Warfarin and Direct Oral Anticoagulants (DOACs) are used to prevent or treat blood clots. Bleeding is the most serious complication of using anticoagulant medication. Several antibiotics are thought to interact with anticoagulants and increase the chance of a major bleed. However, the available research does not clearly identify which antibiotics are safe and prescribing advice is mostly based on anecdote. It is important to know which antibiotics increase the risk of bleeding to enable healthcare professionals to avoid potentially harmful antibiotics but appropriately prescribe those that are safe.
This project will determine the risk of major bleeding associated with antibiotic treatment in anticoagulant medication users by analysing anonymised healthcare data from the Clinical Practice Research Datalink (CPRD). We will identify anticoagulant medication users to address three key objectives. Objective 1 will determine which of the 10 most commonly prescribed antibiotics in the UK are associated with an increased risk of major bleeding and will answer the broad question of which antibiotics are safe to prescribe to anticoagulant medication users. Objective 2 will compare groups of antibiotics recommended for the same infection and will answer the more specific question of which antibiotic is safe to prescribe to an anticoagulant medication user with a respiratory, urinary, or skin infection. Objective 3 will determine the risk of bleeding in people with an infection that did not need antibiotic treatment and will answer the question of whether infection alone is associated with the risk of bleeding.
Background
Depending on the specific drug, 2-4% of anticoagulant medication users have a major bleed needing hospitalisation, of whom 20% die within 30 days. Several studies have implicated antibiotic-anticoagulant drug-drug interactions as the likely cause of a significant number of bleeds in anticoagulant medication users, but previous estimates of the risk of bleeding associated with specific antibiotics are inconsistent and do not fully account for the risk related to the underlying infection.
Aim
This project aims to determine the risk of major bleeding associated with infection and antibiotic treatment in anticoagulant medication users and effectively disseminate the findings to improve care and help reduce the incidence of these adverse outcomes.
Methods
We will address the research aim through a series of epidemiological studies using prescribing, hospital admission, and mortality data from the Clinical Practice Research Datalink (CPRD). We will identify a cohort of new anticoagulant medication users in the CPRD (incident user design). Associations between treatment with different antibiotics and major bleeding will be determined in two ways. First, a retrospective cohort study using logistic regression models to estimate odds of bleeding associated with the 10 most commonly prescribed antibiotics in the UK, using amoxicillin as an active comparator. Second, a propensity score matched cohort study estimating the odds of bleeding for antibiotics recommended for the same indication. A third study will use a self-controlled case series design to determine the impact of infection without antibiotic treatment on bleeding risk. Estimates for warfarin and Direct Oral Anticoagulants will be reported separately. Outcomes will include death or hospitalisation from gastrointestinal or intracranial bleeding and clinically relevant non-major bleeding. Several sensitivity analyses will test the robustness of the findings using methods such as negative control exposures and negative control outcomes.
The primary outcome is major bleeding.
Secondary outcome is clinically relevant non-major bleeding (CRNMB).
To estimate the risk of major bleeding, we will use linked hospital and death registry data to ascertain the following:
1. Hospitalisation with Intracranial bleeding (ICD-10 codes I6xx, S065 and S066)
2. Hospitalisation with Gastrointestinal bleeding (ICD-10 codes I850, K2xx, K3xx, K625, K9xx)
3. Death from either of the above.
To estimate the risk of CRNMB, we will use GP and hospital data to ascertain a clinical presentation with less serious bleeds - haemoptysis, epistaxis, and haematuria
Harry Ahmed - Chief Investigator - Cardiff University
Harry Ahmed - Corresponding Applicant - Cardiff University
Daniel Farewell - Collaborator - Cardiff University
Heather Whitaker - Collaborator - Public Health England
Hywel M. Jones - Collaborator - Cardiff University
Julia Hippisley-Cox - Collaborator - University of Oxford
Simon Noble - Collaborator - Cardiff University
HES Admitted Patient Care;ONS Death Registration Data;Patient Level Index of Multiple Deprivation