GLP-1 drugs may be among the most consequential pharmaceutical developments of this generation, and nothing here disputes that. Semaglutide, tirzepatide and the drugs likely to follow them have demonstrated an extraordinary ability to produce sustained weight loss while improving important metabolic and cardiovascular outcomes. For millions of people, they may materially improve both the quality and the length of life.
A different question receives far less attention: what happens when people stop? For younger adults, that question matters. For someone in their late 60s, 70s or 80s, it may matter considerably more. When I began putting numbers around it, the result surprised me. There may already be 3–4 million Americans over the age of 65 exposed to what I think of as the GLP-1 exit problem.
That number is not a prediction that 3–4 million people will be injured by these drugs. It is something narrower: an estimate of how many older Americans could enter a physiological sequence that remains poorly understood —
flowchart LR
A[Rapid weight loss] --> B[Loss of lean tissue and muscle]
B --> C[Discontinuation]
C --> D[Substantial weight regain]
For an older adult with limited muscular reserve, that sequence deserves far more attention than it is getting.
Start With the Population
The United States had approximately 64.6 million people age 65 and older in 2025, according to the Census Bureau.1 KFF's late-2025 polling found that 9% of Americans 65 and older reported currently using a GLP-1 medication.2 The arithmetic is simple:
64.6 million × 9% ≈ 5.8 million current users age 65+.
Then consider persistence. A large JAMA Network Open study of 125,474 adults with overweight or obesity found that 53.6% had discontinued their GLP-1 within one year and 72.2% within two years.3 The study did not report those figures as age-65-specific cumulative rates, so applying them to the older population is a scenario calculation, not an observed epidemiological result.
But one detail points in the wrong direction:
Important
People 65 and older were more likely to discontinue than younger patients — an adjusted hazard ratio of 1.28 among patients with diabetes and 1.18 among those without — and were significantly less likely to restart afterward.3
Apply the overall discontinuation rates to today's estimated older GLP-1 population:
| Scenario | Calculation | Estimated adults 65+ |
|---|---|---|
| Current GLP-1 population | 64.6M × 9% | 5.82M |
| Potentially discontinuing within 1 year | 5.82M × 53.6% | 3.12M |
| Potentially discontinuing within 2 years | 5.82M × 72.2% | 4.20M |
The 3–4 million figure is not an arbitrary alarming number. It falls almost directly out of current utilization and observed real-world discontinuation.
The potential problem is not that GLP-1s fail. It is that they work extraordinarily well in a population that may not always be able to remain on them.
Weight Loss Is Not All Fat Loss
Weight loss — whether produced by caloric restriction, surgery or medication — normally includes some loss of fat-free tissue, and GLP-1 treatment is no exception. In the SURMOUNT-1 body-composition substudy, participants receiving tirzepatide lost an average 21.3% of body weight, while fat mass fell 33.9% and lean mass fell 10.9%; roughly three quarters of the weight lost was fat and one quarter was lean mass.4 The semaglutide STEP-1 DXA substudy found the same pattern. FDA review data show an average decline of 14.73 kg in body weight, including 8.36 kg of fat mass and 5.26 kg of lean body mass.5
One distinction is routinely lost in public discussion: Lean mass is not synonymous with skeletal muscle.
DXA lean mass includes water and other fat-free tissue, so a scan showing 25% or 35% of lost weight as lean tissue does not mean the patient lost that share in skeletal muscle. A 2026 review that attempted to isolate skeletal muscle found reductions representing roughly 8.5% to 24.5% of total weight lost across the studies examined — and its authors stressed how little direct evidence exists in older adults specifically.6
That uncertainty should not reassure anyone. It should make us want better data.
A 30-Year-Old and a 75-Year-Old Are Not the Same Patient
Human beings lose muscle with age, which means the same percentage reduction in muscle mass does not carry the same functional weight at 30 and at 75. An older adult often begins treatment with less physiological reserve, a weaker anabolic response to dietary protein, more chronic disease, lower activity levels, and an already declining ability to rebuild lost tissue.
Very few older Americans are doing the one thing most likely to offset that.
National Health Interview Survey data show that only 20.7% of adults 65 and older met the federal muscle-strengthening recommendation in 2022 — 6.3% met the strength recommendation alone, and 14.4% met both the strength and aerobic recommendations.7 Roughly 79% were not meeting the muscle-strengthening guideline.
If that prevalence roughly carries over to older GLP-1 users, about 4.6 million of the estimated 5.8 million current users over 65 may not be doing the recommended muscle-strengthening work. Layer the observed discontinuation rates on top — as an illustrative stress test, not a forecast — and the result is roughly 2.5 million people at one year and 3.3 million at two years who could experience both discontinuation and inadequate resistance exercise. Those assumptions are not independent observations, so the calculation should not be mistaken for a measured clinical cohort. What it does is define the scale of the question, and the scale is not small.
