The recorded difference is large and the measured difference is unknown. In a matched cohort of people taking tirzepatide, coded nausea and vomiting ran at a hazard ratio of 0.50 in men against women, 95% CI 0.37 to 0.67 [1]. That counts diagnosis codes rather than symptoms, and nobody has run the study that asks men and women the same question and compares the answers. Since side effects are the commonest stated reason for stopping, the gap matters to how long anybody stays on one.
What the cohort compared
Adults with obesity and heart failure with preserved ejection fraction, all taking tirzepatide, drawn from a network of 110 healthcare organizations. Men and women were matched one to one, leaving 1,654 in each group, and followed for a year. Everyone was on the drug, so there is no untreated arm, and the authors are direct that this must not be read as evidence the drug works differently by sex.
Cardiovascular outcomes showed no significant differences. Heart failure exacerbation came in at 1.14, 95% CI 0.90 to 1.44, major adverse cardiovascular events at 1.28, 95% CI 0.99 to 1.65, and all-cause death at 1.73, 95% CI 0.95 to 3.18. That last one is not significant, and it is also a point estimate approaching double with an interval reaching past triple. A summary sentence reporting no significant differences is accurate and does not convey it.
One feature of the study is worth copying. The authors used two prespecified falsification endpoints, chosen because tirzepatide should have no effect on them, so that an association would signal residual confounding. Neither showed anything.
The skin data, which points the other way
An analysis of FDA adverse event reports from 2018 to 2024 examined skin reactions attributed to these drugs [2]. Cutaneous events appeared in up to 8.16% of GLP-1 reports, more often in women, at a mean age of 60. Against DPP-4 inhibitors the proportional reporting ratio was 0.27, 95% CI 0.257 to 0.284.
Two cautions attach. That 8.16% is a share of adverse event reports rather than of patients, and only a small fraction of people taking any medicine ever generate a report. And DPP-4 inhibitors are associated with bullous pemphigoid, so beating them on skin reporting is a lower bar than beating a neutral comparator.
What survives is the direction: women appear more often in the skin reports and less often in the coded nausea. Both are reporting measures, and reporting is exactly where the sexes are known to differ.
What the trial evidence says, which is nothing about sex
The best-powered evidence on gastrointestinal harm is a meta-analysis of 55 placebo-controlled randomized trials covering 106,395 participants [3]. Gallstones carried a risk ratio of 1.46, 95% CI 1.09 to 1.97, and reflux 2.19, 95% CI 1.48 to 3.25, which work out at roughly 2 and 4 extra cases per 1,000 people. Pancreatitis, bowel obstruction, ileus, perforation, bleeding and gastroparesis all came back null.
None of that is reported by sex. Randomized trials collect symptoms systematically and would be the right place to answer this question, and the pooled analyses have not asked it. The absolute figures are set out in what the side effects are.
What a buyer takes from it
Very little that changes a purchase. The cohort is a narrow clinical population, everybody in it was already on the drug, and the comparison is between sexes rather than between choices anybody makes at a checkout.
What travels is the reading. When a study reports side effects from records rather than from asking people, it is measuring documentation as well as experience. Side effects are what the persistence literature names as the commonest stated reason for stopping, and that is the risk a prepaid term makes a buyer carry.