A trial is a description of what a drug can do under conditions almost nobody experiences: selected patients, free medication, study staff, a protocol for escalation and a schedule for checking in. The question a reader actually has is what it does without any of that. This review is the best available summary, and its answer is a shortfall with a specific cause — one that speaks directly to what is counted in what the semaglutide trials are worth.
What the gap is
Observed weight reduction in clinical practice tends to be lower than in the randomized trials [1]. The interesting part is the qualifier: outcomes approach trial levels when the analysis focuses on highly adherent patients.
That locates the gap. It is not that the drug behaves differently outside a trial; it is that people take it differently. Two mechanisms are named — high discontinuation, reported at 20% to 50% within the first year, and the use of much lower doses than the trials evaluated.
One cohort, for scale
A single matched study puts a concrete figure inside that range. In a cohort of 18,386 propensity-matched adults starting one of these drugs in ordinary care, follow-up ended in discontinuation for 55.9% of those on tirzepatide against 52.5% of those on semaglutide[2]. More than half of each group stopped.
That is one study rather than the literature, and it sits at the upper end of the range the review describes. It is included because a range from a fifth to a half is hard to hold, and a single well-matched number gives it a shape.
The half that is reassuring
The same review reports that observational studies in diabetes and obesity populations found frequent gastrointestinal disturbance — consistent with the trials — but no clear increase in the risks that generate the most alarm: pancreatitis, pancreatic cancer, thyroid disorders, or depression and self-harm.
That belongs in the same summary as the shortfall. An article reporting only that the drug underperforms outside a trial, while omitting that the feared harms did not materialize outside one either, would be selecting evidence rather than reporting it. The eye-disease question is named as still needing evidence, and we cover the two strands of it in sudden vision loss and diabetic retinopathy.
What it means for someone buying online
The two named causes of the gap are both worse in this market than in ordinary care. Doses are frequently unpublished, so a buyer often cannot tell where on the ladder they are — the silence counted in who publishes a dose ladder. And discontinuation is driven substantially by cost, which is the whole subject of this category.
The practical reading is that trial figures are a ceiling rather than an expectation, and that most of the distance between the ceiling and the floor is decided by whether a person keeps taking the drug at an adequate dose. That is a question about price and about supervision, neither of which most sellers here describe.