A search of the MEDLINE and PubMed databases was performed to identify articles describing SSRI-associated bleeding and PK/PD changes when co-administered with anticoagulants and antiplatelets. Combinations of the following Medical Subject Headings and terms were searched: gastrointestinal bleeding, intracranial bleeding, serotonin reuptake inhibitors, warfarin, anticoagulants, clopidogrel, aspirin, platelet aggregation inhibitor, pharmacokinetics, and pharmacodynamics. The search was restricted to abstracts and articles on human studies in English with no date limits. References cited in the articles identified by the search were also evaluated and included if relevant. Fourteen observational studies related to risk of GI bleeding or intracranial bleeding associated with SSRI use were included. Descriptive statistics are reported as mean (SD) unless otherwise specified. 

Several pharmacologic mechanisms may provide an explanation for the actions of SSRI antidepressants to increase the risk of bleeding. Selective serotonin reuptake inhibitors are known to downregulate serotonin (5-hydroxytryptamine [5HT]) receptors not only in the brain but also in blood platelets.⁷ Only neurons and platelets express 5HT receptors in a nonactivated state. The majority of 5HT is synthesized in the GI tract by enterochromaffin cells.⁸ 5HT is then taken up by platelets and metabolized by the liver or pulmonary vascular endothelium. Approximately 99% of the whole 5HT is stored in platelets.⁹ Another mechanism of action for SSRI-induced bleeding is increased gastric acidity leading to potential ulcerogenic effects.¹⁰ 

Selective serotonin reuptake inhibitors inhibit the 5HT transporter protein that blocks the uptake of synaptic 5HT into the presynaptic neuron and, similarly, the entrance of 5HT into the platelet.^11,12 The mechanism of this blockade in the presynaptic neuron is demonstrated by paroxetine and (S)-citalopram, which lock 5HT by lodging into the 5HT transporter.¹³ Based on in vitro and in vivo studies, the 5HT transporter blockade results in a wide spectrum of antiplatelet activity.^11,12 Various mechanisms include the blockade of intraplatelet calcium mobilization in the coagulation cascade; inhibition of nitric oxide synthase; depletion of intracellular 5HT stores; decreased platelet secretion in response to collagen; decreased secretion of platelet factors in response to a chemical or traumatic stimulation; and decreased expression of membrane receptors involved with platelet activation and blood vessel vasoconstriction or vasodilation.^8,11,12,14 Nitric oxide (NO) synthase is required for NO production from the NO donor L-arginine.¹⁵ Nitric oxide activates cyclic guanosine monophosphate, which acts to relax smooth vascular muscle and regulate platelet aggregation.¹⁵ 

Meijer et al¹⁶ proposed a significant correlation between antidepressant potency for 5HT reuptake inhibition and bleeding. Antidepressant potency was divided into 3 groups based on dissociation constant for the 5HT transporter: high (0-1 nmol/L); intermediate (1-10 nmol/L); and low (≥10 nmol/L). The high degree of 5HT reuptake noted for fluoxetine, sertraline, clomipramine, and paroxetine was associated with most of the case reports of abnormal bleeding (53.1%) compared with 15 other antidepressants. The adjusted odds ratio (OR) for bleeding risk for patients who took these high-potency antidepressants was 2.6 (95% CI, 1.4-4.8) compared with patients using the other agents. 

Several SSRIs undergo biotransformation to metabolites that may also influence bleeding and platelet aggregation. Fluoxetine is converted to norfluoxetine, which has an elimination half-life of 7 to 15 days.¹⁷ Physicians should be aware of this factor when prescribing fluoxetine because norfluoxetine is as potent as fluoxetine for inhibiting 5HT reuptake.¹⁸ Sertraline's metabolite N-desmethylsertraline is 5 to 10 times less potent for 5HT reuptake, with lower concentrations detected in patients.^19-21 Both compounds were reported to significantly (P<.05) inhibit platelet aggregation induced by adenosine diphosphate, collagen, and thrombin.²² Paroxetine metabolites possess very weak 5HT inhibition compared with the parent drug.²³ The effects of SSRIs to inhibit platelet aggregation may be dose-dependent, as shown by in vitro studies and clinical case reports.^23,24 

