Drug spectrum and disproportionality signals for diabetes insipidus in FAERS: multi-method analysis, JADER cross-database assessment, and PubMed-based case-level evidence

Objective The aim of this study is to characterize reporting patterns and non-treatment drug safety signals for diabetes insipidus (DI) in the FDA Adverse Event Reporting System (FAERS), evaluate signal stability using crude, adjusted, and shrinkage-based disproportionality methods, and assess signal reproducibility in the Japanese Adverse Drug Event Report (JADER). Methods We analyzed FAERS (2004Q1–2025Q4) and JADER (2004–2025) reports. DI cases were defined by MedDRA Preferred Terms “Diabetes insipidus” and “Nephrogenic diabetes insipidus”; central DI-related lower-level terms mapped to “Diabetes insipidus” and were not separately retrievable. Drugs used for DI diagnosis/treatment or arginine vasopressin (AVP)-deficiency states were excluded during preprocessing. All eligible non-treatment primary suspect drug–DI pairs were screened, and the main multi-method summary focused on the 50 most frequently reported drugs. Published cases were summarized as case-level evidence, not comparative risk meta-analysis. Results FAERS contained 3,444 Preferred Term (PT)-defined DI reports; 3,325 entered the main non-treatment analysis. Raw annual counts increased, whereas normalized reporting proportions fluctuated. Complete screening identified 112 crude reporting odds ratio (ROR)-positive drug–DI pairs. Among the frequency-ranked top 50, 47 met crude ROR criteria, 38 remained positive after adjustment, 32 met empirical Bayes geometric mean (EBGM) criteria, and 40 met information component (IC) criteria; in addition, 36 were positive in at least three methods. In JADER, 696 DI reports entered cross-database assessment; eight drugs met crude ROR criteria, eight met adjusted ROR criteria, three met EBGM criteria, and three met IC criteria. Lithium, hydrocortisone, and propofol were positive across all four JADER methods. Conclusion After treatment-drug exclusion, DI reports clustered mainly among nervous system, antineoplastic/immunomodulating, and systemic hormonal agents. Several non-treatment drugs showed persistent disproportionality, but JADER support was limited. These findings are hypothesis-generating signals, not causal evidence.