Patients with atrial fibrillation (AF), a heart rhythm disorder, are at increased risk of stroke. To prevent strokes, blood thinners (anticoagulants) are important. For years, warfarin was the only anticoagulant. Warfarin is influenced by many factors and therefore requires monitoring of the anticoagulant status, using frequent blood tests (INR-measurements). However, this changed when the non-vitamin K antagonist oral anticoagulants (NOACs) became available. NOACs have fewer interactions, and there is no need for INR-measurements. In large trials, NOACs show a similar effect in preventing strokes, with a reduced risk of bleeding. However, patients included in trials are often carefully selected. In particular this may be a concern for (frail) elderly patients that need to take multiple drugs for various other chronic conditions. Such patients were often not enrolled in the large trials. Therefore, we want to see if the difference in major bleeding when comparing NOACs to warfarin is modified by the number of other drugs someone uses, using data from daily clinical practice. Also, we want to see which variables could predict the occurrence of bleeding in NOAC patients, especially shortly prior to the bleeding. Finally, we want to compare the occurrence of bleeding and stroke in patients who inappropriately received a reduced NOAC dose according to guidelines, with patients with a correct dose.
With this retrospective cohort study, we want to investigate the influence of the number of concomitant drugs used, on the safety and efficacy of NOACs versus warfarin treatment. Patients diagnosed with AF and initiating a NOAC or warfarin between 2010 and 2018 will be included from the UK CPRD. The relative hazards for the different NOACs versus warfarin on the outcomes major bleeding, ischemic stroke and mortality will be quantified using a Cox Proportional Hazard Model. The influence of the number of concomitant drugs used on these outcomes is assessed by including this as an interaction term in the model. Furthermore, we will assess predictor variables for bleeding in patients treated with a NOAC, by validating, and if necessary, updating existing prediction models for bleeding risk. As the time period prior to the occurrence of bleeding is of our particular interest, we will investigate the association between changes in (modifiable) risk factors and bleeding, shortly before the bleeding occurs. Finally, we want to compare the occurrence of bleeding and stroke in patients with an off-label NOAC dose reduction, to patients with an on-label dosage, applying propensity scores to evaluate if the off-label dose reduction was indeed unjustified or not.
Major bleeding
- Stroke (both ischemic and haemorrhagic)
- Clinically relevant non-major bleeding
- Mortality
- Gastro-intestinal bleeding
- Hospitalization
Hendrika van den Ham - Chief Investigator - Utrecht University
Carline van den Dries - Corresponding Applicant - University Medical Centre Utrecht
Arno W Hoes - Collaborator - Maastricht University
Frans Rutten - Collaborator - University Medical Centre Utrecht
Geert-Jan Geersing - Collaborator - University Medical Centre Utrecht
Helga Gardarsdottir - Collaborator - Utrecht University
Karel Moons - Collaborator - University Medical Centre Utrecht
Patrick Souverein - Collaborator - Utrecht University
Romin Pajouheshnia - Collaborator - Utrecht University
Sander van Doorn - Collaborator - University Medical Centre Utrecht