A higher resting heart rate predicts death. Across 46 prospective cohorts covering 1,246,203 people and 78,349 deaths, each 10 beats-per-minute increment carried a relative risk of 1.09 (95% CI, 1.07 to 1.12) for all-cause mortality and 1.08 (95% CI, 1.06 to 1.10) for cardiovascular mortality, essentially unchanged after adjustment for the usual risk factors.[1] These drugs raise resting heart rate. They also reduce death from any cause at an incidence rate ratio of 0.88 (95% CI, 0.84 to 0.92) across 21 randomized trials in 99,599 patients.[2] Both statements are true, and holding them at once is the whole of what is known about this effect.
How many beats, and on which molecule
The labeled means are small and they cluster. Injected semaglutide for weight reduction: 1 to 4 bpm above placebo.[3] The same molecule for diabetes: 2 to 3 bpm, against a mean decrease of 0.3 bpm on placebo.[4] Oral semaglutide: 1 to 3 bpm, against no change on placebo.[5] Tirzepatide for weight: 1 to 3 bpm against no increase on placebo; for diabetes, 2 to 4 bpm against 1 bpm.[6][7] Liraglutide: 2 to 3 bpm.[8]
Nothing in that range separates the molecules. A three-beat difference between a treated arm and a placebo arm is smaller than the difference between sitting down and standing up, which is why the tirzepatide diabetes label states outright that the clinical relevance of the heart rate increases is uncertain.[7] It is also why the effect is almost never the reason someone stops, and why the molecule comparison does not turn on it.
The instrument decides the number
The liraglutide label reports the same effect twice, measured two ways, and the results differ by a factor of three. Routine clinical monitoring across the trial program produced the mean of 2 to 3 bpm. A clinical pharmacology trial that monitored heart rate continuously for 24 hours found treatment associated with a rate 4 to 9 bpm higher than placebo.[8]
A clinic reading catches a patient sitting still by appointment. A 24-hour recorder catches them overnight, after meals, and during the hours nobody schedules a vital sign. The larger figure is the more complete measurement, and it is the one that almost never travels. A wearable study points the same way from outside a trial: 66 adults tracked from the week before starting through 12 weeks on drug showed a resting heart rate change of 3.2 ± 0.8 beats/min against a propensity-matched control group, alongside 10.0% weight loss.[9]
The mean hides a tail
Averages are the wrong summary for a distribution with a long edge, and the semaglutide label prints the edge. More treated adults than placebo adults recorded a maximum change from baseline at any visit of 10 to 19 bpm (41% against 34%) and of 20 bpm or more (26% against 16%). In the adolescent trial, among patients with a normal baseline heart rate, a maximum change of 20 bpm or more was recorded in 54% against 39%.[3]
Read those placebo columns first. Sixteen percent of adults taking nothing recorded a 20-beat swing at some visit, and 34% recorded a 10-to-19-beat one, because heart rate moves for coffee, stairs, sleep debt and a waiting room. The drug adds about ten percentage points on top of a background that is already large. That is the correct frame for a single alarming reading on a home monitor, and the caffeine half of it has its own page in the caffeine article.
Liraglutide’s label records the same shape with a stricter definition — a change at two consecutive visits — and the numbers shrink accordingly: more than 10 bpm in 34% against 19%, and 20 bpm in 5% against 2%. At least one resting rate above 100 bpm was recorded in 6% against 4%, and at two consecutive visits in 0.9% against 0.3%. Tachycardia was reported as an adverse reaction in 0.6% against 0.1%.[8] Persistent is roughly a tenth as common as transient.
The one dose-dependent arrhythmia figure in the class
The tirzepatide diabetes label carries the only rate that climbs with dose. Episodes of sinus tachycardia accompanied by a rise from baseline of at least 15 bpm were reported in 4.3%, 4.6%, 5.9% and 10% of patients on placebo, 5 mg, 10 mg and 15 mg.[7] Among patients enrolled in Japan the same figures were 7%, 7.1%, 9.3% and 23% — three in thirteen at the top dose, against seven in a hundred on placebo.[7]
Those Japanese denominators are 43 patients per arm, so the 23% rests on ten people and should not be read as a population estimate. What it does establish is that the effect is not uniform across bodies, and that the largest dose produces it most — which makes it one of the few reasons to raise a heart-rate question before a scheduled step up rather than after. The schedule itself is in the titration article.
Two labels warn about it. Four do not.
