GLP-1s and Daytime Sleepiness

A real-world cohort found GLP-1 users more likely to be diagnosed with hypersomnolence — excessive daytime sleepiness — with a hazard ratio of about 1.32. Before that becomes alarming, the absolute figures: roughly a 0.4% increase at one year and 0.7% at five.^[1]

What was and was not found

The signal was specific rather than general, which is the most informative part of the study.

One sleep outcome moved and four did not. That is a narrower and more credible finding than “these drugs affect sleep” — a broad disruption of sleep architecture would be expected to show up in more than one place.

Why the absolute numbers matter here more than usual

A 32% relative increase on a rare diagnosis is four extra people per thousand.

Daytime sleepiness is also one of the most confounded symptoms it is possible to study in this population. Obesity causes it. Sleep apnea causes it, and is very common in the same people — our sleep apnea article covers a trial in exactly that group. Depression causes it. Eating far less causes it. Any of those could produce a diagnosis code without the drug doing anything.

There is also a detection effect worth naming: people starting a new medication see clinicians more often, and symptoms that were always present get recorded for the first time.

The iron finding is the practical one

Iron deficiency rose at five years, at a hazard ratio of 1.11 with a tight confidence interval.^[1] That is a small effect and a mechanistically obvious one: eating substantially less, for years, reduces intake of everything, and iron is among the first things to run short.

It is also directly actionable, which the sleepiness finding is not. Iron studies are one of the six tests in the monitoring panel we covered in protein targets and the lab panel, and this is the sort of evidence that panel exists for. Iron deficiency also causes fatigue — so some of the sleepiness signal may simply be the iron finding wearing a different diagnosis code.