One Fewer Event for Every Seventy Patients
A large analysis of United States insurance claims has linked tirzepatide to a lower rate of heart attacks and deaths among people with type 2 diabetes who already have heart disease, and the size of that effect is worth understanding precisely rather than in headline terms.
Researchers led by Nils Krüger at Brigham and Women's Hospital and Harvard Medical School examined 52,971 adults aged 40 and older with type 2 diabetes, a body mass index of at least 25, and established atherosclerotic cardiovascular disease. About 35,353 started tirzepatide, sold as Mounjaro and Zepbound, and 17,618 started sitagliptin, an older diabetes drug chosen because dedicated trials have shown it has a neutral effect on cardiovascular outcomes. Sitagliptin, therefore, serves as a placebo. The findings were published in The BMJ in August 2026.
At one year, the estimated risk of a major adverse cardiovascular event, defined here as heart attack, stroke or death from any cause, was 2.9 percent with tirzepatide and 4.4 percent with sitagliptin. That is a 32 percent relative reduction and, in absolute terms, about one fewer event for every 70 people who start the drug.
The difference was driven by heart attacks and all-cause mortality. Tirzepatide was associated with a 33 percent lower risk of heart attack. Ischemic stroke risk did not differ meaningfully between the two groups, which is a specific limitation rather than a rounding detail.
A Cohort Study, Not a Randomized Trial
This is observational research built on administrative claims. People were not randomly assigned to a drug. Their doctors chose, and the reasoning behind those choices can differ in ways that also affect heart outcomes.
The researchers took the standard steps to address that. They used propensity score overlap weighting to balance baseline characteristics between the groups, accounting for age, sex, race, body mass index, prior heart problems, other chronic conditions, and medication use. They included negative control outcomes, meaning conditions with no plausible connection to the drug, and found no association with those, which argues against a simple healthier-patient effect. They also benchmarked their approach against results from earlier randomized trials.
Those are good methods. They do not convert a cohort study into a trial. An accompanying BMJ editorial by Sourbha Dani, Ashish Jayakumar, and Sarju Ganatra made the point in its title, describing the finding as "a credible signal, not a causal verdict".
That framing sits alongside the randomized evidence, which is more mixed than the new numbers suggest on their own. In SURPASS-CVOT, published in the New England Journal, 13,299 patients with type 2 diabetes and atherosclerotic disease were randomized to tirzepatide or dulaglutide and followed for a median of 4 years, with 13,165 included in the primary analysis. The primary composite of cardiovascular death, heart attack or stroke occurred in 12.2 percent on tirzepatide and 13.1 percent on dulaglutide. That met the statistical bar for noninferiority but not for superiority, as the American College of Cardiology summary noted.
Reconciling the two is straightforward once the comparators are clear. SURPASS-CVOT asked whether tirzepatide beats another drug already known to reduce cardiovascular events. The new cohort study asked whether it beats a cardiovascular-neutral one. Those are different questions, and the answers do not contradict each other.
An Unexplained Signal in Infection Deaths
One finding in the cohort study has drawn attention beyond the cardiovascular endpoints. The authors also tracked infections requiring hospital admission and infection-related mortality as safety outcomes. Hospital admissions for infection were about 36 percent lower among tirzepatide users, and infection-related death was about 60 percent lower.
Krüger said in a Technical University of Munich statement that "The data on infections surprised us because the findings were so striking." He pointed to evidence that obesity promotes inflammatory processes that may impair immune function, while noting that the mechanisms and whether any benefit would occur without weight loss still need to be investigated.
That is a hypothesis, not a demonstrated pathway, and it is precisely the kind of unexpected result that observational data can generate for reasons unrelated to the drug. Sicker and frailer patients are both more likely to die of infection and less likely to be started on a newer injectable medication. Whether the statistical weighting fully removed that imbalance cannot be settled within this study design.
Coverage in the cardiology press, including an analysis at TCTMD, has treated the cardiovascular results as supportive and the mechanistic claims as open. The authors reported support from the National Institutes of Health and the German Heart Foundation for the work, and senior author Shirley Wang reported grants to her institution from the FDA and NIH, along with consulting income for unrelated work. The editorial authors reported no relevant conflicts of interest.
Questions Worth Raising at the Next Appointment
Nothing here changes prescribing guidance today, and nobody should start, stop, or switch a medication based on it. The population studied is specific: adults over 40 with type 2 diabetes, a body mass index of at least 25, and established atherosclerotic cardiovascular disease. That is not the same as people taking tirzepatide for weight loss without diabetes or heart disease, and the results do not transfer to that group.
For patients who do fit the studied population, the finding supports a conversation rather than a decision. Reasonable questions include whether cardiovascular risk is already part of the rationale for the current diabetes regimen, whether another agent in the same class would serve the same purpose at a lower cost, and what the out-of-pocket cost looks like over a full year rather than a first fill.
Cost remains the practical barrier for many households. Insurance coverage for tirzepatide varies substantially by plan and indication, and coverage for diabetes often differs from coverage for weight management, even when the molecule is identical. Patients facing a denial can ask a prescriber about prior authorization, formulary alternatives within the class, generic or older comparators, and manufacturer assistance programs.
The follow-up in this study ran for one year. Cardiovascular disease unfolds over decades, and a twelve-month window cannot capture what happens at five or ten years. Larger and longer randomized trials directly comparing incretin drugs are underway, and those results will carry more weight than any claims analysis can.
Key Questions Answered
What did the study find? Among 52,971 adults with type 2 diabetes and established atherosclerotic cardiovascular disease, the one-year risk of heart attack, stroke, or death was 2.9 percent with tirzepatide and 4.4 percent with sitagliptin, a 32 percent relative reduction.
Does this prove tirzepatide prevents heart attacks? No. This was an observational cohort study using insurance claims, not a randomized trial. An accompanying BMJ editorial described the result as a credible signal rather than a causal verdict.
Did it reduce strokes? Ischemic stroke risk did not differ meaningfully between the groups. The benefit was driven by heart attacks and all-cause mortality.
Who does this apply to? Adults over 40 with type 2 diabetes, a body mass index of at least 25, and established atherosclerotic cardiovascular disease. It does not describe people taking the drug for weight loss without those conditions.
How does this square with the randomized trial? SURPASS-CVOT compared tirzepatide with dulaglutide, a drug already shown to reduce cardiovascular events, and found it noninferior but not superior. This study compared it against a cardiovascular-neutral drug, a different question.
Should anyone change medication based on this? No. Do not start, stop, or switch a prescribed medication without speaking with a clinician. This is a reason to have a conversation during a scheduled visit.
What remains unknown? Folasted ran f1only one year; the reported reduction in infection-relatedlacks ans has no established mechanism, and residual confounding caruled outexcluded in observational data.