This is the part that reveals how the regulators weigh it. Of the six labels in this class, exactly two carry a Warnings and Precautions section headed Heart Rate Increase: the semaglutide weight product at 5.9 and the liraglutide weight product at 5.5. Both instruct monitoring heart rate at regular intervals, both tell patients to report palpitations or a racing heartbeat at rest, and both state that a sustained increase in resting heart rate is a reason to discontinue the drug.[3][8]
The semaglutide diabetes label, which is the identical molecule, has no such section — heart rate appears only among adverse reactions, with no monitoring instruction and no discontinuation threshold.[4][5] Neither tirzepatide label has one either, despite carrying the sinus tachycardia figure above.[6] [7] Two products, one molecule, two different sets of instructions to a prescriber; and the product with the largest dose-dependent tachycardia signal is among the four with no warning at all. Anyone treating a label warning as a ranking of danger is reading a regulatory history rather than a physiology.
What is producing it, and what is not
The obvious explanation is sympathetic activation, and it has been tested directly. A randomized, double-blind trial in 57 adults with type 2 diabetes ran two phases: an acute intravenous exenatide infusion and 12 weeks of liraglutide. The infusion raised resting heart rate by 7.5 ± 0.9 bpm (P < 0.001). Twelve weeks of liraglutide raised it by 6.6 ± 2.1 bpm while systolic blood pressure fell by 12.6 ± 4.7 mmHg (both P < 0.01). Heart rate variability, the marker of sympathetic activity, did not change, and neither did cardiac output or arterial stiffness.[10]
The authors conclude that the acceleration is not explained by sympathetic activation and that their data argue against vasodilation as the cause, pointing instead at direct stimulation of the sino-atrial node.[10] That matters practically: a faster pulse here is not the body compensating for something going wrong, and it is not the signature of the adrenaline response people assume when they feel it. The pressure falling at the same time is the effect covered in the dizziness article.
The wearable study reaches a different mechanism with a different instrument: there, the heart rate rise was statistically mediated by a fall in heart rate variability of 6.2 ± 1.4 ms, and higher weekly physical activity was associated with an attenuated rise.[9] Two plausible pathways, one small controlled trial and one observational cohort, and no reconciliation yet. The exercise finding is a trend rather than a result, and belongs alongside what training does on this class rather than as advice.
Palpitations, and what the reports can say
Palpitations are what people actually notice, and the postmarketing record is thinner than the volume of discussion implies. An analysis of 103,693 tirzepatide reports found no disproportionate reporting for tachycardia among healthcare-professional reports (reporting odds ratio 1.01; 95% CI, 0.68 to 1.51) or for atrial fibrillation (1.02; 95% CI, 0.58 to 1.80). In the full dataset restricted to reports with a documented body weight, tachycardia did reach significance in both weight strata — 1.91 and 1.57 — but that finding was not replicated once the analysis was limited to clinicians.[11]
The reason the two halves disagree is printed in the same paper: 92.3% of those reports were submitted by consumers rather than clinicians.[11] A symptom that a patient can feel and a clinician cannot see is exactly the kind that separates when a database is split by who filed it. A comparison of 56,799 semaglutide against 135,992 tirzepatide reports found tirzepatide generated higher raw counts for palpitations and tachycardia while semaglutide carried the disproportionality after volume adjustment, and its authors attribute that to differences in who takes each drug rather than to toxicity.[12] Neither analysis can produce an incidence, and neither should be read as one.
How it sits beside the outcome data
The trials that measured death, rather than heart rate, measured it in the same people. Across 21 randomized trials of eight agents in 99,599 patients over a mean 2.4 years, the class reduced all-cause death at an incidence rate ratio of 0.88 (95% CI, 0.84 to 0.92; number needed to treat 121), cardiovascular death at 0.87 (95% CI, 0.81 to 0.92; NNT 170) and major adverse cardiovascular events at 0.87 (95% CI, 0.83 to 0.91; NNT 66), all graded high-certainty. Serious adverse events fell 9%, myocardial infarction 15% and heart failure 15%, while gastrointestinal disorders rose 63% and gallbladder disorders 26%.[2]
So the resting heart rate goes up by a few beats and the events that a higher resting heart rate predicts go down by an eighth. The resolution is that resting heart rate is a marker of risk in populations, not a mechanism of it in an individual, and a drug can move the marker without moving what the marker tracks. The populations those trials enrolled were selected for disease, which is the standing caveat on the cardiovascular article and the reason a mortality figure earned in secondary prevention does not transfer to a healthy 35-year-old.
Where the line sits
Every figure above was measured on FDA-approved product at labeled doses with a clinician taking the pulse. 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, and no trial has measured a heart rate on one. More to the point, the monitoring the two warning labels instruct — a pulse checked at regular intervals — is a clinical act, and whether a cash plan includes one is recorded in each provider review.
The threshold the labels themselves set is a sustained increase in resting heart rate, not a single reading. A rate that is consistently 15 or 20 beats above a personal baseline across several days at rest, a resting pulse persistently over 100, or palpitations that last minutes at a time are the presentations both warning labels tell patients to report. Chest pain, fainting, or a fast heartbeat with breathlessness is a different matter and belongs in urgent care the same day.