Skip to content
GLP Loss
← Research
Evidence

SURPASS-CVOT Explained: Noninferior to a Drug That Already Works

SURPASS-CVOT randomized 13,299 patients to tirzepatide or dulaglutide 1.5 mg and returned a hazard ratio of 0.92 whose interval stopped at 1.01 — superiority missed at P = 0.09. The pre-specified kidney analysis, run on the same randomization, cleared its bar at 0.77.

Owen Castellanos9 min read
Two pre-specified analyses, two verdicts13,299 randomized, median 4.0 years, active comparatorHeart: death, heart attack or stroke0.92upper limit 1.01Superiority test returned P = 0.09Kidney: composite of four events0.770.68 to 0.88Cleared its bar at p = 0.0002The comparator was not a placeboDulaglutide 1.5 mg, given weekly for up to five yearsOne third of the maximum approved doseAlso the only dose with an outcome trial behind itNoninferiority margin 1.05. It was met.Events: 801 of 6,586 against 862 of 6,579.Stomach and bowel side effects: 42.5% against 35.9%.

Almost every headline GLP-1 outcome trial answers an easy question: is the drug better than nothing. SURPASS-CVOT answered a harder one. It put tirzepatide against dulaglutide — a drug that had already beaten placebo on cardiovascular events in its own trial — and asked whether the newer molecule is better than the one already in the cabinet. The answer came back as a hazard ratio of 0.92 whose confidence interval stopped at 1.01, which is to say it did not come back at all.[1] Understanding why one hundredth of a point matters here is most of what there is to understand about this trial.

The comparator, and why its dose is the argument

The control arm received dulaglutide at a fixed 1.5 mg weekly for up to five years.[2] Two facts about that dose pull in opposite directions, and a page that prints only one of them is arguing rather than reporting.

The first: 1.5 mg is not the ceiling. A separate randomized trial in 1,842 metformin-treated patients compared dulaglutide 3.0 and 4.5 mg against 1.5 mg and found 4.5 mg superior at 36 weeks — glycated hemoglobin fell 1.77 points against 1.54 on the treatment-regimen estimand, a difference of 0.24 points, with weight down 4.6 kg against 3.0 kg.[7] Against the maximum approved comparator dose, the gap tirzepatide had to clear would have been narrower still.

The second, which is the one that makes the trial defensible: 1.5 mg is the only dulaglutide dose with a cardiovascular outcome trial underneath it. In 9,901 patients followed a median of 5.4 years, dulaglutide 1.5 mg reduced the same three-part composite against placebo — 594 events (12.0%) against 663 (13.4%), hazard ratio 0.88 (95% CI, 0.79 to 0.99; P = 0.026).[6] Choosing 4.5 mg would have meant randomizing thousands of patients against a dose with no proven outcome benefit. The trial chose the proven dose, and that choice is why its result is hard rather than unfair.

What was run

SURPASS-CVOT was an active-comparator-controlled, double-blind, noninferiority trial in adults with type 2 diabetes and atherosclerotic cardiovascular disease, randomized 1:1 to weekly tirzepatide escalated to a maximum tolerated dose of up to 15 mg or to dulaglutide 1.5 mg. A total of 13,299 patients were randomized; 134 were later excluded for not meeting inclusion criteria, leaving 6,586 and 6,579 in the two groups. Mean age was 64.1 years (SD 8.8), 29.0% were women, mean body-mass index was 32.6 (SD 5.5), mean glycated hemoglobin was 8.4% (SD 0.9) and mean diabetes duration was 14.7 years (SD 8.8).[1] That is a diabetes population in the sense the two-indications article describes — fifteen years into the disease, on background therapy, and enrolled because of their arteries rather than their weight. Median follow-up was 4.0 years (IQR 3.7 to 4.4),[3] and median time actually on treatment was 46.9 months (IQR 34.6 to 50.6).[4] Those are two different durations and the second is the one that describes exposure.

One hundredth of a point

The primary endpoint was a composite of death from cardiovascular causes, myocardial infarction or stroke. It occurred in 801 patients (12.2%) on tirzepatide and 862 (13.1%) on dulaglutide: hazard ratio 0.92, 95.3% confidence interval 0.83 to 1.01, P = 0.003 for noninferiority against a margin of 1.05.[1] The protocol set superiority at an upper confidence limit below 1.00. The observed upper limit was 1.01, and the superiority test returned P = 0.09.

That is not a near miss to be rounded away. A confidence interval running to 1.01 is compatible with a small excess of events on tirzepatide, and the trial’s own pre-declared rule says so. What it establishes is that tirzepatide is not worse than a drug that works — a conclusion worth having, and a different sentence from the one marketing tends to write.

