What These Drugs Do Outside a Trial

A retrospective study of 44,025 people measured what these drugs actually achieve outside a trial, alongside bariatric surgery in patients eligible for both. The gap between the published trial figures and ordinary care is not small, and the study also shows exactly where most of it goes.^[1]

The numbers

For comparison, SURMOUNT-1 reported a mean 20.9% reduction on tirzepatide over 72 weeks, and STEP 1 reported 14.9% on semaglutide over 68. Ordinary care produced roughly half to a third of that.

Where the difference goes

The study answers its own question. Restricted to people with one year of continuous therapy, the figures rise: semaglutide to 7.2%, 8.0% and 8.8%, and tirzepatide to 11.7% and 11.9%.^[1]

Continuity recovers about a third of the gap. It does not close it.

That is consistent with everything else we have found on this. Only 58% of people reach maintenance dosing and 21% never leave the starting dose, covered in most people never reach the dose. Discontinuation runs from 37% to 56% depending on the drug. A trial supplies the medicine free, escalates on protocol and chases every missed appointment; ordinary care does none of that.

What remains after adjusting for continuity is presumably the rest of it — dose escalation that stalls below target, gaps in supply, and the difference between a trial population and everyone.

And it complicates something we published

We recently covered a network meta-analysis of randomized trials that found bariatric surgery about 10 points ahead of GLP-1 drugs as a class — and, restricted to tirzepatide alone, not statistically distinguishable from surgery. That is in surgery versus GLP-1 drugs.

This study finds surgery two to three times better. Both are correct, and the reconciliation is the same one that runs through everything above: the trials being pooled in that meta-analysis had trial adherence. This cohort has real adherence. Under trial conditions the newest drug approaches surgery; under ordinary conditions it does not come close.

What it does not settle

• Retrospective and two-center. Not randomized, and surgical and medical patients differ even when both are eligible.
• Surgery carries its own costs — operative risk, permanence, lifelong nutritional consequences — none of which appear in a weight percentage.
• The tirzepatide column is shorter, reflecting a newer drug with less follow-up.
• Weight is not the only outcome. Cardiovascular and metabolic benefits are measured separately and do not scale simply with kilograms lost.