Cross national pharmacovigilance analysis of drug-induced delirium: a comparative study using the US FAERS, Japanese JADER, and canadian CVAR datasets.

Background: Drug-induced delirium is a clinically important and potentially preventable adverse event, especially in older adults, yet large-scale cross-national pharmacovigilance evidence remains limited. Methods: We conducted a retrospective analysis using three major spontaneous reporting systems: the FDA Adverse Event Reporting System (FAERS, 2004-Q3 2025), the Japanese Adverse Drug Event Report (JADER, 2004-Q2 2025), and the Canada Vigilance Adverse Reaction database (CVAR, 2004-Q3 2025). Delirium cases were identified using five pre-specified Preferred Terms (PTs) selected from the narrow scope of the Standardised MedDRA Query (SMQ) Noninfectious encephalopathy/delirium. Disproportionality analyses were performed using reporting odds ratio, proportional reporting ratio, and empirical Bayes geometric mean, with signal strength defined by lower confidence bounds. Time-to-onset (TTO) was summarized using medians, IQRs, and early-onset proportions. Results: A total of 263,327 reports were identified (FAERS: 243,572; JADER: 7,512; CVAR: 12,243). Delirium was more frequently reported in females (49%-57%) and individuals aged ≥60 years (55%-70%). Strong and consistent signals were observed for nervous system drugs, particularly psycholeptics, analgesics, and psychoanaleptics. Duloxetine, lamotrigine, pregabalin, quetiapine fumarate, and zolpidem demonstrated consistent signals across databases. Additional signals involved antipsychotics, benzodiazepines, opioids, and selected anti-infective agents. Time-to-onset analysis demonstrated early-onset clustering, suggesting an acute temporal pattern. Conclusion: Cross-database consistency in drug class signals and early-onset temporal patterns highlights the need for medication review and risk mitigation in high-risk populations.