Two drug classes, both proven against placebo, and a question nobody has run the study to answer. Everything below is inferred by comparing each class’s separate trials with each other — which is a legitimate method and is not the same as a comparison anybody actually performed.
What each class did against placebo
In people with type 2 diabetes and a history of myocardial infarction, SGLT2 inhibitors reduced major adverse cardiovascular events, hazard ratio 0.87, 95% CI 0.77 to 0.97. [1] GLP-1 drugs did too, 0.83, 95% CI 0.77 to 0.89. Those are direct comparisons against placebo and they are the solid part — the half of this literature that a cost-per-event figure can actually be built on.
What comparing them requires
Setting one class’s placebo-controlled result against the other’s, and assuming the two sets of trials enrolled similar enough people for the subtraction to mean something.
On that basis, major cardiovascular events came out at 1.05, 95% CI 0.92 to 1.20, p = 0.44. Heart failure admission at 0.83, 95% CI 0.67 to 1.03, p = 0.09, favoring SGLT2 inhibitors without reaching significance. Myocardial infarction at 1.05, 95% CI 0.83 to 1.33.
Not significant is not equal
The interval for major events runs 0.92 to 1.20. For heart attack, 0.83 to 1.33. Those ranges include differences a person would care about in both directions.
So the finding is that the evidence cannot distinguish the classes, which is a statement about the evidence rather than about the drugs — the distinction every non-inferiority result turns on.
And the population is narrower than it sounds
The authors note the evidence is largely drawn from remote myocardial infarction, limiting what it says about the period immediately after a heart attack.
Nobody buys an SGLT2 inhibitor from a telehealth seller, so this is not a choice made at a checkout — it is a prescribing decision, and the authors say it should be individualized. What makes it worth reading here is the shape: two widely used classes, years of trials, and the comparison that would settle it still waiting on somebody to run it.