Who gets a drug is a different question from whether it works, and it is usually the one that decides outcomes at a population scale. For an expensive drug in the United States, the answer tends to be the people who can pay — which is why what it actually costs is a question about access, not only about value.
What the survey found
The analysis used the 2022 Medical Expenditure Panel Survey, covering 2,542 adults diagnosed with type 2 diabetes and representing roughly 30.3 million US adults. [1] After adjusting for insurance status, education and employment, people with incomes above the median had higher odds of using semaglutide than those below it — an odds ratio of 1.61 with a 95% CI of 1.14 to 2.26. Having private insurance carried an odds ratio of 1.52, 95% CI 1.04 to 2.22.
Both intervals sit above one, so both associations are real in this sample. Note what an odds ratio is not: it is not a percentage, and the study reports no rates to turn it into one. It says the odds differ by that factor, adjusted, and no more.
Why 2022 matters
The survey predates most of what this site covers. In 2022 the cash-pay compounded market was barely a consumer phenomenon, and the realistic ways to obtain these drugs were a covered prescription or a very large bill. So this is a good description of access under insurance and a poor description of buying in 2026.
That shift cuts both ways. A compounded market lowered the entry price substantially — sellers here publish a median of $179 a month for semaglutide by injection, read September 2026 — which puts the drug within reach of people no insurer was going to cover. It also moved the purchase outside the system that would have checked whether it was a good idea, and what this market actually sells is partly that trade.
What it means for a buyer
Directly, not much — you are not a population statistic. Indirectly, two things worth holding.
The first is that coverage remains the single largest determinant of what this costs anybody, which is why the coverage question is worth exhausting before assuming cash is the only route. The second is that where you live changes the answer as much as what you earn, and the state view is the fastest way to see it.
There is a policy argument in the paper too — subsidy programs, expanded coverage, reference pricing. This site has no view on any of that. It notes only that a market where income predicts who gets treated is the market these prices exist inside, and that the worth-it question has a different answer at different incomes.