Although not commonly recognized by physicians, SSRIs have been shown to increase gastric acid secretion directly. This action can potentially enhance the likelihood of ulcerogenic effects leading to GI bleeding.¹⁰ The role of 5HT in gastric acid secretion was investigated by a series of animal studies. One of the studies²⁵ examined the use of a fluoxetine pretreatment that potentiated a metabolically stable thyrotropin analogue RX77368. Thyrotropin-releasing hormone and RX77368 act in the brain to stimulate gastric function and enhance gastric acid and pepsin secretion, motility, gastric emptying, mucosal blood flow, and, therefore, the formation of gastric erosions.^25-27 

The 5HT pathways are known to innervate a number of brain areas, including the dorsal vagal complex.²⁸ Vagal stimulation releases 5HT into the GI tract, where specific 5HT receptor subtypes will modulate gastric acid secretion. Gastric acid inhibition modulated by 5HT occurs with 5HT1 receptor agonists but not 5HT1A receptor agonists.²⁹ Other 5HT receptor agonists (eg, 5HT2, 5HT3) are not involved with gastric acid inhibition.²⁹ Vagal stimulation can increase the basal rate of 5HT release into the gastric lumen and portal circulation by 600% and 265%, respectively.³⁰ Osteopathic manipulative treatment directed to areas of vagus nerve irritation could potentially attenuate gastric acid secretion because of the parasympathetic component of this mechanism. Balancing this treatment with the treatment of somatic areas related to sympathetic tone—originating from the thoracic spine and traveling through the celiac ganglion—is an important consideration as well. 

Both fluoxetine and sertraline may have a dose-dependent effect on gastric acid secretion, possibly attenuated by vagotomy.³¹ Fluoxetine has also been shown to potentiate RX77368-induced gastric acid secretion, which suggests an interaction between 5HT and thyrotropin-releasing hormone to promote gastric acid secretion.²⁵ The concomitant use of paroxetine and aspirin may produce a dose-dependent increase in gastric acid secretion greater than either agent alone.³² This pharmacologic mechanism may account for the increased likelihood of upper GI bleeding when SSRIs and NSAIDs are used together. 

The association between SSRIs and an increased risk of GI bleeding is well described in the literature but yields inconsistent results. A systematic review and meta-analysis by Jiang et al³³ of 22 cohort and case-controlled studies involving more than 1 million people reported 1.55-fold higher odds of upper GI bleeding in SSRI users compared with nonusers (95% CI, 1.35-1.78). In subgroup analyses, the risk was found to be greatest among participants taking concurrent NSAIDs or antiplatelet medications. There was considerable heterogeneity among the included studies (I²=88.9%; P<.001). A meta-analysis of observational studies in 2017 by Laporte et al³⁴ included 42 studies accounting for greater than 1 million people who experienced severe bleeding with SSRI use. A 41% increased risk of severe bleeding was noted for SSRI users (95% CI, 1.27-1.57), which was mostly attributed to GI bleeding (22 studies: OR, 1.55; 95% CI, 1.32-1.81). 

The first epidemiologic study to support this association was conducted by de Abajo et al³⁵ in 2009. Using the United Kingdom–based General Practice Research Database, patients aged 40 to 79 years between April 1993 and September 1997 were matched to 10,000 controls for age, sex, and time based on the source population. Use of SSRIs was present in 52 participants (3.1%) among 1651 incident cases of upper GI bleeding compared with 95 of 10,000 participants (1%) among the control group (adjusted OR, 3.0; 95% CI, 2.1-4.4). Relevant factors included non-SSRI antidepressants (relative risk [RR], 1.4; 95% CI, 1.1-1.9) and concurrent NSAID (adjusted RR, 15.6; 95% CI, 6.6-36.6) or low-dose aspirin (adjusted RR, 7.2; 95% CI, 3.1-17.1) use. No differences were found among individual SSRIs, and use was not associated with the 248 cases of ulcer perforation. The conclusions of de Abajo et al³⁵ are supported by findings in recent observational studies presented in Table 1.^6,36-41 

