Somebody had to ask this question eventually, because health systems buying at national scale now have emissions targets attached to their procurement, and a weekly injection manufactured, refrigerated and shipped for years is not a small object in that arithmetic. The answer the model returns is interesting mostly for the part that does not fit the conclusion, which is a pattern this site keeps meeting in modeled evidence generally.
What was modeled
A cohort state-transition model, fed with results from the STEP and SELECT trials, estimated per-patient emissions across a lifetime for semaglutide 2.4 mg added to diet and exercise, against diet and exercise on their own, in three UK populations. [1] Nothing here was weighed or metered; every figure is an output of the model’s assumptions about how obesity-related complications would otherwise have unfolded.
In the two populations defined by body mass index and complications, the drug came out ahead on everything the model tracked: 0.30 and 0.25 additional life-years, 0.51 and 0.46 additional quality-adjusted life-years, and lifetime emissions lower by 1.8%. The manufacturing footprint of the drug was more than offset by the complications the model expected it to prevent.
The third population
Among people with a body mass index over 27 and established cardiovascular disease, the clinical results were the best of the three — 0.56 additional life-years and 0.59 additional quality-adjusted life-years — and the carbon result went the other way, with modeled lifetime emissions of 14,700 kg CO2e against 14,444 for the comparator. The gap of 256 kilograms, about 1.8% higher, is this desk’s subtraction from the paper’s two published figures.
The stated reason is longer survival and more monitoring. People who do not die of a heart attack go on consuming healthcare, and healthcare emits.
Who wrote it
Two of the authors are employees and shareholders of Novo Nordisk, which makes semaglutide. One sits on the company’s Sustainability Advisory Council under contract. Five are employees of IQVIA, and one works for a health-economics consultancy that has received Novo Nordisk funding in the past three years.
That is the entire author list, and the disclosure is published in full, which is the right way to do it. None of it makes the model wrong — a manufacturer usually holds the data nobody else has — but a reader deciding how much weight to give a favorable environmental conclusion is entitled to know that no independent party checked the assumptions that produced it.
What it has to do with buying one
Almost nothing, which is worth saying plainly on a site whose readers are paying for these drugs out of pocket. This paper argues to a health system deciding what to stock, and a life-year in a model is not a promise to any individual, for the same reason a modeled clock is not a lifespan.
What it is good for is calibration. When a study arrives showing that a drug is good on an axis nobody was arguing about, the question to ask first is who needed that answer — the same question worth asking about any ranking that flatters a seller.