3B-03 - Risk of hemorrhage in patients on oral anticoagulants due to pharmacodynamic drug interactions: a case-crossover analysis

Background: Patients treated with oral anticoagulants (OAC) are at risk of bleeding. These patients may also be on treatment with other drugs that could increase the risk of bleeding due to a pharmacodynamic interaction.

Objectives: To determine the risk of bleeding for patients on OAC when exposed to pharmacodynamic drug interactions.

Methods: A cohort of patients on OAC were identified in the Data Analytics Program for Health Research and Innovation (PADRIS). PADRIS is a database that uses cross-references of health data (such as prescriptions, X-rays, reports and medical visits) generated by the comprehensive public health system of Catalonia, Spain. Diagnoses are coded with the ICD-10 and drugs with the WHO ATC classification. Data are previously anonymized and depersonalized.
Patients experiencing a gastrointestinal hemorrhage (GIH) or intracranial one (ICH) were selected to conduct a case-crossover analysis. From the index date (hemorrhagic event) windows of exposure were the hazard window (≤30 days before) and the not at-risk window (comprising 30 days of follow-up occurring before the hazard window, separated by a 60-day washout period). During these windows patients could be exposed to only OAC or to a pharmacodynamic interaction (OAC + comedications of interest: non-steroidal anti-inflammatory drugs [NSAID], antiplatelets, corticosteroids, and/or antidepressants). Conditional logistic regression was used to estimate odds ratios (ORs) for the association between exposure to interacting drugs during the hazard versus referent windows, adjusted for presence or absence of proton pump inhibitors at each time window.

Results: There was a total of 2,422 patients on OAC experiencing a hemorrhagic event. They had a mean age of 80.7 (SD 8.2) years, half were females (N= 1,172. 48.4%), 35.3% (N= 854) obese and 56.1% (N= 1,359) had an estimated filtration glomerular rate < 60 at the index date (hemorrhage event). Mean OAC exposure was 2.5 years (IQR 12-4.2). More than half of the hemorrhages were GIH (N= 1,461, 60.3%). Antidepressants did not show risk of hemorrhage (OR 1.01, 95% CI 0.73-1.40) while antiplatelets (OR 1.95, 95% CI 1.30-2.93), corticosteroids (OR 1.92, 95% CI 1.32-2.79), and NSAID did (2.92, 95% CI 1.94-4.42). For females antiplatelets and corticosteroids did not show any risk and NSAID showed risk for GIH (OR 3.11, 95% CI 1.89-5.10) but not for ICH (OR 1.63, 95% CI 0.49-5.44).

Conclusions: There is a three-fold increase risk for bleeding when patients are concomitantly exposed to OAC and NSAID, and a two-fold when on OAC and corticosteroids or antiplatelets, with no risk with antidepressants. Differences between females and males exist in the risk of bleeding when exposed to pharmacodynamic interactions.