Dall et al³⁹ reported a more modest association of SSRI use and GI bleeding. Using 3 databases from Denmark, this case-controlled study identified 3652 cases of serious upper GI bleeding from 1995 to 2006. Use of SSRIs was associated with increased risk of upper GI bleeding in current users (adjusted OR, 1.67; 95% CI, 1.46-1.92), recent users (adjusted OR 1.88; 95% CI, 1.42-2.5), and past users (adjusted OR, 1.22; 95% CI, 1.07-1.39). Concurrent users of NSAIDs and SSRIs had an elevated risk of much greater magnitude, contrary to proton-pump inhibitor (PPI) use, which was associated with a decreased risk. 

There is controversy among SSRI-specific factors that may contribute to elevated GI bleeding risk, such as potency of serotonin reuptake or receptor affinity. In a 2012 cohort study, Castro et al⁴¹ identified 36,389 antidepressant users from an electronic medical record database in Massachusetts. The relative affinity of serotonin reuptake of each antidepressant was stratified by low-, moderate-, or high-affinity. The authors observed GI bleeding in 333 (n=14,927) in the low-affinity group vs GI bleeding in 601 (n=21,462) in the high-affinity group (adjusted RR, 1.17; 95% CI, 1.02-1.34). Similar findings were noted among patients who had a stroke. Additionally, a case-control study by Lewis et al³⁸ found increased odds of GI bleeding due to moderate- or high-affinity serotonin reuptake inhibitor use among 359 case participants and 1889 control participants (adjusted OR, 2.0; 95% CI, 1.4-3.0), but a dose-response relationship was not observed (P=.17). 

The association between SSRIs and GI bleeding has been identified by numerous studies and summarized by 3 large meta-analyses.^33,34,42 Inconsistencies may be attributed to enrollment criteria, potential confounders, and exposure definition. Selective serotonin reuptake inhibitors with a moderate to high affinity of serotonin reuptake could increase a person's risk for GI bleeding. However, there is limited evidence on whether this risk is dose-responsive. Concomitant use of NSAIDs, anticoagulants, or antiplatelet medications with SSRIs is associated with a significantly elevated risk of GI bleeding. Concomitant PPI use may be protective, but more evidence is needed. 

Observational studies^43-48 of intracranial bleeding are presented in Table 2. A 2000 study examined the United Kingdom–based General Practice Research Database with participants aged 18 to 79 years who received a first-time prescription for antidepressants from January 1, 1990, to October 31, 1997.⁴³ Antidepressants were classified by SSRIs or tricyclic antidepressants (TCAs), and monoamine oxidase inhibitors were excluded. From the database, 65 confirmed intracranial bleeding cases were identified and matched to 247 controls. This study reported an OR of 0.8 (95% CI, 0.3-2.3) for SSRIs, which implied no increased intracranial bleeding risk. Age, sex, antidepressant dose, and treatment duration were not significantly related to increased intracranial bleeding risk. However, smoking was found to be significantly related (OR, 3.7; 95% CI, 1.7-8.1). 

Smoller et al⁴⁵ described the risk of incident cardiovascular morbidity and mortality among community-dwelling postmenopausal women taking antidepressants. A higher RR of all-cause mortality (hazard ratio [HR], 1.32; 95% CI, 1.10-1.50) and increased stroke risk (HR, 1.45; 95% CI, 1.08-1.97) were associated with SSRIs. Increased all-cause mortality (HR, 1.67; 95% CI, 1.33-2.09) was associated with TCAs. Differences between SSRIs and TCAs were not found. A commentary⁴⁹ was later published stating that depression is an important and underrecognized risk factor for cardiovascular mortality, with a known lower quality of life, unhealthy life choices, and poor adherence to treatment regimens.⁴⁵ These factors can contribute to poor outcomes and leave physicians to make their own clinical judgements and monitor patients carefully. 

