Cost-Effectiveness Models Assume Patients You Don’t Have
Reviewing Betensky et al. The Cost-Effectiveness of Semaglutide and Tirzepatide for PatientsWith Knee Osteoarthritis and Obesity. Ann Intern Med. 2025;178:1549-1560.
Clinical Takeaway: Cost-effectiveness models projecting tirzepatide saves $57,400 per quality-adjusted life-year assume 76% of patients stay on the drug at five years. Real-world retention data shows 50-70% quit within one year. When the entire cost-effectiveness ratio depends on lifetime treatment duration but half your patients can’t afford $10,500 annually or tolerate GI side effects past six months, the model isn’t just optimistic—it’s describing a different universe.
This isn’t about whether GLP-1RAs work for knee osteoarthritis. It’s about whether projections built on pharmaceutical-sponsored adherence fantasies tell you anything useful about your patients.
What They Modeled
Brigham researchers used a Monte Carlo simulation to project lifetime outcomes for patients with knee osteoarthritis and obesity—mean BMI 40, mean age 56, 82% women. They compared usual care, diet and exercise, semaglutide, tirzepatide, and bariatric surgery.
Key assumptions for tirzepatide:
25% BMI reduction at year one (15% if diabetic), maintained for duration of treatment
Annual cost: $10,500 (Zepbound) or $4,300 (Mounjaro for diabetics)
Discontinuation: 21% year one, 3% per year thereafter
Upon discontinuation: all weight regained over two years
Pain reduction: 58% (driven by weight loss)
MACE risk reduction: 20%
Diabetes remission: 79%
That 3% annual discontinuation after year one means 76% of patients are still on tirzepatide at five years in their model.
What They Found
From the healthcare perspective, tirzepatide dominated semaglutide—more quality-adjusted life-years, lower cost. Compared to diet and exercise, tirzepatide had an incremental cost-effectiveness ratio of $57,400 per QALY. Diet and exercise versus usual care: $25,400 per QALY.
Usual care delivered 9.59 QALYs at $222,300 lifetime cost. Tirzepatide: 10.68 QALYs at $280,000.
When bariatric surgery was included (for patients with BMI ≥35), Roux-en-Y gastric bypass dominated both GLP-1RAs with an ICER of $30,700 per QALY versus laparoscopic sleeve gastrectomy. One-time upfront cost ($30,200), sustained weight loss (30% BMI reduction at year four), no chronic medication burden.
In probabilistic sensitivity analysis, tirzepatide was cost-effective in 64% of simulations at a $100,000 per QALY threshold; semaglutide in 34%. When surgery was available, RYGB won 68% of the time.
The Adherence Fantasy
The model assumes 21% first-year discontinuation and 3% annually thereafter. This means 79% stay on tirzepatide through year one, and 76% are still taking it at five years.
Real-world persistence data for anti-obesity medications shows 50-70% discontinuation within one year. Reasons: cost, GI intolerance, injection aversion, lack of perceived benefit. The STEP-1 extension study showed substantial weight regain within one year of stopping semaglutide—not the optimistic two-year trajectory modeled here.
If you plug in 60% first-year discontinuation instead of 21%, the denominator—QALYs gained—collapses. Patients who quit after six months get brief weight loss and temporary pain relief, then return to baseline. The lifetime projection becomes a short-term intervention with minimal durability.
The ICER doesn’t just increase. It explodes.
What Sensitivity Analyses Revealed
The model was most sensitive to tirzepatide cost and BMI reduction efficacy. When annual cost increased to 150% of base case, ICERs crossed $100,000 per QALY. When BMI reduction dropped to 50% of expected (12.5% instead of 25%), semaglutide dominated tirzepatide.
Starting BMI also mattered. At very low (33 kg/m²) or very high (54 kg/m²) BMI, tirzepatide’s ICER increased. At lower BMI, diet and exercise can achieve similar obesity class changes. At very high BMI, even large percentage reductions may not shift patients into lower obesity categories.
What didn’t meaningfully affect ICERs: pain reduction assumptions, adverse event probabilities, demographics. The cost-effectiveness hinged entirely on drug cost and weight loss durability—which means adherence is the whole game.
The Cost No Model Captures
Annual cost in this model: $10,500 for Zepbound, $4,300 for Mounjaro (if diabetic). These are net prices after rebates—optimistic averages that don’t reflect what patients actually pay.
Medicare doesn’t cover weight loss drugs. Commercial insurance increasingly does, but prior authorizations require documented BMI ≥30 with comorbidity or BMI ≥27 with weight-related condition, failed diet and exercise attempts, and sometimes mandatory nutrition counseling.
