Agent-specific bleeding signatures in non-hypertensive versus hypertensive patients exposed to oral anticoagulants: a real-world pharmacovigilance study.

Oral anticoagulants (OAC) are effective in thrombo-embolic disorders, but their real-world bleeding profiles in hypertensive versus non-hypertensive users remain unknown. This study provides an exploratory, real-world assessment of the population-level, agent-specific safety hierarchy of direct OAC (DOACs; apixaban, dabigatran, edoxaban, rivaroxaban) and warfarin. We queried the FAERS database (Q1 2004-Q2 2025) using four disproportionality methods, requiring concordant signals from all four algorithms (a high-specificity, lower-sensitivity approach), and stratified by hypertension status as defined by MedDRA-coded diagnosis terms. A clinical priority matrix (0-8) incorporating salience, reporting density, robustness, and fatality was then applied to tier bleeding events. Among non-hypertensive reporters (n = 101,469), rivaroxaban dominated case volume (50,676; 49.9%), a finding that primarily reflects its market dominance rather than a higher intrinsic risk. In descriptive reporting patterns, dabigatran showed the highest proportion of fatal outcomes among its cases (23.7%). In the much smaller hypertensive subset (n = 4,252), edoxaban (based on only 142 reports) showed a fatality proportion of 28.9%; however, this estimate is based on a small sample and is subject to instability ("small cell" bias), warranting cautious, hypothesis-generating interpretation only. Hypertension was associated with a shift in the anatomical signal pattern from gastrointestinal and vascular-malformation bleeding observed in non-hypertensive toward deep intracerebral, traumatic, and adrenal hemorrhages in OAC. Temporal occurrence revealed early-phase rivaroxaban predominance in both strata, whereas warfarin exhibited a delayed but sustained reporting tail beyond 360 days. Clinical priority scoring identified brain stem hemorrhage as the only universally tier-5 event across DOACs and cerebral microhemorrhage was flagged moderate only in rivaroxaban in non-hypertensive patients. Edoxaban further generated unique tier-5 signals for cerebellar and pulmonary alveolar bleeding in both strata, though these findings are based on small numbers and require replication. Hypertension appears to be a qualitative effect-modifier that reallocates the population burden of OAC-related hemorrhage and redirects bleeding topography in this spontaneous reporting dataset. Agent-specific, blood-pressure-adjusted risk algorithms warrant further investigation and could eventually inform future guidelines, but prospective validation is required before clinical implementation.