The latest population-based cohort study used the Clinical Practice Research Datalink that spanned over 650 general practices.⁴⁸ The number of intracranial bleeding cases found was 3036, which yielded an overall incidence rate of 3.8 per 10,000 persons per year. Use of SSRIs was associated with an increased intracranial bleeding risk (RR, 1.17; 95% CI, 1.02-1.35) relative to the TCAs; the highest risk was during the first 30 days of treatment (RR, 1.44; 95% CI, 1.04-1.99). Selective serotonin reuptake inhibitors classified as potent or strong 5HT inhibitors were also associated with an increased intracranial bleeding risk (RR, 1.25; 95% CI, 1.01-1.54). Use of SSRIs can place a small number of patients at increased intracranial bleeding risk, especially with higher 5HT potency agents during the first 30 days of therapy. 

Antiplatelet or anticoagulant medications are often prescribed for patients with stroke, and their concomitant use with SSRIs may further increase intracranial bleeding risk. In a study by Renoux et al,⁴⁸ concurrent anticoagulants were found to increase intracranial bleeding incidence (RR, 1.73; 95% CI, 0.89-3.39), but the increase was not statistically significant. The concurrent use of antiplatelet agents, however, did not increase intracranial bleeding risk (RR, 1.08; 95% CI, 0.65-1.34). The small number of patients with intracranial bleeding who used SSRIs and agents typically associated with bleeding (anticoagulants [n=291], antiplatelets [n=852], NSAIDs [n=708]) was acknowledged as a potential study limitation. Another study⁵⁰ reported an increased intracranial bleeding incidence when SSRIs were prescribed with anticoagulants compared with anticoagulants alone (RR, 1.56; 95% CI, 1.33-1.83). Patients should be carefully assessed when SSRIs and TCAs are used with anticoagulants, and prudent judgment should be used because of the routine prescription of antiplatelets and NSAIDs. 

Studies published from 1995 to 2005 reported a lack of association between SSRIs and intracranial bleeding, but later studies^43-48 reported a slight increase in likelihood (Table 2). A large population-based cohort study reported the incidence to be 3.8 per 10,000 persons per year (0.038%).⁴⁸ One feature that may influence study results is the number of controls used per case; Renoux et al⁴⁸ used a strict approach, with up to 30 controls matched to each case and a varying number of covariate factors. An association between 5HT potency and intracranial bleeding has been inconclusive. The concurrent use of SSRIs and anticoagulants or antiplatelets may increase intracranial bleeding risk, but the reported incidence is low, which makes definitive answers challenging. Overall, physicians should be cognizant of intracranial bleeding symptoms especially within the first 30 days of treatment. Patients should be evaluated emergently if the clinical presentation of intracranial bleeding is suspected. 

The combination of warfarin and SSRIs has been shown to increase bleeding risk. Proposed mechanisms behind this drug-drug interaction include altered platelet aggregation and CYP isoenzyme inhibition by SSRIs. Potency of CYP inhibition varies among SSRIs, though the effect of potency on bleeding risk is unclear. S-warfarin is the more potent isomer of warfarin that is primarily metabolized by CYP2C9. Priskorn et al⁵¹ evaluated the interactions between warfarin and a weak CYP2C9 inhibitor, citalopram. Participants (N=12) either received a single 25-mg dose of warfarin alone or on day 15 of a 21-day regimen of citalopram 40 mg/d. Blood samples were then analyzed during a 168-hour period after warfarin dosing. Citalopram did not result in any pharmacokinetic changes in warfarin. However, compared with warfarin alone, the addition of citalopram increased the maximum mean (SD) prothrombin time (PTT) from 25.1 (3.7) seconds to 26.7 (5.1) seconds, which represents an increase of 1.6 (3.0) seconds (90% CI, 1.01-1.10). Additionally, the area under the PTT-time curve (AUC_PT) increased by a mean (SD) of 5.0% (5.7%) after the addition of citalopram to warfarin (90% CI, 1.03-1.07). This study was powered to detect a 3% difference in AUC_PT using a 90% CI. Given the lack of pharmacokinetic changes, the authors suggested that citalopram affects anticoagulation by a mechanism outside the CYP isoenzyme system. The interaction resulting in a small change in PTT may not be clinically significant. 

