Bleeding is an adverse drug reaction (ADR) largely discussed with serotonin reuptake inhibitors (SRIs) [1]. The problem of an increased bleeding when SRIs are combined with oral anticoagulants remains open to debate, mainly due to methodological problems [2, 3]. Since multi-source approaches are mandatory in pharmacoepidemiology [4], the present study used VigiBase®, the largest WHO pharmacovigilance database in the world, to investigate putative drug-drug interactions between antidepressants and antithrombotics in general with regards of bleeding.

We performed disproportionality analyses [5,6,7] with all “serious” reports (defined according to WHO) [8] with antithrombotics (B01, according to Anatomical Therapeutic Chemical (ATC) classification, registered between January 01, 2000 and December 31, 2022) as “suspected/interacting” in Vigibase® in adults (≥ 18 years) with age and sex known and reported by physicians. Antithrombotics were divided into Vitamin K Antagonists VKA (B01AA), direct factor Xa inhibitors (B01AF), direct thrombin inhibitors (B01AE), platelet aggregation inhibitors (PAIs) (B01AC), and heparins (B01AB). Cases were reports of bleeding (according to Standardized MedDRA Queries classification) with the drug(s) of interest and non-cases all other reports with the same drug(s) of interest. Antidepressants (N06) were divided into non-selective monoamine reuptake inhibitors (NSMRIs, N06AA) and SRIs (N06AB), thus excluding monoamine oxydase inhibitors (N06AF and N06AG). Data with antidepressants (and their different pharmacological classes as described above) + antithrombotics (and their different pharmacological classes) were compared with antithrombotics (and their different pharmacological classes) alone. Results are presented as reporting odds ratios (ROR) [6, 7].

Among the 2,396,829 reports registered in VigiBase®, according to the criteria defined above, 235,813 were bleeding. Table 1 shows the number of bleeding reports with antithrombotics (and their different pharmacological classes) and their association with antidepressants (and their two different pharmacological classes). “Serious” bleeding with antidepressants + antithrombotics was mainly observed in patients aged more than 74 years: 67.5% for antidepressants + antithrombotics, 53.9% for antithrombotics alone. They caused/prolonged hospitalization in 66.2 and 81.2%, were fatal in 13.7 and 14.4%, respectively. A slight significant association was found for bleeding reports with the combination antidepressants + antithrombotics, driven by SRIs. The reporting risk was found with all antithrombotics, i.e., the 3 main clases of oral anticoagulants, as well as heparins and PAIs (Table 1).

Table 1 Case non-case study of “serious” bleeding with the association antidepressants
Full size table

The present study is, as far as we know, the first to investigate not only bleeding associated with SRIs alone [1] or with direct oral anticoagulants (DOACs) [2, 3] but also bleeding associated with antidepressants and antithrombotics in general. It confirms the excess risk for the association SRIs + DOACs (factor Xa inhibitors, thrombin inhibitors) and also extends this result to all antithrombotics. Interestingly, we found an increased reporting risk not only for VKA but also for heparins and PAIs, data not widely previously described in the literature. An association was only found for SRIs and not for NSMRIs, i.e., imipraminic antidepressants and related compounds. Once again, this result is new. The mechanism of this interaction is pharmacodynamic, i.e., addition of hemorrhagic effects of antithrombotics and SRIs, the latter being explained by the well-known role of serotonin and 5HT2A receptors on platelet aggregation [9].

Strengths and limits are well known; for example, the generalizability of VigiBase®, underreporting, and the fact that ROR evaluates the risk of bleeding reporting but not the true bleeding risk [5].

In conclusion, using an original and validated method, the present study allows generalizing the results of other previous studies: an excess of bleeding can be expected when SRIs (but not NSMRIs) are associated with antithrombotics (all antithrombotics and not only DOACs). Further studies using other pharmacoepidemiological methods are needed to confirm this pharmacovigilance signal.