No trial of this drug class has measured fatigue as an endpoint, so the honest answer is assembled from adjacent measurements rather than read off a result. The closest direct one is daytime sleepiness, pooled across fifteen studies and 1,877 participants with sleep apnea, where the Epworth score improved by 0.20 points with a 95% confidence interval from 0.14 to 0.26 [1]. Two points is the smallest change on that scale anybody notices, so the measured effect is a tenth of the threshold of perception, and it runs toward less sleepiness rather than more. What the same literature does show is depletion: ferritin 26% to 30% below a comparison group on a different diabetes drug, and more than 60% of users eating below the estimated requirement for iron [2].
Tiredness is one of the commonest things people report on these drugs and one of the least studied, which is an uncomfortable combination. The trials count nausea, vomiting, diarrhea and constipation, and the time course of those is set out in the nausea timeline. Energy was not on the list.
The one scale that touches it
The Epworth Sleepiness Scale asks how likely someone is to doze off in eight ordinary situations, and it is the only instrument in this literature that measures anything close to how tired a person feels. A network meta-analysis of fifteen studies and 1,877 participants with moderate to severe obstructive sleep apnea pooled it, and GLP-1 drugs improved the score by 0.20 points against placebo [1]. In the same analysis, breathing events fell by 15.28 an hour, 95% CI 8.35 to 22.22, which is a large effect on the thing a machine counts.
The subgroup analyses behind the sleep apnea approval point the same way. Reductions in the apnea-hypopnea index appeared in every subgroup examined — by age from 27.7 to 34.1 events per hour, by body mass index from 25.2 to 34.4, by neck circumference from 23.9 to 30.8 [3]. Those analyses are descriptive, with no interaction tests and no subgroup confidence intervals reported, and the full reading is in the subgroup post hoc. For somebody whose tiredness comes from untreated apnea, this is the mechanism by which a GLP-1 would make them less tired rather than more.
The mechanisms that would push the other way
Three findings in this literature are consistent with feeling worse rather than better, and none of them was collected as a fatigue measurement. The first is iron. A narrative review covering six studies and 480,825 adults found ferritin running 26% to 30% below a comparison group taking an SGLT2 inhibitor, and more than 60% of users eating below the estimated requirement for iron and for calcium [2]. Low iron stores are among the most ordinary causes of tiredness there is, and the full nutritional picture is in the micronutrient evidence.
The second is thiamine, at the severe end of the same mechanism, where prolonged vomiting and rapid weight loss have preceded every published case of Wernicke’s encephalopathy in this setting. Fatigue and confusion are early features, and the staging of it is in the thiamine cases. Six published cases cannot produce a rate, and nothing here says this is common.
The third is energy expenditure itself. In 83 adults followed for twelve months at one obesity unit, skeletal muscle mass fell from 29.9 kg to 28.7 kg and fat-free mass from 53.8 kg to 52.7 kg, both at p < 0.001 [4]. Handgrip strength was preserved, in the treated group and in the lifestyle-only group alike. Resting energy expenditure fell further than the change in body composition predicts, and it did so equally in both groups, which makes it a feature of losing weight rather than of the drug.
What to do with an answer this thin
Someone who feels tired on one of these drugs is not contradicted by the evidence, because the evidence never asked. The measurable causes worth ruling out are the ones a blood test settles: iron and ferritin, vitamin B12, vitamin D, and thyroid function. A reader losing lean mass quickly has a second, separate question to ask, and it is laid out in the muscle-loss evidence.
The one thing this page will not do is convert an absence of data into a reassurance. A drug that reduces food intake by roughly 271 calories at a measured meal is reducing the intake of everything, as the intake measurements show. The trials that would have detected a fatigue signal were not designed to collect one.