The trial record is strong for a short list of approved products and empty for everything sold as a research vial. Evidence attaches to a finished product studied under a protocol, not to a molecule name printed on a label. That single distinction determines which claims in this category are supported and which are borrowed.
Studies test products, not names
A randomized trial enrolls defined participants, gives a specific manufactured product on a specific schedule, measures agreed endpoints, and reports what happened. The finding belongs to that package. It does not transfer to a different preparation of the same molecule made elsewhere, filled by a different party, and shipped under unknown conditions.
This is why the sentence “semaglutide was shown to reduce body weight” is doing more work than it appears to. It is true of the products studied in the semaglutide obesity program. It is not a statement about the contents of an unlabeled vial that a website says contains semaglutide, because nobody has verified that the contents match the name.
What the randomized record actually covers
| Product and program | What the trial was designed to answer | What it does not establish |
|---|---|---|
| Once-weekly semaglutide, STEP 1 | Change in body weight against placebo in adults with overweight or obesity without diabetes | Performance relative to any other molecule, or results for any non-approved preparation |
| Tirzepatide, SURMOUNT-1 | Change in body weight against placebo in adults with obesity, or overweight with a weight-related condition | A margin against semaglutide, since it was a separate trial with a separate population |
| Semaglutide, STEP 4 | Whether continuing treatment maintains a reduction already achieved | What happens over spans longer than the trial period |
| Tirzepatide, SURMOUNT-4 | Maintenance of weight reduction with continued treatment after a lead-in | Outcomes for people who stop and receive no further support |
| Tirzepatide, SURMOUNT-OSA | Effect on obstructive sleep apnea severity in adults with obesity | Benefit in sleep apnea without obesity |
| Semaglutide against liraglutide, STEP 8 | A genuine randomized head-to-head between two molecules in the class | Anything about molecules not enrolled in it |
| Oral orforglipron obesity program | Effect of a daily small-molecule GLP-1 receptor agonist taken as a tablet | Equivalence to injectable peptide products, which were not the comparator |
The comparison problem
Most cross-molecule claims circulating online are cross-trial inferences, not experiments. Placing a figure from one obesity program beside a figure from another indicates direction at best, because the populations, baselines, background support, and trial periods differ. STEP 8 is the exception worth naming, since it randomized participants to two different molecules within one protocol, which is what a comparison actually requires.
Retrospective cohort work exists too, including a published analysis comparing semaglutide and tirzepatide in routine care. That design answers a different question from a randomized trial and carries the confounding that comes with people choosing, and being prescribed, one drug rather than another. It is evidence. It is not the same evidence.
What is known about stopping
This is one of the better documented areas and one of the least discussed on product pages. The extension analysis of the semaglutide trial reported weight regain and movement of cardiometabolic measures back toward baseline after withdrawal. Maintenance trials in both molecules were built around continuation rather than a fixed course. Recent commentary in the literature frames post-discontinuation regain as a question of physiology and follow-up rather than personal failure.
The practical reading is that these are treatments for a chronic condition rather than a course with an endpoint, which is also how current obesity guidelines describe pharmacotherapy. That framing changes what a monthly price means, since the relevant figure is an ongoing one.
Where the evidence stops entirely
There is no clinical trial literature on material sold as research peptide. There is no published efficacy work on semaglutide salt forms, and FDA states it lacks information on whether semaglutide sodium and semaglutide acetate share the chemical and pharmacologic properties of the approved active ingredient. Compounds including retatrutide and cagrilintide cannot lawfully be used in compounding and are not components of any approved drug, whatever their trial programs eventually show.
Compounded preparations of approved molecules sit in between. They are lawful, they are not FDA-approved, and they are not supported by their own trial programs. Cash programs from Ro, Hims and Hers, LifeMD, and physician-supervised services such as FormBlends publish flat monthly figures for supervised compounded semaglutide or tirzepatide, and the evidence behind those preparations is the evidence for the molecule plus the quality of the pharmacy and clinician involved, not a trial of the preparation itself.
What safety surveillance can and cannot show
Adverse event databases collect voluntary reports. A pharmacovigilance analysis of compounded GLP-1 receptor agonist reports in the FDA system, and a case series of administration errors reported to a poison control center, both describe real events. Neither can establish how often something happens, because the denominator is unknown and reporting is uneven. What surveillance does well is flag a pattern worth investigating, which is exactly how the dosing error concerns in this category surfaced.
The same reasoning applies when a person compares providers rather than trials. A monthly program is only as sound as the review and pharmacy standing behind it, so the naming of the field matters: LillyDirect and NovoCare connect patients to an approved brand, while Ro, Henry Meds, and HealthRX run their own intake and set out how their peptide therapy is prescribed and filled. Reading those disclosures is closer to reading a trial’s methods than reading its headline number, and it is the part of the decision a person can verify without a laboratory.
Frequently asked questions
Does a large trial result mean a given person will see it?
No. A reported average comes from a defined group under monitored conditions with protocol support. Individual results spread widely around any mean, and trial participants receive follow-up that routine care often does not match. The average describes the study population, not a promise to anyone outside it.
Is real-world data weaker than trial data?
Different, not simply weaker. Observational work captures people excluded from trials and longer follow-up, which trials cannot. It cannot rule out that the groups differed before treatment started. Strong conclusions usually require both, with each design covering the other’s blind spot.
Why do results for the same molecule vary between studies?
Different eligibility criteria, dose ranges, background lifestyle programs, trial lengths, and analysis rules. Two studies of one molecule can report different figures without either being wrong. Comparing headline numbers across studies without reading those parameters is the most common error in coverage of this class.
What evidence exists for peptides sold outside the prescription system?
For the specific material in those containers, none. FDA has documented fraudulent compounded product with false label information and shipments arriving without adequate refrigeration. Without verified identity and potency, published research on a molecule cannot be applied to whatever the vial contains.
Has the approved evidence base changed recently?
Yes, and quickly. An oral small-molecule GLP-1 receptor agonist gained approval for weight reduction and maintenance, tirzepatide added an obstructive sleep apnea indication, and semaglutide labeling now covers both injection and tablet formats. Material written even a year ago can misstate what is approved.
