Cough was reported 1.46 times as often within six months of starting, in a matched cohort of 617,296 adults [1]. Any laryngeal symptom came in at 1.19 (95% CI 1.16–1.21). The absolute differences are small, and the authors say so themselves.
Nausea gets a warning on every intake form. A hoarse voice and a persistent cough do not. Adults with obesity on these drugs were matched to controls on age, sex, race and ethnicity. Head and neck cancer, prior radiation, autoimmune disease and airway disorders were excluded first. Keeping the molecules straight matters here for the same reason it does in the super-responder brand comparison.
Which drugs, and how much
Cough came in highest at 1.46, and voice and resonance disorders at 1.19. Exenatide, liraglutide, semaglutide, dulaglutide and lixisenatide each showed a significant six-month elevation. Tirzepatide and albiglutide did not.
Two nulls complicate a class-wide story. A drug with fewer users has less power, and albiglutide has been off the market for years. Tirzepatide is different: it is widely used, so its null result carries more.
The likely mechanism was not measured
Reflux is the obvious candidate. It classically presents as throat clearing, a lump-in-the-throat sensation and a cough rather than heartburn. This study tested none of that — no pH testing, no endoscopy, no reflux diagnosis in the analysis.
The reflux evidence sits elsewhere and it is randomized. Pooling 55 trials and 106,395 participants, reflux carried a risk ratio of 2.19 (95% CI 1.48–3.25) [2]. The authors translate that as about 4 extra cases per 1,000 people treated. Gallstones came in at 1.46, or about 2 per 1,000.
The rest of that list came back null: pancreatitis, obstruction, ileus, perforation, bleeding, gastroparesis. So two things went up and almost nothing else did, on trial denominators this cough cohort does not have.
The other symptom nobody warns about
Skin is the same shape of finding read the other way. In a FAERS analysis, skin reactions appeared in up to 8.16% of reports mentioning a GLP-1 [3]. The proportional reporting ratio was 0.27 against DPP-4 inhibitors.
That unit is reports, not patients, so no rate among users follows from it. What it does show is that this class is not the worst offender on skin. The comparator had a blistering condition of its own.
What to do about a cough
Mention it. A cough or a change in your voice within a few months of starting is worth raising. Do not assume it is unrelated.
That is the kind of thing a thorough intake would ask about at follow-up. How much follow-up a seller actually provides varies more than the prices do. That is the thread running through sellers publishing two prices and what a GLP-1 actually costs.
One question does most of the work before ordering. If something starts, who answers, and how fast? The side-effect ledger is longer than any product page publishes.