Even with coverage, copays can hit $200-500 monthly. For a patient with knee osteoarthritis already paying for NSAIDs, PT, and periodic steroid injections, adding $2,400-6,000 annually out-of-pocket isn’t trivial.
The model assumes everyone who starts can sustain cost indefinitely. Your Medicaid patient paying $400/month out-of-pocket? Not in this simulation.
Who This Doesn’t Apply To
Base case cohort: 82% women, mean age 56, mean BMI 40, mean WOMAC pain score 71 (out of 100). This matches the STEP-9 trial population.
If you’re treating younger patients with BMI 33 and moderate pain, the benefit-cost calculation shifts. If you’re treating older patients with multiple comorbidities who can’t tolerate GI side effects, adherence assumptions break down. If your patient has BMI 29 and knee osteoarthritis, this model excluded them entirely.
The model also doesn’t capture opportunity cost. Every dollar spent on tirzepatide is a dollar not spent on PT, aquatic therapy, or addressing the social determinants keeping patients sedentary.
What You Actually Need to Know
Cost-effectiveness analyses answer: “If patients took this medication perfectly, paid for it indefinitely, and experienced average trial outcomes, would it be worth it for a payer with a defined threshold?”
They don’t answer: “Will my patient be able to afford this, tolerate it, and stay on it long enough to benefit?”
The ICER of $57,400 per QALY assumes lifetime treatment with 3% annual discontinuation after year one. If 60% of your patients quit within a year, you’re not getting lifetime QALYs. You’re getting six months of weight loss and temporary pain relief, followed by regain.
If you have a patient with BMI 40 and severe knee osteoarthritis pain who can afford tirzepatide, isn’t diabetic (so they’re paying $10,500/year instead of $4,300), tolerates injections, and has failed diet and exercise, this model suggests it might be worth it—if they stay on it.
If your patient is on Medicaid (doesn’t cover weight loss drugs), has marginal insurance with high copays, or is one of the 60% who quit within a year, the cost-effectiveness evaporates.
If they’re eligible for bariatric surgery and willing to do it, that’s probably the better long-term play from a value standpoint. But only 1-2% of eligible patients undergo bariatric surgery annually. Stigma, fear of surgery, insurance barriers, and lack of access all limit uptake. So yes, surgery wins in the model. But if your patient won’t do it, the comparison is academic.
What About Diet and Exercise?
Diet and exercise had an ICER of $25,400 per QALY versus usual care. First-year cost: $5,100. Ongoing cost: $1,400 annually.
The model assumed 8% BMI reduction with 21% discontinuation and 49% pain reduction. Real-world data from structured programs like WE-CAN show this is achievable in research settings with intensive support. Your patient trying to lose weight on their own? Much harder.
So diet and exercise is “cost-effective” if patients can access structured programs, sustain behavior change, and lose weight. Most can’t.
When you see cost-effectiveness analyses projecting lifetime benefits from chronic medications, check the discontinuation rates they assumed. If they’re assuming 75%+ retention at five years for a drug that costs $10,000 annually, they’re modeling a world that doesn’t exist. And if the entire ICER depends on lifetime treatment duration, then real-world adherence isn’t a limitation—it’s everything.
References
Ganguly R, Tian Y, Kong SX, et al. Persistence of newer anti-obesity medications in a real-world setting. Diabetes Res Clin Pract. 2018;143:348-356. doi:10.1016/j.diabres.2018.07.017
Betensky DJ, Smith KC, Katz JN, et al. The cost-effectiveness of semaglutide and tirzepatide for patients with knee osteoarthritis and obesity. Ann Intern Med. 2025;178(11):1549-1560. doi:10.7326/ANNALS-24-03609
Wilding JPH, Batterham RL, Davies M, et al; STEP 1 Study Group. Weight regain and cardiometabolic effects after withdrawal of semaglutide: the STEP 1 trial extension. Diabetes Obes Metab. 2022;24(8):1553-1564. doi:10.1111/dom.14725
Bliddal H, Bays H, Czernichow S, et al; STEP 9 Study Group. Once-weekly semaglutide in persons with obesity and knee osteoarthritis. N Engl J Med. 2024;391(17):1573-1583. doi:10.1056/NEJMoa2403664
Office of the Assistant Secretary for Planning and Evaluation, Office of Health Policy, U.S. Department of Health and Human Services. Medicare coverage of anti-obesity medications. Accessed February 1, 2026. https://aspe.hhs.gov/reports/medicare-coverage-anti-obesity-medications
Campos GM, Khoraki J, Browning MG, et al. Changes in utilization of bariatric surgery in the United States from 1993 to 2016. Ann Surg. 2020;271(2):201-209. doi:10.1097/SLA.0000000000003554