "Prevalence of knee osteoarthritis in the United States: arthritis data from the Third National Health and Nutrition Examination Survey 1991-94," Journal of Rheumatology, 2006.
12. National Center for Health Statistics, "Obesity and Severe Obesity Prevalence in Adults: United States, August 2021 to August 2023," Data Brief 508, September 2024.
13. Zheng H, Chen C, "Body mass index and risk of knee osteoarthritis: systematic review and meta-analysis of prospective studies," BMJ Open, 2015.
14. Blagojevic M, et al., "Risk factors for onset of osteoarthritis of the knee in older adults: a systematic review and meta-analysis," Osteoarthritis and Cartilage, 2010.
15. Messier SP, et al., "Effects of Intensive Diet and Exercise on Knee Joint Loads, Inflammation, and Clinical Outcomes Among Overweight and Obese Adults With Knee Osteoarthritis," JAMA, 2013 (the IDEA trial).
16. Atukorala I, et al., weight loss dose-response and knee osteoarthritis symptoms, Arthritis Care and Research, 2016.
17. Fitch K, Tomicki S, Bazell C, Milliman, "Knee Osteoarthritis in a Commercially Insured Population," January 2020. MarketScan and Milliman CHSD, 2013 to 2017. The $35,605 figure is the average allowed cost per patient undergoing primary total or partial knee arthroplasty, in 2017 dollars. It excludes revisions; the all-arthroplasty figure including revisions is lower, at $32,454.
18. Sloan M, Premkumar A, Sheth NP, projected US total joint arthroplasty volume, Journal of Bone and Joint Surgery (American), 2018.
19. Betensky DJ, Smith KC, Katz JN, Losina E, et al., "The Cost-Effectiveness of Semaglutide and Tirzepatide for Patients With Knee Osteoarthritis and Obesity," Annals of Internal Medicine, September 15, 2025.
20. Institute for Clinical and Economic Review, "Semaglutide and Tirzepatide for Obesity," Final Evidence Report, December 16, 2025, for net prices, health-benefit price benchmarks, and the budget impact finding.
21. Institute for Clinical and Economic Review, "Indication-Specific Pricing of Pharmaceuticals in the United States," 2015 Policy Summit report, March 2016.
22. US Senate Committee on Finance, "The Price of Sovaldi and Its Impact on the U.S. Health Care System," December 1, 2015.
23. Novartis statements on Kymriah pricing relative to allogeneic stem cell transplant, August 2017.
24. Amgen, "Amgen Makes Repatha (evolocumab) Available in the US at a 60 Percent Reduced List Price," October 24, 2018.
25. Systematic review and meta-regression of weight regain following discontinuation of GLP-1 receptor agonists, eClinicalMedicine, March 2026.
26. Peterson Health Technology Institute, "Employer Approaches to GLP-1 Coverage," Market Trend Report, December 2025.
27. Gallup, "In U.S., GLP-1 Usage Reaches New High," June 2026.
28. US Food and Drug Administration, letter to the Federation of State Medical Boards and the National Association of Boards of Pharmacy regarding compounded drug products containing retatrutide, March 31, 2025.
29. Global Burden of Disease 2021 knee osteoarthritis analysis, PLOS One, 2025, for the share of knee osteoarthritis burden attributable to high body mass index in high-income North America.
30. Eli Lilly and Company, "FDA approves Zepbound (tirzepatide) for moderate-to-severe obstructive sleep apnea," December 20, 2024, and the Lilly Colorado WAC disclosure sheet for Zepbound.
31. Eli Lilly and Company, second quarter 2026 earnings call, August 12, 2026. Retatrutide is discussed at length; no pricing is disclosed.
32. Eli Lilly and Company and the US government, agreement on obesity medicine pricing, November 6, 2025.
33. Reporting on Eli Lilly litigation against sellers of illicit retatrutide, August 12, 2026. The referral and takedown counts are Lilly's own statements as reported; the underlying complaints were not reviewed.
34. Nelson Mullins Riley and Scarborough, "Health Plan Coverage of GLP-1s: Is a Simple Weight Loss Exclusion Enough?", Compensation and Benefits Brief, July 9, 2026.
35. Prime Therapeutics, GLP-1 Pipeline Update, February 19, 2026, which lists retatrutide indication timelines including knee osteoarthritis with obesity.
36. Optum Rx, "The GLP-1 revolution continues," February 3, 2026.
37. Losina E, et al., "Lifetime Risk and Age at Diagnosis of Symptomatic Knee Osteoarthritis in the US," Arthritis Care and Research, 2013.
38. American Joint Replacement Registry, 2024 Annual Report, data 2012 to 2023, for the mean age at primary total knee arthroplasty of 67.6 years.
39. Fang H, Gavazza A, "Dynamic Inefficiencies in an Employment-Based Health Insurance System: Theory and Evidence," American Economic Review, volume 101, number 7, 2011.
40. McCullough JM, "Declines in Spending Despite Positive Returns on Investment: Understanding Public Health's Wrong Pocket Problem," Frontiers in Public Health, volume 7, 2019.
41. "Trends in Disenrollment and Reenrollment Within US Commercial Health Insurance Plans, 2006 to 2018," JAMA Network Open, 2022.
42. TRIUMPH-7, ClinicalTrials.gov NCT07035093, obesity or overweight with chronic low back pain. N=586, first patient dosed May 29, 2025, primary completion September 2027. Co-primary endpoints are pain intensity and percent change in body weight. Lilly has not stated an intention to seek this indication.
43. Funck-Brentano T, Nethander M, Moverare-Skrtic S, Richette P, Ohlsson C, "Causal Factors for Knee, Hip, and Hand Osteoarthritis: A Mendelian Randomization Study in the UK Biobank," Arthritis and Rheumatology, volume 71, number 10, 2019, for the causal odds ratios of 1.76 knee, 1.52 hip, and 1.00 hand.
44. Wegovy (semaglutide) FDA approval history: obesity June 2021, cardiovascular risk reduction March 2024, metabolic dysfunction-associated steatohepatitis August 2025.

Fringe Theory is independent and unaffiliated. Views expressed are my own and do not represent those of my employer. Nothing here is legal, tax, medical, or investment advice.