The event columns the registry posts

Since the paper is paywalled, the per-outcome breakdown has to come from the trial’s posted results, and those results carry counts without any ratio or significance test attached. Over the full follow-up window, death from any cause was recorded in 566 patients on tirzepatide against 669 on dulaglutide — 8.6% against 10.2%. Cardiovascular death ran 367 against 408, myocardial infarction 311 against 357, stroke 229 against 249, coronary revascularization 527 against 617, and new or worsening nephropathy 373 against 442. The expanded four-part composite occurred in 1,089 against 1,217.[2]

Every one of those columns leans the same direction, which is the honest reason the primary result reads as it does: a consistent small advantage spread across many outcomes, none of it individually tested here. A published post hoc analysis bundled six of them — all-cause death, myocardial infarction, stroke, revascularization, heart-failure hospitalization and the kidney composite — and reported 1,559 events (23.7%) against 1,803 (27.4%), hazard ratio 0.84 (95% CI, 0.79 to 0.90; P < .001).[4] A composite assembled after the results are known is a description, not a test, and the authors label it that way.

The endpoint that did clear its bar

The trial carried a second pre-specified exploratory analysis, on kidneys rather than arteries, and it did not miss. Among the 13,165 analyzed patients — 2,948 of whom had high-risk chronic kidney disease at baseline — the composite of persistent macroalbuminuria, a persistent 50% or greater fall in estimated glomerular filtration rate, end-stage kidney disease or death from kidney disease occurred in 396 (6.0%) against 498 (7.6%): hazard ratio 0.77 (95% CI, 0.68 to 0.88; p = 0.0002). In the low-to-moderate-risk group it was 0.70 (0.58 to 0.84) and in the high-risk group 0.79 (0.64 to 0.96).[3]

The mechanism differed by group, which is the part worth noticing: the benefit in lower-risk patients came from fewer new cases of persistent macroalbuminuria, and in higher-risk patients from a slower fall in filtration rate. The annual difference in filtration decline was 0.29 mL/min per 1.73 m² overall and 0.93 in the high-risk group.[3] The registry’s own 36-month columns point the same way: urinary albumin-to-creatinine ratio fell 26.42% against 7.23%.[2] What this means for anyone with reduced kidney function is set out in the kidney disease article.

The placebo that was never randomized

Because there was no placebo arm, the question buyers actually ask — how much does tirzepatide reduce events compared with not taking it — cannot be answered by this trial at all. A pre-specified indirect analysis tried anyway, multiplying the tirzepatide-versus-dulaglutide ratio from this trial by the dulaglutide-versus-placebo ratio from the earlier outcome trial, using 2,055 of the 9,901 earlier participants who would have qualified for this one. It produced hazard ratios against an imputed placebo of 0.72 (95% CI, 0.55 to 0.94) for the three-part composite and 0.61 (95% CI, 0.45 to 0.82) for all-cause death.[5]

Those are the most favorable numbers in the entire SURPASS-CVOT literature, and they belong to a comparison nobody ran. They chain two trials conducted a decade apart in different populations with different background therapy, and their confidence intervals do not carry the uncertainty of the chaining itself. They are worth knowing and they are not evidence of the same kind as a randomized column.

Weight, glucose and who stopped

At 36 months the registry records glycated hemoglobin down 1.66 points (SE 0.0211) on tirzepatide against 0.88 (SE 0.0235), and body weight down 11.6% (SE 0.122) against 4.82% (SE 0.0995).[2] That weight gap is roughly seven percentage points in favor of tirzepatide in a population selected for cardiac disease rather than for obesity, and it is the clearest separation anywhere in the trial — larger, proportionally, than anything in the event columns.

No study-drug discontinuation rate is published. The abstract does not report one, the full text cannot be opened, and the registry posts completion of the study rather than of the drug: 6,524 of 6,648 assigned to tirzepatide and 6,505 of 6,651 assigned to dulaglutide completed, with 45 and 59 lost to follow-up.[2] Those figures describe follow-up discipline and say nothing about how many people stopped injecting. What is published is the tolerability gap behind such a rate: gastrointestinal adverse events affected 2,827 patients (42.5%) on tirzepatide against 2,387 (35.9%) on dulaglutide, with other adverse events similar.[4] Even the comparator arm ran high, as it did against placebo in the earlier trial, at 47.4% against 34.1%.[6]

What SURPASS-CVOT did not establish

It did not establish superiority over dulaglutide on cardiovascular events. It did not compare tirzepatide with semaglutide, which is a separate trial with a separate comparator problem, covered in the SURPASS-2 article. It did not enroll anyone without diabetes, so it says nothing about cardiovascular risk in people using these drugs purely for weight — the trial that asked that question is discussed in the SELECT article, and it used a different molecule. It did not test dulaglutide at 4.5 mg. And it did not measure anything at the doses and durations a cash-pay buyer typically starts on, which are set out in the tirzepatide dose article.

It also did not produce a mortality claim. The 566-against-669 split is a posted count with no ratio, no interval and no test attached to it, and the one place a mortality ratio does appear is the imputed-placebo analysis, where the comparison is constructed rather than randomized. Where the pooled evidence on this class sits is covered in the heart benefits article.

What was in the syringe

Everything above describes branded, FDA-approved tirzepatide supplied free, escalated under protocol to a maximum tolerated dose of up to 15 mg, injected weekly for a median of nearly four years alongside standard diabetes and cardiovascular care in patients selected for established artery disease. Sellers listed on the compounded tirzepatide board dispense compounded preparations, which are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before they reach a patient.

