Integrated Evidence from VigiBase and Clinical Trials: A Comprehensive Pharmacovigilance Analysis of Seven Glucagon-Like Peptide 1 Receptor Agonists (GLP-1 RAs)
Li J, Liang J, Zhang W, He J, Ye X
Diabetes Therapy (2026) · n=348,649
Across 348,649 GLP-1RA adverse-event reports in the WHO VigiBase database through January 2025, the seven approved GLP-1 receptor agonists show distinct drug-level safety signatures — tirzepatide with abdominal pain (ROR025 53.54), liraglutide with drug-ineffective reports (31.14) and pancreatitis (4.24), exenatide with injection-site pain (70.14) — that complicate the simple class-effect framing and inform personalised prescribing.
This 2026 Diabetes Therapy paper from the Naval Medical University Shanghai pharmacovigilance group is the most comprehensive published disproportionality analysis of the seven approved GLP-1 receptor agonists in the WHO VigiBase database, designed to identify drug-specific rather than class-wide safety signals across the full approved-product landscape — semaglutide, liraglutide, tirzepatide, dulaglutide, exenatide, lixisenatide, and albiglutide. The analysis covers 348,649 GLP-1RA-related adverse event reports through January 2025, with cross-validation against published randomised controlled trial data.
The headline finding is that the seven drugs show distinct rather than uniform safety signatures, complicating the simple "class effect" framing that has dominated GLP-1RA safety communication:
- Tirzepatide stood out for abdominal pain reporting (ROR025 53.54), consistent with the trial-population GI tolerability concentration of the drug.
- Liraglutide showed elevated "drug ineffective" reports (31.14) and pancreatitis (4.24), with the pancreatitis signal anchoring the long-standing class concern at the molecule with the longest post-marketing exposure history.
- Exenatide showed elevated injection-site pain (70.14), reflecting the device-formulation history of the earliest-approved member of the class.
- Albiglutide showed extreme device-use-error signal (1424.33), reflecting the now-withdrawn product's pen-device-specific user-experience problems.
Gastrointestinal disorders were the most frequent system-organ-class for all GLP-1RAs, consistent with trial evidence. Male patients and younger adults (18-44 years) showed higher positive reporting rates across most signals. The authors conducted parallel meta-analysis of RCT data to validate the VigiBase mining results, generally finding directional consistency between spontaneous reporting and trial-level adverse-event rates.
This is a pharmacovigilance disproportionality analysis, not a controlled cohort study — VigiBase data is subject to severe selection bias, reporter-attribution bias, and detection bias from media-driven attention. The reporting odds ratio (ROR) is a relative-frequency-of-reports metric within the database, not a population-incidence estimate; an elevated ROR for one drug does not establish absolute risk per treated patient, and inter-drug comparisons are confounded by differential prescription volume, post-marketing time-on-market, geographic distribution of approvals, and population-level prescription patterns.
The seven-drug landscape spans 15+ years of approval timing — exenatide (2005), liraglutide (2010), albiglutide (2014, withdrawn 2018), dulaglutide (2014), lixisenatide (2016), semaglutide (2017), tirzepatide (2022) — which means the underlying reporter populations, geographic distributions, prescription patterns, and post-marketing surveillance regimes are very different across the seven drugs. Higher reporting rates for older drugs may reflect longer surveillance history rather than higher true risk, and the reverse may apply to newer drugs.
The albiglutide device-use-error signal is real but historical — the product is withdrawn from market and the signal is principally a device-design artefact rather than a generalisable GLP-1RA finding. Including it in the cross-drug comparison potentially distorts the "seven drugs" framing.
The cross-validation against RCT meta-analysis is methodologically appropriate but RCTs systematically under-detect rare adverse events that VigiBase reporting may over-represent, and the two evidence sources should be read as complementary rather than fully concordant. The authors do not have access to denominator-level prescription data for the relevant drug-years, so the analysis cannot generate absolute incidence rates.
The Chinese-language reporter contribution is substantial in recent VigiBase data; cultural and regulatory differences in reporting practices may differentially affect signal detection across drugs.
The authors declare no conflicts of interest. The methodology is appropriate for the disproportionality-analysis tradition but inherits the underlying limitations of the data source.
07·Member discussion
No member discussion yet.
Member-only conversation lives here — cycle notes, practitioner commentary, pattern-matching. Be the first paying member to start the thread.