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

Abstract

Inhibition of serotonin uptake in platelets seems to be the crucial mechanism underlying SSRI-associated haemorrhages. This effect is also present in antidepressants featuring non-selective serotonin reuptake inhibition (non-SSRI). Impact of selectivity of serotonin reuptake and/or affinity to the serotonin reuptake transporter on the bleeding risk have not yet been studied sufficiently. We retrieved country- and SSRI-/non-SSRI-specific data from the Uppsala Monitoring Centre and used a case/non-case approach to calculate substance-specific reporting odds ratios (ROR) to evaluate the statistical association of treatment with SSRI/non-SSRI and haemorrhages. Country-specific analysis revealed no clear trends towards an increased risk of bleeding related to particular agents of group SSRI/non-SSRI (sporadically ROR>1 for citalopram, duloxetine, escitalopram, fluvoxamine, paroxetine, sertraline, St. John's wort). There was a clear trend in the total dataset towards a "reduced protective effect" (suggested by ROR<1) on the development of haemorrhages with agents featuring comparatively high affinity to the 5-HTT and/or selective serotonin reuptake inhibition (as with escitalopram, citalopram, duloxetine or venlafaxine) in comparison to agents with lower affinity or non-selective serotonin reuptake inhibition (as with mirtazapine or doxepin). Comparison of group-specific aggregated data (SSRI vs. non-SSRI) revealed significant differences regarding the "protective effect" on the development of haemorrhages between groups SSRI vs. non-SSRI in favour of non-SSRI in nearly all countries as well as in the total dataset. Our findings provide preliminary evidence that agents with increased affinity to the 5-HTT and/or selective serotonin reuptake inhibition may be associated with an increased risk of bleeding.

Keywords: Depression; Haemorrhages; Pharmacovigilance; Reporting odds ratio; Signal detection.