The gap between this trial and that purchase is not mainly about the molecule. It is that the trial ran for four years in people with established heart disease and high glycated hemoglobin, measured hard clinical events, and reported that the drug was not worse than another drug. None of that transfers to a claim about protecting the heart of a person buying a vial for weight loss, and the trial itself declined to make the stronger claim it was designed to test.

Frequently asked

Did SURPASS-CVOT show tirzepatide is better for the heart than dulaglutide?
No. The composite of cardiovascular death, myocardial infarction and stroke occurred in 12.2% on tirzepatide against 13.1% on dulaglutide, a hazard ratio of 0.92 with a 95.3% confidence interval of 0.83 to 1.01. The protocol required an upper limit below 1.00 for superiority and the superiority test returned P = 0.09. What the trial did establish is noninferiority against a margin of 1.05.
Why was the comparator dulaglutide 1.5 mg rather than the highest dose?
Dulaglutide is approved up to 4.5 mg, and a randomized trial in 1,842 patients found 4.5 mg superior to 1.5 mg on both glycated hemoglobin and weight. But 1.5 mg is the only dose with a cardiovascular outcome trial behind it, where it beat placebo at a hazard ratio of 0.88 over a median 5.4 years. Testing against the proven dose made the comparison harder, not easier.
Did anything in SURPASS-CVOT reach significance?
Yes, on the kidney side. A pre-specified exploratory analysis of the composite of persistent macroalbuminuria, a persistent 50% or greater fall in filtration rate, end-stage kidney disease or kidney death found 6.0% against 7.6%, a hazard ratio of 0.77 with a 95% confidence interval of 0.68 to 0.88 and p = 0.0002. The same randomization produced a clear kidney result and an inconclusive cardiovascular one.
Were there fewer deaths on tirzepatide?
The trial's posted results record death from any cause in 566 patients on tirzepatide against 669 on dulaglutide, which is 8.6% against 10.2%. No hazard ratio, confidence interval or p-value is attached to that comparison in the posted results, and the abstract does not report one. A mortality ratio appears only in a separate indirect analysis against a placebo arm that was never randomized.
How many people stopped the injection?
That figure is not published. The abstract does not give a study-drug discontinuation rate and the full text is paywalled with no PubMed Central record, while the registry posts study completion rather than drug discontinuation — 6,524 of 6,648 and 6,505 of 6,651. What is reported is that gastrointestinal adverse events affected 42.5% of the tirzepatide group against 35.9% of the dulaglutide group.
Does SURPASS-CVOT support a heart claim for a compounded tirzepatide vial?
No. The trial used branded, FDA-approved tirzepatide supplied free, escalated under protocol to up to 15 mg, in patients with type 2 diabetes and established artery disease followed a median of four years. Compounded preparations are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before dispensing, and a trial that declined to claim superiority over another drug cannot be borrowed to claim cardiac protection for a different product.

Sources

  1. [1] Nicholls SJ, Pavo I, Bhatt DL, et al. (2025). Cardiovascular Outcomes with Tirzepatide versus Dulaglutide in Type 2 Diabetes. N Engl J Med. PMID 41406444
  2. [2] Eli Lilly and Company (2026). A Study of Tirzepatide (LY3298176) Compared With Dulaglutide on Major Cardiovascular Events in Participants With Type 2 Diabetes (SURPASS-CVOT): posted study results, NCT04255433. ClinicalTrials.gov. Source
  3. [3] Zoungas S, D'Alessio D, Pavo I, et al. (2026). A comparison of the effects of tirzepatide and dulaglutide on major kidney events in people with type 2 diabetes: pre-specified exploratory analyses of the SURPASS-CVOT trial. Lancet Diabetes Endocrinol. PMID 42114520
  4. [4] Nissen SE, Wolski K, D'Alessio D, et al. (2026). Cardiorenal Outcomes With Tirzepatide Compared With Dulaglutide in Patients With Diabetes and Cardiovascular Disease: A Post Hoc Analysis of the SURPASS-CVOT Randomized Clinical Trial. JAMA Cardiol. PMID 41903177
  5. [5] Sattar N, Gerstein HC, D'Alessio D, et al. (2026). Estimating the True MACE Benefits From Tirzepatide in SURPASS-CVOT Using an Imputed Placebo Analysis of REWIND. Diabetes Care. PMID 41940793
  6. [6] Gerstein HC, Colhoun HM, Dagenais GR, et al. (2019). Dulaglutide and cardiovascular outcomes in type 2 diabetes (REWIND): a double-blind, randomised placebo-controlled trial. Lancet. PMID 31189511
  7. [7] Frias JP, Bonora E, Nevarez Ruiz L, et al. (2021). Efficacy and Safety of Dulaglutide 3.0 mg and 4.5 mg Versus Dulaglutide 1.5 mg in Metformin-Treated Patients With Type 2 Diabetes in a Randomized Controlled Trial (AWARD-11). Diabetes Care. PMID 33397768

Where to get it

Best GLP-1 injections

Every injectable seller we can verify, with the price each one publishes and an honest read of what the trials measured.

Compare providers →

More in Evidence