checkedPsychiatry & Psychotherapy Podcast2026-11-21last verified 2026-11-21

did a 2026 pharmacovigilance study find a 2.35 reporting odds ratio for wernicke's with glp-1 drugs?

verdict: confirmed
the primary source says it, at that magnitude
“they saw a relative odds ratio of 2.35”

a guest, a medical student said on Psychiatry & Psychotherapy Podcast, 2026-11-21, at 02:42. hear it in context → the speaker cited: 2026 narrative review of pharmacovigilance data.

the episode: Wernicke's Encephalopathy: Why We Miss It, How to Diagnose It, and How to Dose Thiamine (at 02:42). we link the show's own page; we do not re-host audio.

first, the part that holds

the number is real and correctly attached. a pharmacovigilance study published in january 2026 found a reporting odds ratio of 2.35 (95% ci 1.38-4.01) for wernicke's encephalopathy with glp-1 receptor agonists compared with other medications in the fda adverse event reporting system.

what we found when we went and looked

we went and looked, and the figure matches the primary record. the study (clinical nutrition, vol. 57, article 106571; published online 2 january 2026, february 2026 issue) combined a faers pharmacovigilance analysis with a narrative literature review. its abstract reports "increased reporting of we with glp-1 ras compared with other medications (ror = 2.35 [95%ci, 1.38-4.01])", with an ic025 of 0.29. the ratio rests on a small base: 15 cases in total, 13 of them from faers, mostly with semaglutide (8) or tirzepatide (6), and 14 of the 15 reported in 2023-2024. a reporting odds ratio measures how often an event shows up in spontaneous reports relative to other drugs. it is a safety signal, not a measure of how much risk goes up for a person taking the drug, and the speaker's quote calls it a "relative odds ratio," which is a looser label.

assertion by assertion

assertionverdictwhat the record says
a pharmacovigilance disproportionality analysis found a reporting odds ratio of 2.35 for wernicke's encephalopathy with glp-1 agonists versus other drugsconfirmedthe abstract reports ror = 2.35 (95% ci 1.38-4.01), ic025 0.29, for wernicke's encephalopathy reporting with glp-1 receptor agonists compared with other medications in faers. the lower bound of the interval is above 1 but not by much, and the count behind it is small (13 faers cases). [1]
the analysis is from 2026confirmedpubmed metadata gives an electronic publication date of 2026-01-02 and the february 2026 issue (clin nutr vol. 57:106571). the doi carries a 2025 prefix (10.1016/j.clnu.2025.106571) because the paper was accepted in 2025, but it was published in 2026. [1]
the source is a 2026 narrative review of pharmacovigilance dataconfirmedthe methods sentence reads: "we conducted a pharmacovigilance study using the fda adverse event reporting system (faers) and a narrative literature review." pubmed indexes it as both journal article and review. the 2.35 comes from the original faers analysis inside the paper, not from the review portion, so "narrative review" undersells what kind of paper it is, but the description is fair. [1]
a 2.35 odds ratio means people on glp-1 agonists have about 2.35 times the risk of wernicke's encephalopathy (the popular reading of 'relative odds ratio')misattributedthe 2.35 is a reporting odds ratio from spontaneous reports. it compares how often wernicke's encephalopathy appears among reports for glp-1 agonists with how often it appears among reports for all other drugs. it has no denominator of people exposed, so it cannot give incidence or relative risk. the authors call the event "potentially rare" and stop at a signal. [1]
the signal is consistent with other independent sourcesconfirmeda separate 2025 analysis of the who vigibase database plus 19 cases (one new case and 18 reviewed) reported disproportionate reporting of wernicke's encephalopathy for semaglutide, tirzepatide and the glp-1 agonist class. a 2026 prisma systematic review in obesity found six semaglutide case reports and cites the faers study as its pharmacovigilance reference. that review reports no disproportionality figures of its own. [2]

both directions

for the speaker: the exact number, its confidence interval, the comparator ("other medications"), the database (faers) and the 2026 publication date all match the indexed abstract. the speaker's description of the source as a 2026 review of pharmacovigilance data also fits: the paper describes itself as a pharmacovigilance study plus a narrative literature review, and pubmed tags it as a review. an independent vigibase analysis found the same direction of signal, and so did case series in which most patients had vomiting, poor intake or rapid weight loss before onset. that matches the proposed mechanism of thiamine depletion.

against: the 2.35 rests on 13 faers reports, and the lower confidence bound (1.38) and ic025 (0.29) are only modestly above the null. reporting odds ratios can be pushed up by publicity and by stimulated reporting. 14 of 15 cases came in 2023-2024, when use of these drugs and media attention both rose sharply. the comparator is every other drug in faers, not a matched population, so the ratio says nothing about absolute risk or about risk compared with other weight-loss approaches. bariatric surgery and alcohol use disorder are the classic causes of wernicke's encephalopathy. the quote's phrase "relative odds ratio" invites a reading of 2.35 times the risk, which the data cannot support. the systematic review notes that large semaglutide trials did not report wernicke's encephalopathy.

what this cannot say

spontaneous-report databases have no count of people exposed, contain duplicates and incomplete records, and cannot establish causation or incidence. a disproportionality ratio is a prompt for investigation, not a risk estimate. on our side, we read the abstract and indexing metadata through the national library of medicine's pubmed records. we did not read the full text or its tables, so we cannot check how the faers query was built, how duplicates were handled, or whether the comparator group was restricted. we also could not confirm which point in the episode the timestamp refers to, so we checked the claim as transcribed.

sources, each one fetched and read

  1. Lev D, Leibowitz A, Lang A, et al. Glucagon-like peptide-1 receptor agonists and Wernicke encephalopathy: a pharmacovigilance study and literature review. Clinical Nutrition 2026;57:106571. doi:10.1016/j.clnu.2025.106571. PMID 41534460 (PubMed XML record) What we read there: "We conducted a pharmacovigilance study using the FDA Adverse Event Reporting System (FAERS) and a narrative literature review." "Disproportionality analysis showed increased reporting of WE with GLP-1 RAs compared with other medications (ROR = 2.35 [95%CI, 1.38-4.01]...)". Electronic publication 2026-01-02; publication types Journal Article; Review. Fetched 2026-11-21 (HTTP 203). https://pubmed.ncbi.nlm.nih.gov/41534460/
  2. Bidesie J, Oudman E. Wernicke's encephalopathy following semaglutide treatment for obesity: a systematic PRISMA review of case-based evidence. Obesity (Silver Spring) 2026;34(Suppl 2):64-69 What we read there: PRISMA 2020 systematic review of six case reports; reports no ROR, IC or PRR of its own; cites the Lev et al. Clinical Nutrition 2026 pharmacovigilance study (its reference 27) for multiple WE cases with other GLP-1 RAs including tirzepatide. Fetched 2026-11-21 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/PMC13535703/

speakers are described by role, not named. one sentence is quoted under fair use, with attribution and a link to the original at the timestamp; we do not re-host audio or video. nothing on this page is a reason to start, stop or change a medication. if you have the file behind anything marked unverifiable, the research board is where people can answer. adam