GLP-1 Reality Check: Doctors Underestimate the Upside, Patients Underuse the Drugs
New survey and claims data suggest a quiet mismatch in cardiometabolic care: physicians may be undercounting how much weight and risk these drugs move, while many eligible adults never start them.
The strangest thing about the GLP-1 era is not how fast these drugs arrived, but how unevenly the evidence about them has settled into everyday medicine. Semaglutide and tirzepatide have, in a few short years, rewritten what is possible in the treatment of obesity and cardiometabolic disease. Yet a new survey of attending physicians suggests that many of the doctors writing — or declining to write — those prescriptions are working from a quieter, more conservative picture of the drugs than the trials describe. At the same time, claims data show that uptake among the adults most likely to benefit is patchy. The result is a gap between evidence and practice that is worth naming plainly, before it hardens into habit.
- Physicians underestimated weight loss. Surveyed doctors guessed average weight loss of roughly 9% on GLP-1s; the STEP and SURMOUNT trials reported 14.9% and 18.5%.
- Cardiovascular benefit is widely under-recognized. Only about 39% of physicians perceived a cardiovascular benefit in nondiabetic patients — below what random guessing would predict.
- Side effects were underestimated, discontinuation overestimated. Doctors expected milder tolerability problems but more quitting than trials actually documented.
- Uptake is uneven. Among commercially insured adults with new obesity diagnoses and no diabetes, only a narrow slice initiated semaglutide within six months.
- The evidence is moderate, not settled. Trial populations, real-world adherence and long-term safety still need filling in — a reason for careful conversations, not blanket enthusiasm or blanket caution.
What the doctors thought — and what the trials showed
In a questionnaire study of 165 internal and family medicine attendings, with 122 responding, physicians estimated that patients on semaglutide or tirzepatide lose, on average, about 9.22% of their body weight. The pivotal trials tell a more striking story: 14.9% mean weight loss in STEP for semaglutide and 18.5% in SURMOUNT for tirzepatide. That is not a rounding error. For a 220-pound adult, the gap between a doctor's mental model and the trial average is roughly twenty pounds of expected loss.
The cardiovascular picture is more unsettling. Only 48.4% of surveyed physicians perceived a cardiovascular benefit for patients with diabetes — statistically indistinguishable from a coin flip — and just 39.3% perceived one in nondiabetic patients, a result the authors note is significantly below what random guessing would produce. In other words, when it came to the heart, doctors were not merely uncertain; a meaningful share were leaning the wrong way.
Tolerability perceptions were skewed too, though in a different direction. Physicians estimated side effect rates around 32.6%, far below the 89.7% and 80.5% reported in the STEP and SURMOUNT trials, while expected discontinuation (8.59%) ran higher than the trial data. The composite portrait is a profession that, at least in this sample, underestimates how much weight the drugs move, underestimates how often patients feel something, and overestimates how often they quit.
Doctors were not merely uncertain about cardiovascular benefit. A meaningful share were leaning the wrong way.
The conversation in the room is shaped by what the clinician believes the drug can — and cannot — do.
Why the perception gap matters
Survey studies have well-known limits. A questionnaire of 122 internal and family medicine attendings is not a verdict on American medicine, and self-reported perceptions are not the same as prescribing behavior. The evidence here is best described as moderate: suggestive, internally consistent, and worth taking seriously, but not the last word. Even with those caveats, the direction of the gap matters. A clinician who quietly expects 9% weight loss may be less likely to raise GLP-1s with a patient who would, in trial conditions, lose closer to 15% or 18%. A clinician unsure whether the drugs help the heart may frame them as cosmetic rather than cardiometabolic.
That framing has consequences downstream. Patients who are told the drugs are modest and risky tend to behave accordingly: they wait, they hesitate, they stop at the first wave of nausea. Patients who are told the drugs are powerful but demanding — meaningful weight loss, real gastrointestinal side effects, slow titration, long-term commitment — tend to enter treatment with a more accurate map.
Who actually starts the drug
The other half of the gap is on the patient side. A retrospective cohort study in JAMA Network Open used the Merative MarketScan commercial claims database to follow adults with a first obesity diagnosis between June 2021 and July 2022 who had no prior antiobesity medication, no prior GLP-1, no bariatric surgery and no diabetes-related claim in the year before diagnosis. The question was simple: who, in this eligible-on-paper population, actually initiated semaglutide within six months?
The answer, in broad strokes, is: not many, and not evenly. Using a cross-validated random forest model, the authors identified sociodemographic factors, prior medication classes and co-occurring diagnoses among the top predictors of initiation. Translation: who gets started on semaglutide for obesity is shaped less by a single clean clinical signal and more by a thicket of context — insurance, prior prescribing relationships, comorbidity load, and the patient's broader medical footprint. That is not unusual for a new, expensive therapy, but it is a reminder that "eligible" and "treated" are different categories.
Who walks out with the prescription is shaped by more than the diagnosis on the chart.
How to read this if you are the patient
None of this is a recommendation to start, stop, or switch a medication. It is context for a better conversation. If you are weighing a GLP-1 with a clinician, a few questions follow naturally from the evidence. What weight loss does the trial data actually show for the specific drug being discussed, and how does that compare with what your clinician expects in practice? What is known — and not yet known — about cardiovascular benefit in people without diabetes who share your risk profile? How will side effects be managed in the first months, when the trials suggest most patients feel something? What is the plan if the drug works, and the plan if it does not?
If you are already on a GLP-1 and tolerating it, the perception data is a quieter kind of reassurance: the trial weight loss numbers your clinician may be implicitly anchoring to could be lower than what the studies report. If you are struggling with side effects, the same data argues for patience and titration rather than assuming you are an outlier.
The gap worth closing
The GLP-1 story is genuinely new, and new drugs always arrive faster than the collective clinical intuition that surrounds them. The survey data suggest that intuition is still catching up — undercounting the weight loss, undercounting the cardiovascular signal, miscounting the tolerability picture in both directions. The claims data suggest that, in the meantime, the people most likely to benefit are reaching the drugs through an uneven set of pathways. Neither finding calls for hype. Both call for more careful conversations, in more exam rooms, with more accurate numbers on the table.
Sources
- Physician Perceptions of the Safety and Efficacy of GLP-1 Receptor Agonists: Underestimation of Cardiovascular Risk Reduction and Discrepancies with Clinical Evidence. — Journal of cardiovascular development and disease
- Factors Associated With Semaglutide Initiation Among Adults With Obesity. — JAMA network open