Yes, in one specific population. Pooled data from two placebo-controlled trials found semaglutide improved heart-failure symptom scores by 7.5 points more than placebo. That was over 52 weeks, in 1,145 people with obesity-related heart failure with preserved ejection fraction [1]. A separate trial found tirzepatide cut a composite of cardiovascular death or worsening heart failure by 38%, from 15.3% to 9.9%. That was in a comparable population, followed for a median of 104 weeks [2] — the full component breakdown is below. Neither number transfers whole: one hides a death count that ran the wrong way, and both apply to a narrower group of patients than most readers asking this question.
Heart failure with preserved ejection fraction is a condition with very little that works. The pooled analysis of STEP-HFpEF and STEP-HFpEF DM is one of the few trials to move it. It combined individual patient data from two randomized placebo-controlled trials across 129 sites in 18 countries, every participant with a BMI of at least 30 and heart failure symptoms. The effect sizes are modest in absolute terms and the direction is consistent across everything measured — including one number that looks wrong until you understand it. Before weighing it against what a seller charges, it is worth knowing who was studied.
What improved in STEP-HFpEF
The dual primary endpoints were symptom score and body weight at 52 weeks. The Kansas City Cardiomyopathy Questionnaire clinical summary score improved by 7.5 points more on semaglutide than placebo (95% CI 5.3 to 9.8). Body weight fell 8.4 percentage points further (95% CI −9.2 to −7.5). Both at P < 0.0001.
Six-minute walk distance improved by 17.1 meters more (95% CI 9.2 to 25.0). A hierarchical composite of death, heart failure events and changes in those two measures gave a win ratio of 1.65 (95% CI 1.42 to 1.91). Effects held consistently across subgroups defined by age, race, sex, BMI, blood pressure, baseline CRP and ejection fraction.
The number that looks like a typo
There were 161 serious adverse events in the semaglutide group and 301 in the placebo group. That is not a transcription error, and it is not unusual in a sick population. A drug that prevents some of what the underlying illness would otherwise do can leave the treated arm with fewer serious events than the untreated one. It is also a reminder that the harms of a condition and the harms of its treatment are measured on the same page.
A second trial, and a composite worth taking apart
SUMMIT enrolled 731 patients with the same diagnosis and randomized them to tirzepatide or placebo for a median of 104 weeks. The composite of cardiovascular death or worsening heart failure fell from 15.3% on placebo to 9.9% on tirzepatide, a hazard ratio of 0.62. Almost all of that came from the worsening-heart-failure half: 14.2% against 8.0%, on its own.
Who inside these trials it worked for
Re-cutting the 1,145 STEP-HFpEF participants by frailty separated two things that are usually assumed to move together. Weight loss was statistically the same across every frailty stratum (P for interaction 0.38). The symptom benefit was not (P for interaction below 0.001). The most frail gained 11.0 points against placebo. The nonfrail group gained nothing measurable, a mean difference of −1.5 with an interval from −8.4 to 5.4. The full breakdown, including why that negative point estimate is printed rather than dropped, is in the frailty analysis[3].
A separate, prespecified look at the SUMMIT population by sex found the opposite pattern: no difference at all in the headline benefit. Cardiovascular death or worsening heart failure gave hazard ratios of 0.66 in women and 0.61 in men, an interaction P of 0.81. The one difference that reached significance concerned mechanism, not size — among people on tirzepatide, weight loss tracked symptom improvement more closely in women than in men. That comparison is set out in the sex-stratified SUMMIT analysis[4].
How this compares with semaglutide
No trial has randomized semaglutide against tirzepatide on cardiovascular outcomes, so the comparison runs on claims data. A 217,920-person target trial emulation found tirzepatide ahead on heart failure at one year. The two ways of reading that result disagree about who benefits more. The risk ratio was 0.60 in people without diabetes against 0.82 with it. The absolute risk difference ran the other way: 0.9 percentage points without diabetes against 1.5 points with it. Both numbers are correct, and which one matters depends on baseline risk — the full working is in the semaglutide-versus-tirzepatide comparison[5].
The sickest patients trials do not enroll
None of the trials above included people with a heart failing badly enough to need a mechanical pump. A matched cohort of 141 pairs with implanted left ventricular assist devices found favorable directions for infection, mortality and transplant listing among those with prior GLP-1 exposure. The abstract publishes no hazard ratios, confidence intervals or percentages for any of it. Exposure was also recorded before implantation, which selects for people already stable enough to be on the drug. That gap between direction and magnitude is detailed in the assist-device cohort[6].
What it does not transfer to
These were people with diagnosed heart failure, a BMI of 30 or more, and symptoms bad enough to score under 90 on the questionnaire. The dose was 2.4 mg weekly, which is the weight-management dose most compounded sellers titrate toward — so unlike the kidney trial at 1.0 mg, the dose here does transfer. The population does not.
If you have heart failure and are considering a telehealth purchase, the gap is not the evidence. It is that this is a condition needing monitoring. Most sellers tracked here publish nothing about who reviews a case — the census is in what sellers publish and in what they will not tell you. A seller that cannot tell you who reads your history is not the right route into a cardiac diagnosis.