Five millimeters of mercury is a meaningful quantity. Individual participant data from 48 randomized trials covering 344,716 people put the hazard ratio for a major cardiovascular event per 5 mmHg of systolic reduction at 0.91 (95% CI, 0.89 to 0.94) in people with no prior cardiovascular disease and 0.89 (95% CI, 0.86 to 0.92) in people with it — about a tenth of the risk, at any starting pressure, including normal ones.[1] That is the yardstick every number below has to be held against, and it comes from blood-pressure drugs rather than from these ones.
The size of the effect, and which instrument recorded it
On the tirzepatide weight label, systolic pressure at week 72 fell by 5.6, 6.7 and 6.4 mmHg below placebo at 5, 10 and 15 mg, with diastolic down 4.1, 4.2 and 3.7 mmHg, every interval clear of zero.[2] On the semaglutide label, the cardiovascular outcome trial recorded a smaller systolic difference at week 104 — 3.3 mmHg — and a diastolic difference of 0.5 mmHg.[3]
Every one of those figures carries the same footnote: the parameter was not included in the trial’s prespecified hierarchical testing.[3] Blood pressure was a secondary reading taken while the trial measured something else, and no product in this class carries an indication for treating hypertension. What these drugs are approved to do, and on what evidence, is in the indication article.
The two halves of a blood-pressure reading also behave differently, and the newest semaglutide dose makes that visible. At 7.2 mg in adults with obesity, systolic pressure fell 5.3 mmHg below placebo and diastolic 2.7 mmHg. In adults with obesity and type 2 diabetes at the same dose, systolic fell 4.0 mmHg and diastolic 0.2 mmHg — two-tenths of a millimeter of mercury, which is nothing.[3] A pressure effect that is almost entirely systolic widens pulse pressure rather than narrowing it, and the trials were not designed to say whether that matters.
Twenty-four-hour monitoring makes the effect larger and stranger
A prospectively planned substudy of SURMOUNT-1 fitted 600 participants with ambulatory monitors — readings every 30 minutes by day and every 60 by night — and analyzed the 494 with valid data at baseline and week 36. Mean baseline 24-hour systolic pressure was 124.6 mmHg, and 30.0% reported hypertension with 29.0% on at least one antihypertensive.[4]
The placebo-adjusted 24-hour systolic change was −7.4 mmHg (95% CI, −10.0 to −4.7) at 5 mg, −10.6 (95% CI, −13.2 to −8.0) at 10 mg and −8.0 (95% CI, −10.6 to −5.4) at 15 mg.[4] The middle dose produced the largest reduction. Nighttime pressure fell alongside daytime pressure, which matters because nighttime systolic pressure predicts cardiovascular and all-cause death better than the daytime or 24-hour average does.
Diastolic pressure went somewhere else entirely. It fell 2.0 mmHg (95% CI, −3.6 to −0.3) at 5 mg and 2.9 mmHg (95% CI, −4.5 to −1.3) at 10 mg, and at 15 mg it fell 0.5 mmHg on an interval of −2.0 to 1.1 — indistinguishable from nothing.[4] The same top dose produced the largest heart-rate increase in the substudy, 5.4 beats per minute against 2.1 and 2.3 at the lower doses, which is the subject of the heart-rate article. Whatever is producing the pressure fall, it is not a quantity that scales with dose, and the highest dose bought the worst combination of the two.
Weight explains most of it, and the interval reaches past all of it
In the same substudy, change in 24-hour systolic pressure correlated with change in body weight at r = 0.31 (P < 0.0001), and a mediation analysis put the share of the pressure change carried by weight change at 70.0% (95% CI, 47.0 to 102.6).[4] An upper bound above 100% means the data cannot exclude weight loss accounting for the entire effect, and the lower bound means it cannot exclude nearly a third of the effect arriving some other way. The authors also note that food intake and 24-hour urine sodium were never measured, so the contribution of eating less salt cannot be separated out at all.
