FAERS Screens 17 Drug-Associated IBD Safety Signals
Key Takeaways
- Across 52,395 FAERS reports representing 50,426 patients, the pharmacovigilance screen identified a defined set of suspected drug-associated inflammatory bowel disease signals.
- Seventeen drugs reported at least 50 times met all four prespecified signal criteria in FAERS.
- Anti-TNF-α inhibitors accounted for the greatest reporting volume, while isotretinoin showed the strongest signal intensity.
- JADER suggested similar major class-level patterns despite a lower minimum report threshold than FAERS (at least three reports vs at least 50), and among the 16.02% of FAERS reports with usable onset dates, most median onsets occurred within 1 year and most modeled drugs in FAERS showed an early-failure pattern, although later-onset cases also occurred.
- Because FAERS and JADER are spontaneous-reporting systems, the observed associations remain hypothesis-generating rather than causal.
A retrospective pharmacovigilance analysis in Ying et al. pharmacovigilance study of drug-associated inflammatory bowel disease in FAERS and JADER used FAERS from 2004 Q1 through 2025 Q2 with external validation in JADER from January 2004 through November 2025. Case finding used Medical Dictionary for Regulatory Activities (MedDRA) preferred terms for inflammatory bowel disease, ulcerative colitis, and Crohn's disease, with analyses limited to primary suspect drugs. Reports listing IBD as an indication were excluded, as were drugs approved exclusively for IBD treatment, and a positive signal required concordance across reporting odds ratio (ROR), proportional reporting ratio (PRR), Information Component (IC), and Empirical Bayes Geometric Mean (EBGM). Prespecified subgroup analyses examined age, sex, and IBD subtype, and investigators also evaluated time to onset with Weibull modeling.
The FAERS search identified 52,395 reports involving 50,426 patients, and 17 drugs reported at least 50 times met all four positivity criteria. Reports were concentrated in adults, and anti-tumor necrosis factor-alpha (anti-TNF-α) inhibitors accounted for the greatest reporting volume.
Signal strength did not track directly with report count. Isotretinoin showed the strongest disproportionality signal in FAERS at ROR 110.51, while adalimumab led by report count and infliximab also ranked prominently. Notably, this analysis was the first to detect positive IBD signals for upadacitinib and risankizumab, contrasting with an earlier FAERS-based study using data through Q4 2023 that found no signal for either drug — a difference the authors attribute to the longer follow-up window (through Q2 2025) allowing more case reports to accumulate for these more recently approved agents.
JADER suggested similar class-level patterns for anti-TNF-α inhibitors, anti-leukocyte or anti-interleukin biologics, conventional immunosuppressants, and antibiotics, although its signal detection was performed on drugs reported at least three times rather than at least 50 times in FAERS, and substantial heterogeneity between JADER and FAERS was also noted. Subgroup findings were heterogeneous rather than uniform, with stronger signals in older adults for most drugs, younger-skewing signals for isotretinoin and ozanimod, stronger ulcerative colitis signals for isotretinoin, and stronger Crohn's disease signals for most anti-TNF-α agents.
Among the 5,673 reports with complete time-to-onset data—16.02% of the cohort, because 83.98% lacked usable dates—most median onsets fell within 1 year, and 11 of 14 analyzed drugs showed an early-failure Weibull pattern.
In practical terms, the analysis suggests reported events tended to cluster earlier after treatment initiation and then decline over time, but this does not mean all events clustered in the first year; some drugs had median onset beyond 400 days and cases also occurred after year 1.
Interpretation remains limited by the structure of spontaneous-reporting systems. FAERS and JADER are vulnerable to underreporting, variable reporter identity, incomplete baseline clinical information, and the absence of exposure denominators, so they cannot establish absolute risk or causality. Residual confounding from underlying disease activity, comorbidities, and concomitant medications also remains unresolved, and differences in case volume, exposure patterns, and reporting practices limit direct FAERS-JADER comparison.
Across both databases, several drug classes showed broadly similar class-level disproportionality patterns for IBD reporting, with isotretinoin standing out in FAERS by signal strength. The overall pattern remains hypothesis-generating and still requires prospective validation to distinguish true drug-associated IBD from reporting artifacts or clinical confounding.
Clinician Questions
How can anti-TNF or other IBD therapies still appear in drug-associated IBD safety-signal searches?
Anti-TNF agents and other therapies used in IBD or related immune-mediated diseases can still appear because spontaneous reports name a medication as the primary suspect drug for an IBD event, not as proof that the drug caused the event. Excluding reports that listed IBD as an indication and removing drugs approved only for IBD reduced some misclassification, but paradoxical reactions and indication-related confounding still make background disease activity hard to separate from a true drug-associated event.
What counted as a drug-associated IBD case in the FAERS and JADER search?
Cases were defined with MedDRA preferred terms for inflammatory bowel disease, ulcerative colitis, and Crohn's disease, and analyses were restricted to primary suspect drugs. Records were excluded if IBD appeared as an indication, and drugs approved exclusively for IBD treatment were removed, making this a pharmacovigilance signal-detection definition rather than a clinical diagnostic standard.
Which patient subgroups showed different drug-associated IBD signal patterns?
Most drugs showed stronger disproportionality in older adults, whereas isotretinoin and ozanimod skewed younger. Several anti-TNF-α agents, isotretinoin, ozanimod, and upadacitinib showed stronger male signals, doxycycline signaled only in females, isotretinoin was stronger in ulcerative colitis, and most anti-TNF-α signals were stronger in Crohn's disease. These subgroup differences remain exploratory because spontaneous-reporting data cannot separate biology from reporting behavior.
What does an early-failure Weibull pattern mean for reported onset of drug-associated IBD?
An early-failure Weibull pattern means reported IBD events tended to cluster relatively soon after treatment initiation and then taper over time rather than becoming more common later in treatment. The random-failure pattern described for ixekizumab, risankizumab, and teduglutide instead suggests a more even distribution of reported onset across the treatment course.
Recommended Reading
- For more on anti-TNF–experienced Crohn’s disease: Risankizumab Beats Ustekinumab for Steroid-Free Crohn’s Remission
- For more on anti-TNF use in IBD care: IBD Masterclass: Choosing the Right IBD Therapy - A Patient-Clinician Connection
- For more on positioning anti-TNF therapy in IBD: IBD Care Reimagined: Personalized, Proactive, and Target-Driven Management