Better than safe, on the evidence. Three large randomized trials were pooled at participant level, covering 30,787 people over more than three years. The primary kidney composite favored semaglutide. There were 973 events against 1,134, a hazard ratio of 0.84.
That composite counts cardiovascular death as one of four components. Strip it out and the kidney-only result holds. It ran 347 events against 416, hazard ratio 0.80.
Severity changes the size of it. A FLOW analysis broke the result down by stage of chronic kidney disease [2]. The evidence is thinner at the edges of the population. After a transplant it comes down to 39 people.
Cost and benefit part company here. A German model of second-line diabetes drugs for preventing nephropathy put tirzepatide first on outcome and last on value [3]. That is the best result losing on the money.
A pooled analysis of three trials is stronger than any one of them, and harder to read. This one is worth the effort. Part of that is what it includes. One of the three trials also used the tablet, the format this roster prices separately and usually charges differently for.
What was pooled
Participant-level data from SELECT, FLOW and SOUL, prespecified rather than assembled afterward. [1] One trial enrolled people with chronic kidney disease and used 1.0 mg semaglutide by weekly injection. One enrolled people with established atherosclerotic cardiovascular disease and used 2.4 mg weekly by injection. The third also enrolled people with cardiovascular disease and used 14 mg of oral semaglutide daily. Together, 30,787 participants with a mean follow-up between 39.5 and 47.5 months.
The result, and the component that is not a kidney
The primary endpoint was a composite. A persistent 50% or greater fall in kidney filtration, kidney failure, kidney-related death, or cardiovascular death. On that, 973 first events occurred on semaglutide against 1,134 on placebo. The hazard ratio was 0.84, 95% CI 0.77 to 0.91.
One of those four components is death from a heart cause, which is not a kidney outcome by any reading. Including it in a kidney composite is defensible. Cardiovascular death is a competing risk. It removes people from the chance of reaching a kidney endpoint. It also means the headline number is not purely about kidneys. How an endpoint is assembled does more work than readers notice, and so does which comparison a trial chooses to publish.
What the pooling costs
Three doses, two routes of administration and two different patient populations went into one estimate. The authors say so plainly, noting differences in baseline characteristics and in dose and route.
This is the trade every pooled analysis makes. You gain precision and lose specificity. The hazard ratio of 0.84 describes none of the three trials on its own. It does not say whether 1.0 mg injected and 14 mg oral produce the same kidney benefit. Nobody ran that comparison. What the pool shows is that an analysis including an oral arm came out favorable. That is not the claim that the tablet matches the injection. This page will not make the second one.
It matters commercially because the two formats are not priced alike here. Take only the 71 sellers publishing a figure for both, read September 2026. The tablet runs a median of $9 more a month than the injection. Of those sellers, 37 charge more for it. Whether that gap reflects anything about outcomes is exactly what nobody has tested, and the price check shows the current spread.
Who funded it, and who it describes
Novo Nordisk, which makes semaglutide and ran all three trials. The pooled analysis was prespecified, which is the safeguard that matters most, and the funding is still worth knowing.
The population is people with cardio-kidney-metabolic disease — kidney disease or established cardiovascular disease, mostly with diabetes. If you are buying for weight loss, this describes somebody else’s risk profile. The honest unit for your own is how many people have to be treated for one event to be prevented. In a lower-risk person that number gets much larger.