GLP-1 drugs and diabetic foot reports

A 2026 pharmacovigilance study found that diabetic foot complications made up a smaller share of adverse-event reports for GLP-1 receptor agonists than for other glucose-lowering drugs. The result is a reporting pattern, not proof that GLP-1 drugs prevent foot ulcers or amputations. The researchers analysed reports rather than enrolled patients, then added a drug-target Mendelian randomization analysis. Neither method was a randomized treatment trial^[1]. Readers following semaglutide should keep that distinction in view.

What the researchers analysed

The paper used AERSMine to search the FDA Adverse Event Reporting System from the second quarter of 2005 through the second quarter of 2024. It identified 1,819 adverse-event reports for GLP-1 receptor agonists and 17,206 reports for other medicines in the ATC-A10 class, which covers drugs used in diabetes^[1]. These are counts of submitted reports, not counts of everyone who used each medicine. The abstract does not provide an exposure denominator such as prescriptions or person-years.

FDA says its reporting database contains adverse-event reports, medication-error reports and product-quality complaints associated with adverse events. A report does not establish that a medicine caused the event. Reports may lack enough detail for evaluation, some events are never submitted, publicity can change reporting behaviour, and duplicates can occur^[2]. That is why the agency says these data cannot calculate adverse-event incidence in the US population. Our semaglutide safety overview applies the same caution to spontaneous reports.

The main reporting result

Diabetic foot complications appeared in 6.38 per 1,000 reports in the GLP-1 receptor agonist group and 11.31 per 1,000 reports among the other ATC-A10 drugs. The proportional reporting ratio was 0.56, with a 95% confidence interval from 0.54 to 0.59. The reported P value was below 0.001^[1]. A PRR below 1 means the selected event occupied a smaller fraction of reports for the GLP-1 group than it did in the comparison group. It does not mean that treatment cut a patient's risk by 44%.

The same direction appeared in the molecule-level analyses reported in the abstract. Semaglutide had a PRR of 0.45 (95% CI 0.39 to 0.51), dulaglutide 0.49 (0.45 to 0.54), and liraglutide 0.30 (0.26 to 0.35)^[1]. Those estimates compare reporting proportions, not randomized groups. They should not be used to rank medicines for diabetic-foot protection. The clinical and regulatory context for one of these drugs is available on our semaglutide evidence page. Its evidence limitations are particularly relevant when a reporting signal is mistaken for a clinical outcome.

Why the denominator changes the meaning

A denominator of all submitted reports answers a narrow question: among reports associated with a drug group, how prominent was the selected event? It does not tell us how many treated patients developed a diabetic foot complication. That would require reliable exposure data and a design able to account for baseline disease severity, diabetes duration, vascular disease, kidney disease, neuropathy, foot care and other treatment differences. The distinction also matters when reading the semaglutide research record.

Reporting can also differ between older and newer medicines or after media attention. FDA specifically lists time on the market and publicity as factors that influence whether an event is reported^[2]. Treatment selection can also shape the groups. People receiving a GLP-1 receptor agonist may differ from those receiving another glucose-lowering drug in ways the report database cannot fully reconstruct. A lower reporting proportion can therefore be compatible with several explanations, including confounding and reporting behaviour. It is not a treatment recommendation. See the semaglutide regulation section for the prescription boundary.

What the Mendelian randomization adds

The authors also conducted a two-sample drug-target Mendelian randomization analysis to investigate whether genetically proxied GLP-1 receptor agonism was associated with diabetic foot complications. The abstract says those results generally corroborated the retrospective reporting analysis^[1]. It does not provide the genetic instruments, outcome definitions, effect estimates, sensitivity results or assumptions needed for an independent appraisal. This summary therefore does not attach an effect size to that part of the work.

Mendelian randomization can test whether genetic variants used as proxies are consistent with a causal hypothesis. It does not randomize patients to semaglutide or another drug, and a lifetime genetic proxy is not identical to taking a medicine at a clinical dose. Agreement between the two analyses makes the signal more interesting. It does not turn spontaneous-report data into evidence that a prescription drug prevents foot complications. Our semaglutide mechanism section describes the drug separately from this genetic proxy.

What this means for diabetic foot care

The study supports a research question, not a change in care. Diabetic foot complications are clinically serious and need assessment based on the individual patient, not a database ratio. Someone with a new wound, colour change, swelling, warmth, drainage, loss of sensation or worsening pain should seek prompt clinical advice rather than infer protection from a GLP-1 prescription. Do not start, stop or switch a medicine on the basis of this analysis. The semaglutide safety section is context, not a substitute for assessment.

For readers using semaglutide, the practical conclusion is deliberately limited. The paper did not test a foot-care strategy, compare ulcer healing, or report amputation outcomes from a prospective cohort. It examined how often a grouped event appeared within submitted reports. Medication decisions and diabetic-foot prevention belong with clinicians who can assess glycaemic control, circulation, sensation and the foot itself.

What we do not yet know

The PubMed abstract does not show how each diabetic-foot term was defined, how duplicate or incomplete reports were handled, or how results changed across reporting eras. It gives no treatment-exposure denominator and no prospective clinical endpoint. The Mendelian randomization methods and estimates are also not reported in enough detail there to judge instrument strength or possible pleiotropy^[1]. Full-text appraisal may answer some of these questions, but it cannot remove the basic limitations of spontaneous reporting. The same evidence hierarchy applies across our semaglutide overview.

A useful next step would be a prospective study or a carefully designed active-comparator cohort with validated foot outcomes, known exposure time and adjustment for baseline diabetic-foot risk. Until such evidence exists, the honest reading is that GLP-1 receptor agonists had a lower proportional reporting signal for diabetic foot complications in this database. Whether that reflects a clinical protective effect remains unproven. Our semaglutide research summary keeps observational signals separate from treatment evidence.