Nobody ran the trial that would answer it. The two classes have each been proven against placebo and never against each other, so the comparison below is an inference from separate studies [1]. Setting one class’s placebo-controlled result against the other’s puts major cardiovascular events at 1.05, 95% CI 0.92 to 1.20. That is a statement about the evidence rather than about the drugs, and it is not a comparison anybody actually performed.
What each class did against placebo
Eleven randomized trials enrolled 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. GLP-1 drugs did too, at 0.83, 95% CI 0.77 to 0.89.
Those are direct comparisons against placebo. They are the solid part of this literature, and the half a cost-per-event figure can be built on, as the number needed to treat sets out.
What comparing them requires
It requires 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, p = 0.44. Heart failure admission came out at 0.83, 95% CI 0.67 to 1.03, p = 0.09, favoring SGLT2 inhibitors without reaching significance. Myocardial infarction came out at 1.05, 95% CI 0.83 to 1.33.
Not significant is also not equal. The interval for major events runs 0.92 to 1.20 and the one for heart attack runs 0.83 to 1.33. Both ranges include differences a person would care about, in both directions, which is the distinction every non-inferiority result turns on.
Where the two classes do separate: price
A German Markov microsimulation modeled kidney disease progression across four options [2]. Quality-adjusted life years ran 10.28 on standard care, 10.48 with empagliflozin, 10.58 with semaglutide and 10.75 with tirzepatide. On the health axis the ordering is exactly what a clinician would expect.
The cost column runs the other way. Modeled lifetime direct costs were €66,343.64, €79,767.42, €84,328.59 and €155,992.59. Net monetary benefit came out at €961,445.98, €968,406.95, €973,647.74 and €919,972.95. Semaglutide is the model’s winner, and tirzepatide is the only option scoring below standard care.
Those are modeled lifetime totals for a simulated 62.8-year-old German cohort, built on German drug and complication costs. None of them is a price a US self-pay buyer faces. What carries across is the structure. A more effective drug that costs enough more can be worse value while being better medicine, which is the mechanism behind what insurance actually covers.
Taking both is a third option, and harder than it reads
A pragmatic trial randomized 173 insured adults to an SGLT2 inhibitor, a GLP-1 drug, or both. They filled the prescriptions through their own insurance [3]. By four months, 84% assigned one drug had filled it against 53% assigned two, P < .001. Over a ten-month median the figures were 87% against 68%.
Among those who filled, 22% on one drug and 49% on two discontinued a study medication, mostly because of side effects, P = .002. Each additional medication is another chance for the chain to break. The fill step alone is measured in whether people fill a GLP-1 prescription at all.
What a buyer does with this
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. The combination question, where it arises, is covered in taking a GLP-1 alongside an SGLT2 inhibitor.
The evidence base is also narrower than the question sounds. The authors note it is largely drawn from remote myocardial infarction, which limits what it says about the period immediately after a heart attack. What the GLP-1 side of it establishes on its own is in the cardiovascular result.