Then They Stop
What happens next is much better documented: when GLP-1 treatment ends, substantial weight regain is common. Participants in the STEP-1 extension who stopped semaglutide regained approximately two-thirds of their prior weight loss within a year.8 Tirzepatide taught the same lesson. In SURMOUNT-4, participants first lost an average 20.9% of body weight; those then switched to placebo saw an average 14% increase in body weight over the following 52 weeks, while those who stayed on tirzepatide lost an additional 5.5%.9
This supports what clinicians increasingly say outright: for many patients, obesity pharmacotherapy should be treated as chronic therapy, not a temporary course of medicine. That is pharmacologically understandable. It also creates a second problem, because long-term medical treatment exists inside a short-term economic system.
Biology May Say "Continue." Economics May Say "Stop."
A drug can be intended for chronic use while access to it remains anything but permanent, and this is the most underestimated part of the GLP-1 story. KFF found that 56% of people who had used GLP-1s described them as difficult to afford, despite most having at least some insurance coverage, and 14% of those who had used the drugs said cost was specifically why they were no longer taking them.10
That situation is changing for some older Americans, in a way that illustrates the problem rather than solving it.
As of 2026, Medicare's temporary GLP-1 Bridge gives qualifying Part D beneficiaries access to certain obesity drugs for a $50 monthly copayment through the end of 2027, and KFF estimates 3.8 million Medicare beneficiaries could meet its eligibility criteria based on 2023 claims.11 That figure lands, independently, at nearly the same population scale. But the Bridge is exactly what its name says. It is scheduled to end December 31, 2027, and KFF notes there is presently no clear Medicare coverage pathway for these patients afterward unless another mechanism is put in place.12
So picture the patient. She is 72. She finally gains access to a highly effective medication, her appetite falls dramatically, and she loses 20%, perhaps 25%, of her body weight. Her metabolic health improves. She is not resistance training consistently, and she loses some lean tissue along the way. Then her formulary changes, or her insurance changes, or the demonstration ends, or her copayment rises, or the side effects become too much, or she simply can no longer afford it.
The molecule doesn't care why she stopped.
Her physiology only registers that the pharmacological signal maintaining the weight loss is gone.
The Question Nobody Has Answered
There is currently no evidence demonstrating that millions of older Americans will suffer sarcopenia, frailty or serious medical deterioration because they discontinue GLP-1 therapy. There is not enough older-adult-specific data to support that claim.
Recent reviews say essentially the same thing: muscle loss is plausible enough and observed often enough to warrant attention, while evidence on strength, function, and long-term clinical outcomes in older adults remains thin. Researchers already recommend resistance exercise, adequate protein intake and monitoring of muscle health as components of treatment in higher-risk patients.613
That leaves an enormous open question.
When an older person loses substantial weight, loses some muscle along with it, stops treatment, and then regains much of that weight, what exactly comes back? Does muscle recover proportionately, or does fat return faster? What happens after a second cycle, or a third? What happens to gait speed, balance, grip strength, and fall risk? What happens to a 75-year-old who starts the process with borderline sarcopenia?
None of these are arguments against GLP-1s. They are questions created by their success.
A drug that produces extraordinary weight loss also creates an obligation to understand the composition of that loss — and the composition of the weight that returns.
This Became Personal
I did not come to this subject out of hostility toward pharmaceutical medicine. Quite the opposite — I am aggressively pro-science. But I am currently helping care for my mother in hospice, and I watched a version of this sequence unfold in someone I love.14
I cannot prove that a GLP-1 drug caused what ultimately happened to her, and I will not make that claim; there were other medical problems involved. But I watched dramatic weight loss. I watched strength disappear. I watched what happened after treatment stopped. Once I had seen it up close, I began asking whether what looked like an individual tragedy might expose a population-level vulnerability, and that question led me to the numbers. Not 20 million. Not 50 million. Approximately 3–4 million older Americans potentially passing through the discontinuation side of this treatment cycle, based on today's utilization and persistence data.
I sincerely hope the eventual clinical evidence shows this concern is overstated. But hoping something is benign is not science. Measuring it is.
The Drug Isn't the Problem. The Protocol May Be Incomplete.
The takeaway is not "don't take GLP-1s." That would ignore overwhelming evidence that these medications produce meaningful health benefits. The lesson is that weight cannot be the only metric we manage, particularly in older patients. Muscle mass matters. Strength matters. Protein intake and resistance training matter. The rate of weight loss and the maintenance dose may matter.
Above all, the exit strategy matters.
A 70-year-old beginning GLP-1 treatment should not simply receive a prescription and a target body weight. The treatment plan should anticipate what happens if that patient cannot stay on the medication three years from now. That means treating muscle preservation as part of the therapy rather than optional lifestyle advice, studying lower maintenance doses and individualized strategies, and measuring strength and function rather than celebrating the number on the scale. It also means recognizing that prescribing a medication intended for chronic use, without asking whether the patient can keep access to it, is itself a clinical variable.
GLP-1s may prove every bit as transformative as their advocates believe; I suspect they will. But transformative medicines mature. Dosages evolve, patient selection improves, monitoring improves, adjunctive treatment improves, and we learn which populations need different protocols.
That is not evidence that a medicine failed. That is what successful medicine looks like.
For younger patients, preserving muscle while losing weight is desirable. For older patients, it may be the difference between becoming lighter and becoming frail. If several million older Americans are now entering this experiment with us, we should find out which one we are producing.15