Eli Lilly's orforglipron, the oral GLP-1 pill the company sells as Foundayo, matched injected insulin on major heart events in a 2,749-person trial. Over two years, 57 of the people on the pill (4.2 percent) had a cardiovascular death, heart attack, stroke, or hospitalization for unstable angina, against 67 (5.0 percent) on insulin glargine. That is the headline from ACHIEVE-4, published September 30 in The Lancet ↗01865-9) and presented the same day at the EASD meeting in Milan. What the trial proved is narrower than how it is being sold.
ACHIEVE-4 was a safety trial, not a benefit trial. Its job was to rule out that orforglipron raises heart risk, not to show it lowers it. The design says so: participants were randomized against insulin glargine, a drug already known to be neutral on cardiovascular outcomes, and the statistical bar was non-inferiority with a margin of 1.8. In plain terms, the pill passed if the data could rule out an 80 percent relative increase in events. It cleared that bar easily (hazard ratio 0.84, p less than 0.0001 for non-inferiority).
What the confidence interval allows
The point estimate, a hazard ratio of 0.84, is where the benefit story comes from. Lilly's release ↗ frames it as a 16 percent lower risk of four-component heart events. But the 95 percent confidence interval runs from 0.59 to 1.20. Read plainly, the data are compatible with anything from 41 percent fewer events to 20 percent more. An interval that crosses 1.0 does not establish benefit. It establishes that the drug did not obviously make things worse, which is exactly what a safety trial is built to show.
That distinction is not pedantry. The injected peptide GLP-1 drugs earned their cardiovascular claims the hard way, in dedicated outcome trials that beat placebo. Semaglutide ↗ did it in SELECT and SUSTAIN-6, dulaglutide ↗ in REWIND, liraglutide ↗ in LEADER. Each showed a real reduction against an inert comparator. Orforglipron, the first non-peptide agonist of the GLP-1 receptor ↗ to run a cardiovascular trial, has so far shown only that it does not raise risk against a comparator that was neutral to begin with. Safety is a real result. It is not the same result.
The mortality number needs a caveat
The most striking figure is deaths. There were 19 (1.4 percent) on orforglipron and 43 (3.2 percent) on insulin, which the EASD presentation ↗ put at a hazard ratio of 0.43. Almost all of the deaths in both arms were judged unrelated to the study drugs. The number is large and it points the right way, but it is a secondary endpoint, not corrected for the many comparisons the trial ran, and the trial was open-label: everyone knew who was taking a pill and who was injecting insulin. A nominal mortality signal from an unblinded safety trial is a reason to run a superiority trial, not a reason to claim one was already won.
The safety profile otherwise tracked the class. Gastrointestinal side effects hit 62.1 percent on orforglipron versus 14.2 percent on insulin, the most common reason people stopped the pill. The trade ran the other way on low blood sugar: clinically significant or severe hypoglycemia occurred in 6.8 percent on the pill against 19.2 percent on insulin, the expected penalty of titrating insulin to target.
Two weeks ago this section covered a Lilly post hoc analysis that said orforglipron lowered a predicted risk score while counting no actual events ↗. ACHIEVE-4 is the trial that finally counted events, 124 of them across both arms. The counting matters, and the verdict it returns is clean: an oral pill that controls blood sugar without the hypoglycemia of insulin and without a heart-safety penalty. For a once-daily tablet in people with established heart or kidney disease, that is worth something. It is just not the protection the word "benefit" implies, and the gap between the two is the whole story.