In the high summer of 2025, between July and August, the Centers for Disease Control and Prevention put a set of questions to 8,118 American adults through its Rapid Surveys System. The answers, published on November 12 in the agency’s Morbidity and Mortality Weekly Report, describe a country that has taken up a new class of medicine in enormous numbers and then, without much announcement, started rationing it for itself.

The headline figure is simple. According to the CDC report, 9.8 percent of adults said they had used a glucagon-like peptide-1 receptor agonist in the past year. As Beth Mole reported for Ars Technica, that works out to roughly 27 million people. These are the peptide drugs now prescribed for diabetes and weight loss, and increasingly for related chronic conditions such as cardiovascular disease and obstructive sleep apnea. They work very well. They are wanted everywhere.

Twenty-seven million is a nation’s worth of patients. The survey asked what they actually did with the medicine.

The household arithmetic

Of the 976 respondents who reported using a GLP-1, 44 percent said they had taken it in some way other than the way it was prescribed. Skipping doses was the most common change, at 28 percent. Others put off refilling the prescription, or took less than the doctor had ordered. A prescription is written as a schedule and a dose. For close to half of these users it had turned into something closer to a suggestion, adjusted to whatever the month allowed.

Two dark blue hands stretch a cream-colored syringe into a thin middle with a red streak against a yellow background.

The authors were clear about why. Among those who modified their use, 38 percent named cost and 36 percent named availability. Ars Technica combined the two and reported that more than half gave one or both reasons. The report’s own summary puts it soberly: “Modifications in medication use reflect, in part, responses to cost- and access-related challenges.”

The burden fell hardest on the young. Among adults aged 18 to 39 who changed how they took the drug, 58 percent cited cost; among those 65 and older, only 20 percent did. (Ars Technica printed the younger bracket as starting at 14; the CDC’s tables begin at 18.) In that youngest group, 54 percent of users modified their use at all, against 30 percent of the oldest. Thirty-seven percent of the younger users delayed filling a prescription. Among those 65 and over, the figure was 6 percent.

The pattern has an obvious explanation, though the report does not spell it out. Americans over 65 have Medicare. Younger adults have whatever their job, their plan or their bank balance gives them. The CDC found that only 4.8 percent of adults under 40 used these drugs at all, compared with 13.7 percent of those aged 40 to 64. The young are the smallest group of users, and they are the most likely to stretch the medicine to make it last.

A cream hourglass with yellow sand wears a navy belt with a gold buckle, its loose end flowing right against a bright blue background.

The pharmacy outside the gate

Then there is the medicine that comes from somewhere else. Among all users, 22.6 percent said they had taken a compounded version, which Ars Technica rounded to “at least 23 percent.” A further 8.8 percent did not know whether theirs was compounded. That second number deserves attention: nearly one user in eleven was injecting a potent drug without knowing where it came from.

The report is blunt about what that means: “Compounded GLP-1 medications are not reviewed by the Food and Drug Administration for safety, effectiveness, or quality.” Here, too, the young were overrepresented. Twenty-eight percent of users aged 18 to 39 took a compounded version, compared with 18.6 percent of those 65 and older. Compounded users also bent the rules more often: 62 percent modified their use, against 36 percent of those taking other versions.

Seen from the patient’s side, the two findings form one story. Someone priced out of the branded drug turns to the cheaper copy, and then cuts back even on that. The drug is unregulated, and the dose is improvised. These are the conditions under which the government has lately had to take a closer look at the class: regulators have ordered a vision-loss warning on the branded products, and the federal health department has been revising its rules on peptides sold outside the usual trials. The safety record of these drugs comes from trials in which people took them as directed. Many people are not taking them that way.

A red-handled dropper releases two cream-colored dice toward a dark blue spoon against a teal background.

What the survey cannot see

The authors list the study’s limits plainly. The panels behind it had cumulative response rates of 13.3 percent and 3.8 percent, and they warn that such “online panel surveys have low cumulative response rates, which could affect generalizability.” The surveys “might be less accessible to persons without reliable internet or computer access,” which may leave out exactly the people for whom cost bites hardest. And “measurement error from response or recall biases was possible.” The numbers are estimates and should be read as estimates.

Not every change was about money. About 12 percent of users reduced or stopped their medication because of side effects or symptoms, which is a reminder that these drugs are hard on some bodies and that patients sometimes adjust for reasons no pharmacy bill explains.

As for what should follow, the authors write that the findings “might signal an opportunity for clinicians and health care payers to better understand patient experiences,” and that “future research could evaluate the implications of real-world medication use.” In other words, the agency measuring the gap between the prescription and the patient is not yet in a position to say what that gap costs. Insurers and prescribers will set the price and the dose. Millions of patients will go on deciding, week by week, how much of either they can afford.