ByJohn A
What the Current Evidence Can and Cannot Say About the GLP-1 Drug List
The evidence here is strong in narrow places and thin everywhere else. Randomized trials support each approved product for the indication printed on its label. Very little supports direct ranking between them, outcomes measured in decades, or compounded copies, which have never been tested in a clinical trial at all.
What randomized trials established, drug by drug
Semaglutide for weight rests on the STEP program. The principal trial randomized adults with overweight or obesity and no diabetes to weekly semaglutide 2.4 mg or placebo for 68 weeks, reporting mean weight change of about 14.9 percent against about 2.4 percent for placebo. Companion studies tested the same drug alongside intensive behavioral therapy and tested continued dosing after an initial run-in.
Tirzepatide for weight rests on the SURMOUNT program. Its lead trial ran 72 weeks and reported mean reductions of about 15.0 percent at 5 mg, 19.5 percent at 10 mg, and 20.9 percent at 15 mg against about 3.1 percent for placebo. Later entries examined maintenance after a lead-in and reproduced the effect in a Chinese adult population.
Liraglutide has weight trial evidence and a randomized comparison against semaglutide. Dulaglutide and exenatide carry diabetes evidence, and dulaglutide’s cardiovascular indication is supported by a dedicated outcomes trial reflected on its label. Neither has weight approval, so neither has a weight evidence base to argue about.
| Drug | Brand | Mechanism | Approved indication | Anchor randomized evidence |
|---|---|---|---|---|
| Semaglutide | Wegovy | GLP-1 receptor agonist | Chronic weight management, cardiovascular risk reduction, MASH with fibrosis | STEP program, 68 weeks, placebo controlled |
| Semaglutide | Ozempic | GLP-1 receptor agonist | Type 2 diabetes, plus cardiovascular and kidney outcomes | Diabetes and outcomes program reflected on the label |
| Tirzepatide | Zepbound | GIP and GLP-1 receptor agonist | Chronic weight management, obstructive sleep apnea with obesity | SURMOUNT program, 72 weeks; separate sleep apnea trial |
| Tirzepatide | Mounjaro | GIP and GLP-1 receptor agonist | Type 2 diabetes | Glycemic control program from age 10 upward |
| Liraglutide | Saxenda, Victoza | GLP-1 receptor agonist, daily | Weight management and type 2 diabetes respectively | Weight program plus a randomized comparison against semaglutide |
| Dulaglutide | Trulicity | GLP-1 receptor agonist | Type 2 diabetes and cardiovascular risk reduction | Cardiovascular outcomes trial supporting the label |
| Compounded semaglutide or tirzepatide | None, not FDA approved | Same molecules as above | No approved indication | No clinical trial evidence for the compounded preparation |
Where direct comparison actually exists
Only one weight comparison in this class is a clean randomized head-to-head at approved weight doses. It assigned 338 adults with overweight or obesity and no diabetes to weekly semaglutide 2.4 mg or daily liraglutide 3.0 mg for 68 weeks, and reported mean weight change of 15.8 percent against 6.4 percent, with pooled placebo at 1.9 percent.
Semaglutide against tirzepatide has been examined differently. A propensity-matched cohort study using linked health record and dispensing data followed more than 18,000 matched adults on formulations labeled for type 2 diabetes and found greater on-treatment weight reduction with tirzepatide at three, six, and twelve months, with similar gastrointestinal adverse event rates. A published subpopulation analysis from a randomized comparison of the two agents reported the same direction. Neither of those is equivalent to a full randomized trial of the weight-labeled products across a general population.
The comparison people most often make is the one evidence does not support
Placing roughly 15 percent next to roughly 21 percent and calling it a result is a cross-trial comparison. Those figures came from different studies with different enrollment criteria, different durations, and different placebo responses, and the placebo arms alone differed enough to distort any subtraction between them.
The direction of that difference has held across several evidence types, which is worth something. The size of it, for a specific person at a specific dose, is not something any current publication can supply.
What is known about stopping
This is one of the better-evidenced areas. A withdrawal extension of the semaglutide obesity trial found participants regained roughly two thirds of lost weight within a year of stopping. A tirzepatide maintenance trial randomized participants after a 36-week lead-in and found continued treatment held and extended reductions while the placebo group regained substantially. Recent reviews frame that regain as expected physiology rather than treatment failure.
The practical consequence is that the evidence describes ongoing therapy. That makes affordability an evidence-relevant variable, not a separate commercial one. Cash programs from Ro, Hims and Hers, and LifeMD compete on monthly price precisely because continuation drives outcomes, and picking a compounded GLP-1 provider with physician oversight and posted pricing solves an access problem rather than an evidence one. The trials behind the published percentages were run on approved brand products, and that fact travels with the data regardless of which route someone uses.
Where the evidence is genuinely absent
Compounded preparations have no efficacy trials. Not weak trials, none. What exists is pharmacovigilance work, and a disproportionality analysis of adverse event reports from 2018 to 2024 found higher reporting odds for preparation errors, contamination, prescribing errors, and hospitalization compared with approved formulations. Disproportionality analysis cannot establish cause and rests on voluntary reporting, so it is a signal rather than a measurement.
Long-horizon data is also missing. The weight trials ran 68 to 88 weeks. For a treatment framed as chronic, that leaves questions about outcomes at five and ten years largely unanswered, particularly around lean mass, bone, and what happens across repeated cycles of stopping and restarting.
That absence does not stop a cash market from forming around the same molecules. Named services differ mainly in oversight and sourcing rather than in any proven edge, and a provider such as HealthRX lists monthly pricing for GLP-1 medications next to competitors like Henry Meds, LillyDirect, and Hims and Hers. Because no compounded preparation carries efficacy data, what distinguishes one storefront is the pharmacy behind it and whether a clinician reviews each order, not a figure from a trial.
What the evidence does support without argument
Each approved product produces clinically meaningful effects on its own endpoint against placebo. Tirzepatide improves moderate to severe obstructive sleep apnea in adults with obesity, measured by breathing events during sleep rather than by weight. Semaglutide reduces cardiovascular events in a defined population and reduces liver fibrosis measures in MASH under an accelerated approval. Gastrointestinal effects dominate the adverse event profile across the class and cluster during dose escalation.
Frequently asked questions
Does trial evidence for one brand transfer to the other brand of the same molecule?
Only partly. The molecule is identical, but dose ranges and studied populations differ between the diabetes and weight versions. Regulators treat the evidence packages separately, which is why the indications differ even though the chemistry does not.
Are the reported percentages what a patient should expect?
They are group averages under trial conditions that included structured support and close monitoring. Individual results in those cohorts spread widely in both directions, and real-world adherence is lower, so a trial mean is a reference point rather than a forecast.
How reliable is observational comparison data?
Useful and limited. Propensity matching reduces confounding but cannot remove it, since prescribing choices reflect factors no database records. Such studies are best read as consistent or inconsistent with randomized findings, not as a replacement for them.
Is there evidence that any product on the list works better for a particular type of patient?
Subgroup analyses exist but are mostly exploratory and underpowered. No approved product currently has a validated marker predicting who will respond well, which is why prescribing decisions still hinge on indication, tolerance, dosing rhythm, and access.
Sources