Quinn et al⁵² evaluated 9186 participants from the Anticoagulation and Risk Factors in Atrial Fibrillation (ATRIA) trial who were exposed to warfarin during the median 6-year study period. Compared with warfarin users who did not use SSRIs, concomitant SSRI users (n=1743) spent a greater proportion of time at a supratherapeutic international normalized ratio (12.3%, P<.001) and had a higher mean bleeding risk score (2.99 for SSRI users vs 2.45 for nonusers; P<.001). A multivariable model adjustment for the bleeding risk score showed an elevated risk of major bleeding for warfarin patients taking SSRIs compared with nonusers. Warfarin users taking concomitant SSRIs may necessitate closer monitoring with regard to increased international normalized ratio levels and elevated risk for bleeding. 

Certain SSRIs may affect clopidogrel concentrations, resulting in decreased clopidogrel efficacy. Clopidogrel is a prodrug that undergoes a 2-stage activation process mediated by CYP450 enzymes, most notably CYP2C19. The SSRIs fluoxetine and fluvoxamine are potent inhibitors of CYP2C19. In a PK/PD study of rats, pretreatment with high-dose fluvoxamine (27 mg/kg) resulted in significant increases in AUC₀ and half-life of clopidogrel carboxylic acid (P<.05 for both).⁵³ The difference in mean (SD) platelet aggregation percentage before and after fluvoxamine (21.63% [6.05%] vs 45.98% [5.11%], respectively, P<.01) administration suggested a significant inhibition of clopidogrel's effects.⁵³ In a large population-based cohort study of CYP2C19-inhibiting SSRI use (n=9284) vs non-CYP2C19-inhibiting SSRI use (n=45,073), researchers found an increased risk of ischemic events (HR, 1.12; 95% CI, 1.01-1.24) in patients taking CYP2C19-inhibiting SSRIs.⁵⁴ The difference was more pronounced in older adults. This finding suggests that the co-administration of CYP2C19-inhibiting SSRIs such as fluoxetine and fluvoxamine may result in clinically significant reductions in clopidogrel efficacy. 

Both aspirin and SSRIs are associated with an independent increased risk of bleeding. The PK/PD interactions between SSRIs and aspirin are not well described. Serotonin was added to the platelets of 12 healthy participants in an ex vivo pharmacometabolomics study⁵⁵ of collagen-induced platelet aggregation. Before aspirin use, collagen-stimulated platelet aggregation was inhibited to a lesser extent among participants with higher serotonin levels compared with participants with lower serotonin levels (mean [SD], 61% [11%] vs 72% [8%], respectively, P=.02).⁵⁵ This study suggested that aspirin may further enhance serotonin's effects on platelets. Inhibition of serotonin reuptake into neurons and platelets is known to reduce platelet aggregation. Each patient's level of serotonin could potentially mediate the already additive effects of SSRIs and aspirin. 

Limitations inherent to this narrative review include limited data sources, possible selection bias, and no measures of heterogeneity among included studies. Although references cited by identified studies were evaluated if relevant, we did not use hand searching or include gray literature in this review. 

This review describes the rare but serious adverse effects associated with SSRI use: GI bleeding and intracranial bleeding. Mechanisms of SSRI-associated bleeding risk include increased gastric acid secretion and the inhibition of serotonin entrance into platelets. Concomitant use of warfarin, clopidogrel, and aspirin may further elevate a patient's risk of bleeding through various PK/PD interactions. Some studies suggest PPI use in patients taking these concomitant medications to mitigate GI bleeding risk. The application of OMT to associated areas of somatic dysfunction may also attenuate gastric acid secretion via modulation of the autonomic nervous system. However, more evidence is needed in this area. Regarding the risk of intracranial bleeding, prudent monitoring is highly recommended for patients taking anticoagulants or antiplatelets with or without SSRI use.