FormBlends Research · formblends network
Plain-English digests of the GLP-1 and GIP/GLP-1 trials: SURMOUNT, SURPASS, STEP, SELECT, FLOW, SUSTAIN, PIONEER, retatrutide, orforglipron ATTAIN and CagriSema REDEFINE. Design, population, primary endpoint, result with confidence interval, adverse events, discontinuation, funding, and what each trial does not show, with the PubMed id on every page.
Interactive
Each runs in your browser, shows its formula, and prints a dated assumptions note. Nothing you type leaves the page.
Reading
Written for people doing the arithmetic themselves: where the mistakes happen, and how to check your own work.
Methods
The five things to check before you quote a weight-loss trial number: which estimand, what the placebo arm did, how the confidence interval reads, what a responder threshold means, and whether the endpoint was weight at all.
Updated September 4, 2026
Methods
One table for every trial digested on this site: drug, population, size, duration, primary endpoint, result with confidence interval, estimand, funding and PubMed id. Plus the interactive explorer.
Updated September 4, 2026
Trials
The pivotal tirzepatide weight-management trial. 2539 adults without diabetes, three doses against placebo, mean weight change of -15.0 to -20.9 percent versus -3.1 percent at week 72.
Updated September 4, 2026
Trials
A randomised withdrawal trial. After 36 weeks on tirzepatide, 670 adults either continued or switched to placebo for 52 weeks. Continued: -5.5 percent more. Switched: +14.0 percent regained.
Updated September 4, 2026
Trials
The first direct comparison in obesity without diabetes. 751 adults, maximum tolerated dose of each drug for 72 weeks. Tirzepatide -20.2 percent, semaglutide -13.7 percent.
Updated September 4, 2026
Trials
The pivotal semaglutide weight-management trial. 1961 adults without diabetes, 2:1 against placebo, mean weight change -14.9 percent versus -2.4 percent at week 68.
Updated September 4, 2026
Browse
Each digest starts from the PubMed abstract of the primary paper, fetched directly from the NCBI database, and where the full text is open access, from that too. We report the estimand the authors used, keep the confidence interval next to the point estimate, and separate the coprimary endpoints from the secondary ones. Numbers that appear only in press releases, conference slides or secondary reviews are left out. If a figure you expect is missing, it is because the abstract does not state it.
A digest is not a recommendation and it is not a comparison between drugs unless the trial itself was head-to-head. Population averages from selected, monitored trial participants do not predict any one person's result, and none of these trials studied compounded products, which are not FDA approved and not interchangeable with the brand products that were tested. Talk to a prescriber about what any of this means for you.
From the FormBlends network