A 2025 study looked for compounded semaglutide and tirzepatide in the medical records of primary care patients across the United States (n=153,044), documented in 8.2% of them [1]. The share rose over the study window, which ran from January 2021 to December 2024.
How that number was obtained matters more than the number. Brand-name prescriptions were identified from structured data, the coded fields every system records automatically. Compounded use could not be, because a compounded preparation does not travel through the channel that produces a structured prescription record. It was found instead by reading clinical notes — free text, written by a clinician who happened to mention it. That is the same visibility problem the gaps in what sellers publish create from the other direction.
So 8.2% is not a measurement of how many people were using compounded GLP-1 drugs. It is a measurement of how often somebody wrote it down. Anyone who bought from an online seller and never told their primary care physician, or told one who did not record it, is not in that figure. The real proportion can only be higher, and nothing in the study can say by how much.
Two differences between the groups are worth reporting carefully. Compounded-only patients stayed on treatment longer on average, 10.0 months vs 7.8 months for brand-name-only, and were more likely to be female. Neither is evidence that compounded works better. Cost, refill mechanics and who selects a compounded route in the first place all plausibly produce that gap, and a retrospective cohort cannot separate them. How long people actually stay on these drugs is its own question, taken up in what happens when they stop.
The window closed at the end of 2024, during the semaglutide and tirzepatide shortages that made large-scale compounding lawful. Those shortages have since been declared over, and the market changed with them, so this describes a period rather than the present. What survives the change is the measurement problem: the people buying this way are the hardest people to count, which is worth remembering whenever a national figure for GLP-1 use is quoted.
It also bears on how this site is read. Almost every seller ranked here dispenses a compounded preparation, and the care standards comparison and what sellers publish both assume a buyer who may never appear in a dataset like this one at all.