A study finding that a drug lowers medical costs is the most quotable kind of study there is, and it is the kind the affordability question most wants to be settled by. This one reports a real and sizable reduction. It also contains, in its methods, the reason to read the number carefully.
What was compared
Using a large US claims database, researchers identified people with overweight or obesity and at least one diagnosis of atherosclerotic cardiovascular disease, with twelve months of coverage on either side of an index date. [1] 770 patients who started branded semaglutide 2.4 mg were matched one to four against 3,080 randomly selected untreated controls, on demographics, clinical characteristics, prior medical costs and prior resource use.
Over the following year, treated patients incurred 22% lower mean medical costs — $4,639 less per patient per year, cost ratio 0.78 with a 95% CI of 0.67 to 0.89. Most of that came from hospital stays: inpatient costs were 65% lower at $3,593 less, cost ratio 0.35, 95% CI 0.21 to 0.49, and the inpatient visit rate was 48% lower, 0.08 against 0.15, rate ratio 0.52.
Why that matters more here than usual
Because this site has already measured how rare that is. In a clinic program where the drug was free and a team was managing patients, median persistence was under eleven months. Most people who start do not become the cohort this study measured. A saving computed on adherent patients is a saving available to adherent patients.
Medical costs are not all costs
The endpoint is all-cause medical costs and healthcare resource use. The abstract does not say whether the cost of the drug itself is netted against the $4,639, so this page cannot tell you either way — and it is the first thing a buyer would want to know, because the answer decides whether this is a saving or a transfer.
There is also a product difference that cannot be waved away. This studied branded semaglutide 2.4 mg at US list-adjacent prices. Most of what this roster sells is compounded, at a fraction of that. Those are not the same purchase, and a cost result computed on one does not port to the other — which is a separate question from what a cardiovascular result costs in the first place.
What it is good evidence for
Something narrower than the headline and still worth having. Among people with established cardiovascular disease who take this drug consistently for a year, hospital use goes down substantially — a 48% lower admission rate is not a small signal, and it points the same direction as the randomized cardiovascular trials rather than resting on its own.
What it is not good evidence for is that buying this drug will save you money. That claim needs the drug price included, a population like yours, and an adherence assumption you can actually meet. None of those three is established here, and the number needed to treat is the more honest unit for a low-risk buyer. If you want the arithmetic that does apply, start with what six months actually costs.