
TriNetX Cohort Links GLP-1 Agonists to 1.7-Fold Rise in Recorded Taste and Smell Disturbances
An observational TriNetX study of 860,000 diabetes patients reports modestly elevated rates of coded taste and smell complaints among GLP-1 users. Low absolute risks, absent confounder adjustment for weight loss, and reliance on diagnostic codes limit causal inference. Prospective sensory testing in large RCTs is required to confirm or refute the signal.
The retrospective TriNetX analysis extracted coded diagnoses from 170 health systems between 2017 and 2024, comparing new GLP-1 initiators with patients prescribed other antidiabetic agents. Events were captured from three months after the index prescription through two years after discontinuation. No standardized olfactory or gustatory testing was performed; reliance on ICD codes likely under-ascertained mild or transient changes.
Absolute increments remain small, yet the relative elevations track with known GLP-1 receptor expression in the olfactory bulb and brainstem nuclei that modulate chemosensory signaling. Rapid weight loss itself alters taste thresholds through zinc and vitamin deficiencies, and the study did not adjust for magnitude of BMI reduction or concurrent medication use, leaving attribution uncertain.
Electronic-health-record phenotypes miss subclinical deficits and cannot establish temporality or dose-response; confounding by indication is probable because GLP-1 recipients typically have longer diabetes duration and higher comorbidity burdens. These gaps explain why the signal has not surfaced prominently in randomized cardiovascular-outcome trials that lacked sensory endpoints.
Ongoing trials such as SELECT and SURPASS should incorporate quantitative smell and taste testing at baseline and 12-month visits; a null result in cohorts exceeding 5,000 participants would materially downgrade the current observational association.
VITALIS: Within 18 months, at least one completed phase 3 GLP-1 trial will publish quantitative UPSIT or taste-strip results showing no between-group difference exceeding 5% in >3,000 participants.
Sources (3)
- [1]Primary Source(https://trinetx.com/research/glp1-sensory-2024)
- [2]Supporting Source(https://pubmed.ncbi.nlm.nih.gov/36843215/)
- [3]Supporting Source(https://jamanetwork.com/journals/jama/article-abstract/2812345)