Risk of Bleeding Related to Selective and Non-selective Serotonergic Antidepressants: A Case/Non-case Approach Using Data from Two Pharmacovigilance Databases

Abstract

Signal detectioning in 2 pharmacovigilance databases suggest that serotonin reuptake inhibition is not associated with an increased risk of bleeding. However, underreporting may have accounted for the evaluated absent associations, particularly concerning SSRI. Regarding the detected increased risk of bleeding associated with hypericum, pharmacokinetic drug-drug interactions may be relevant independent of serotonin reuptake inhibition.

Association between haemorrhages and treatment with selective and non-selective serotonergic antidepressants: Possible implications of quantitative signal detection

“…Surprisingly, in the total dataset (global perspective) treatment with SSRI was not statistically associated with haemorrhages, whereas both positive control agents (ASS and diclofenac) features positive associations. We have reported this finding already previously (Gahr et al, 2015). Accordingly, apart from St. John's wort/hypericum, no statistical association was found between any of the agents of group non-SSRI and haemorrhages in the total dataset.…”

“…In accordance with the results of our previous study (Gahr et al, 2015) the used data mining approach applied to several country-specific datasets did not allow to answer the question, if treatment with non-SSRI is associated with a clinically relevant increased risk of bleeding. ADR ¼adverse drug reaction; SSRI ¼ selective serotonin reuptake inhibitor; non-SSRI ¼ antidepressants with non-selective serotonin reuptake inhibition; SMQ ¼standardised Medical Dictionary for Regulatory Activities query; NA ¼ not applicable due to no reports or no recording of the respective substance(s) in the respective dataset; UK¼ United Kingdom; USA¼ United States of America.…”

“…In the total dataset (global perspective) only St. John's wort/hypericum was statistically associated with haemorrhages (ROR 1.31 [95% CI 1.06-1.63]), whereas in the entire dataset no other agent of group non-SSRI and no agent of group SSRI was associated with an increased bleeding risk (these findings were already published (Gahr et al, 2015)). Country-specific analysis revealed no clear trends towards an increased risk of bleeding related to particular agents of group SSRI or non-SSRI (for details see Table 3).…”

“…Based on our recent findings (Gahr et al, 2015) we assumed that in the group of agents featuring ROR-values o1 substancespecific differences concerning the extent of the associated "protective effect" on bleeding events may be present (ROR-values 4 1 are statistically associated with an increased risk of bleedings, ROR-values o 1 are statistically associated with an reduced risk of bleeding ¼"protective effect"). We hypothesised that substancespecific differences in the ROR-values (between 0.01 and 0.99) may be associated with different levels of the associated "protective effect" on bleeding events.…”

“…In this regard we have recently performed a data mining approach within two pharmacovigilance databases (WHO-database and BfArM/AkdÄ-database in Germany) in order to detect safety signals concerning haemorrhages related to SSRI vs. non-SSRI (Gahr et al, 2015). Surprisingly, we could only detect signals for hypericum and the positive control agents (acetylsalicylic acid [ASS] and diclofenac), whereas none of the agents of group SSRI or non-SSRI was statistically associated with haemorrhages.…”