The trial number is net of prescriptions being taken away
An individual-patient-data meta-analysis pooled three 68-week randomized trials of semaglutide 2.4 mg — 2,109 treated and 1,027 on placebo — and found the between-arm systolic difference at −4.95 mmHg (95% CI, −5.86 to −4.05).[5]
Alongside that, the analysis tracked what happened to everyone’s other medication. The antihypertensive treatment intensity score fell in the semaglutide arm relative to placebo by 0.51 (95% CI, 0.32 to 0.71 lower).[5] Participants were having blood-pressure drugs reduced or stopped as their weight came down, and the 4.95 mmHg that remains is what survived that unwinding. The true pharmacological effect is larger than the published difference, and the published difference is only reproducible in a setting where somebody is watching the rest of the prescription list.
That is the practical consequence worth stating plainly. A person who takes two antihypertensives, buys a vial through a checkout form, and loses fifteen percent of their body weight has acquired an unsupervised pressure reduction on top of an unchanged regimen. Nobody deprescribed anything. The symptoms that can follow are covered in the dizziness article.
Stop the drug and the pressure comes back
Both programs ran a randomized withdrawal, and both published what happened to blood pressure. In the tirzepatide withdrawal study, participants took open-label drug to week 36 — systolic pressure fell 11.2 mmHg over that run-in — and were then randomized to continue or switch to placebo. Over the following 52 weeks systolic pressure rose 8.2 mmHg in the withdrawn group against a rise of 2.0 in those who continued, a difference of −6.2 mmHg (95% CI, −8.2 to −4.3); diastolic rose 3.2 against a fall of 0.7, a difference of −3.8 (95% CI, −5.2 to −2.4).[2]
The semaglutide withdrawal ran the same way from week 20 and found systolic pressure rising 4.4 mmHg in the withdrawn group against 0.5 in those continuing, a difference of −3.9 mmHg.[3] Nearly the whole effect is contingent on the next injection, which puts blood pressure in the same category as the weight it follows — the stopping article sets out what else returns, and the tapering article what is and is not known about coming off slowly.
Who this actually matters for
A meta-analysis of seven randomized trials put the tirzepatide systolic reduction at a median of −4.20 mmHg (95% CI, −5.17 to −3.23) at 5 mg, −5.34 (95% CI, −6.31 to −4.37) at 10 mg and −5.77 (95% CI, −6.73 to −4.81) at 15 mg.[6] Set against the 5 mmHg yardstick, that is roughly a tenth off major cardiovascular risk for someone whose risk is high enough for a tenth to be worth something.[1] For someone at 118 over 74 it is a laboratory result.
The ambulatory substudy also defines its own population by exclusion: entry required blood pressure below 140/90 and, if antihypertensives were used, a stable dose for at least three months.[4] The most rigorous blood-pressure measurement in this literature was therefore taken in people whose pressure was already controlled. The one trial population where a pressure question is most urgent — uncontrolled hypertension — is the one the substudy did not enroll.
For anyone weighing whether lowering pressure is worth the lightheadedness that sometimes comes with it, the closest evidence comes from outside this drug class. An individual participant meta-analysis of nine hypertension trials in 29,235 people, 9% of whom had orthostatic hypotension at baseline, found more intensive treatment reducing cardiovascular disease or death whether or not orthostatic hypotension was present — hazard ratio 0.81 (95% CI, 0.76 to 0.86) without it and 0.83 (95% CI, 0.70 to 1.00) with it, P = 0.68 for interaction.[7] Standing dizziness was not a reason the benefit disappeared. In the smaller group with standing hypotension at baseline the estimate was 0.94 (95% CI, 0.75 to 1.18) and not significant. Those were people taking antihypertensive drugs, not incretins, and the transfer is an inference.
What is left after the caveats
The effect is real, it is roughly the size of adding a low-dose antihypertensive, it is largest when measured over 24 hours rather than in a clinic, and it is almost entirely conditional on continuing treatment. It is also not what the drug is sold for, not tested hierarchically, not larger at the top dose, and not accompanied by any instruction to review the medicines it interacts with.
Most sellers covered here dispense compounded semaglutide or tirzepatide, which is not FDA-approved and is not reviewed by the FDA for safety, efficacy or quality before it is dispensed. No ambulatory monitoring substudy has been run on a compounded preparation, and the cardiovascular outcome evidence that gives a pressure change its meaning was generated on branded product — the outcome trial article sets out what that trial did and did not establish.