Nobody has shown that in a randomized trial. What the randomized evidence supports is narrower, and it can be stated as a number. Pooling 99,599 patients, 66 people have to take one of these drugs for about two and a half years to prevent one major cardiovascular event [2]. The 77% reduction in all-cause death that circulates alongside it comes from a matched cohort, and its size is the reason to distrust it [1].
Most of what this site does is check whether a price is what it appears to be. The same discipline applies to a result, and the arithmetic needed here is arithmetic anyone can do — the same instinct applied to a headline price in the headline is not the bill.
What the mortality study reported
Researchers matched 12,123 people prescribed a GLP-1 receptor agonist against 12,123 who were not, among individuals with obesity and without type 2 diabetes, and followed them. All-cause mortality carried a hazard ratio of 0.23, 95% CI 0.15 to 0.34. Rates of ischemic heart disease, heart failure, arrhythmias, hypertension, stroke and atrial fibrillation were all lower, as were acute kidney injury and allergic reactions.
Read plainly, that is a 77% reduction in the chance of dying of anything at all. The authors describe it as evidence of long-term protective effects. Nothing in the paper is dishonest, and the result is what their data produced. It is also several times larger than anything in three and a half years of tirzepatide.
The eight-outcome tell
The more useful signal is breadth rather than size. Heart disease, heart failure, arrhythmia, hypertension, stroke, atrial fibrillation, kidney injury and allergic reactions are not one mechanism. A drug that improved all eight would be doing something no drug does.
What produces that pattern is a difference between the groups that existed before the prescribing did. People who get prescribed an expensive weekly injection, and who keep collecting it, differ from people who do not. They differ in how often they see a doctor, in what else they are being treated for, and in whether they can afford to. Propensity matching narrows those differences using what was recorded, and it cannot touch what was not.
The second tell: who was watched for longer
A different cohort reported an 83% lower death rate among GLP-1 users with chronic pancreatitis, across two matched groups of 1,422 [3]. One of those groups was followed 206 days longer than the other.
Time under observation is time in which a death can be recorded. A group watched for two hundred fewer days will accumulate fewer of them, whatever the drug does, and the follow-up gap is the first thing to check in any survival claim built from records.
What the randomized number looks like instead
A systematic review pooling 99,599 patients produced a number needed to treat of 66 for one major adverse cardiovascular event over roughly two and a half years. That figure belongs to trial populations far sicker than most people buying from this market.
Sixty-six people for one prevented event is a real benefit and a modest one. It is also the honest unit for any value calculation, so the arithmetic behind it is worth doing before reading what the economic studies measured.
What this has to do with a price
A claim this large is what justifies a large bill. A seller quoting a 77% mortality reduction is quoting a real paper, and a buyer has no way to tell from the quotation that the design cannot support it.
The test worth carrying is short. Ask whether the result came from a trial that randomized people. Ask whether both groups were watched for the same length of time. Ask whether the effect got bigger as the evidence got weaker. Hold all three while reading any storefront’s summary of the evidence beside its own monthly figure.