What Do GLP-1s Do to Wearable Data? Heart Rate Rises, HRV Falls, and Some of the Changes Are Not What They Seem

In the clinical trials behind Wegovy, 26% of adults had a resting heart rate at least 20 bpm higher than when they started, at some visit, against 16% on placebo [1]. To a wearable, a sustained rise like that looks like stress, illness or poor recovery. To a readiness score, it is a reason to tell someone to take it easy.

GLP-1 drugs change almost every number a wearable reports. Some changes are real and benign, some are real improvements, and some are artifacts of how the numbers are calculated. For the growing share of health app users on semaglutide or tirzepatide, a product that cannot tell them apart will give confident, wrong advice. This is what the evidence says about each metric, and what to do about it.

Do GLP-1s raise resting heart rate?

Yes, by 1 to 4 bpm on average, and by much more in some people. The evidence is consistent across drugs and study designs:

This is a known, labelled effect of the drugs. Wegovy’s label asks clinicians to monitor heart rate and patients to report palpitations or a racing heartbeat at rest, and to stop the drug if a sustained increase persists [1]. For a product, that means two things at once: a modest rise is expected and should not be read as strain, and a large sustained rise is exactly what a user should be told to raise with their clinician.

Do GLP-1s lower HRV?

Yes, in the best evidence available. In a randomised, double-blind, placebo-controlled crossover trial, 12 weeks of liraglutide lowered SDNN by 33.9 ms and RMSSD by 0.3 log-ms against placebo, and the fall in SDNN remained significant after adjusting for metabolic and heart rate changes [3]. In the WHOOP study, people starting a GLP-1 saw HRV fall by about 6 ms over 12 weeks compared with matched controls [4].

Lower HRV usually means more physiological stress or less recovery. On a GLP-1, it partly reflects the drug itself, alongside better weight, blood sugar and blood pressure. Reading it the usual way gets the story backwards.

Why do readiness and recovery scores drop on a GLP-1?

Because they are built for the opposite situation. Most readiness and recovery scores compare overnight resting heart rate and HRV with the user’s own recent baseline, and treat higher heart rate and lower HRV as signs of strain. A GLP-1 moves both in that direction for reasons unrelated to training load or illness.

The size matters. Across 92,457 Fitbit users, resting heart rate varied by about 3 bpm a day within the same person [5]. A rise of about 3 bpm is roughly one standard deviation, and in our simulation of personal baselines, a lasting shift of that size kept being flagged as unusual for about 5 days with a 7-day baseline and up to 41 days with a 60-day one, as we covered in how many days a wearable baseline needs. GLP-1 doses are usually increased in steps over several months, so a user can face a new shift every time the dose goes up.

The WHOOP study found one moderating factor: people who exercised more saw smaller increases in resting heart rate [4]. It is a correlation, not proof, but it is the kind of finding a coaching product can use.

Does VO2 max improve when you lose weight on a GLP-1?

The estimate may rise. Fitness often does not. Wearables report VO2 max per kilogram of body weight, and Apple says its estimate takes weight, height, age, sex and heart-rate-affecting medications into account [6]. Lose 15% of body weight with no change in the heart’s capacity, and a per-kilogram figure rises. A GLP-1 also raises heart rate, which pushes heart-rate-based estimates the other way. The trend is hard to read in either direction.

The physiology is clearer. A 2025 review of a decade of studies found no evidence that GLP-1 drugs improve cardiorespiratory fitness, and noted that 25% to 40% of the weight lost is fat-free mass, including muscle [7]. A rising VO2 max estimate on a GLP-1 is more likely to be the denominator moving than the engine improving. Fitness, and the muscle behind it, has to be trained for separately, which is why a coaching product should track performance, such as pace or power at a given heart rate, rather than the estimate alone.

Are calorie estimates still accurate on a GLP-1?

Only if the weight in the user’s profile is current. Apple tells users to keep their personal information up to date because its calorie estimates depend on it [8], and most wearables work the same way. Energy spent walking and moving scales with body weight, so a profile still showing the starting weight after a 15% loss overstates movement energy by roughly that much. For a weight-loss product built on calorie balance, that is an error in the direction of false reassurance.

The same applies to body composition. Smart scales cannot track lean mass changes of a few hundred grams a month, which is the scale of muscle loss on these drugs, as we covered in smart scale accuracy on a GLP-1.

What happens to sleep apnea and blood oxygen readings?

They should improve, and this change is real. In the two SURMOUNT-OSA trials, tirzepatide reduced the apnea-hypopnea index by 25.3 and 29.3 events an hour over a year, against 5.3 and 5.5 on placebo, and 42% of participants not using CPAP reached remission, against 16% on placebo [9]. The FDA approved tirzepatide for moderate to severe obstructive sleep apnea in adults with obesity in December 2024.

Wearables that estimate breathing disturbances or overnight blood oxygen dips should register that improvement. A product can show it as progress, while making clear that a diagnosis and its follow-up belong with a clinician, which we covered in what smartwatch sleep apnea detection can and cannot do.

What should a product do when a user starts a GLP-1?

All of this depends on knowing that a user has started treatment, which is sensitive health information. Ask for it explicitly, explain what it changes, and treat it with the care that state health privacy laws and the FTC expect, which we covered in HIPAA-compliant health data APIs.

GLP-1s are changing the health of millions of people, and the wearable data those people generate changes with them. Most of the shifts are predictable: heart rate up a little, HRV down a little, apnea down a lot, weight and lean mass down together. The products that serve these users well will be the ones that know which of those changes to celebrate, which to explain and which to stop reading as a warning.