Treating epilepsy with anticoagulants?

Introduction
Atrial fibrillation (AF) is associated with significant morbidity, mortality and increased healthcare costs. The main complication of AF is stroke. As oral anticoagulants markedly reduce the risk of stroke, they are a central pillar of the guideline-based therapy for AF. Routine application of this therapeutic strategy has become easier with direct oral anticoagulants (DOACs), the use of which is steadily increasing while vitamin K antagonists (VKA) tend to become less important. Clinical trials and post-marketing surveillance programmes have indicated that, overall, DOACs have a comparable efficacy to VKAs in preventing thromboembolic events but cause less intracranial haemorrhages.1,2

A caveat, however, relates to the fact that studies on oral anticoagulation in AF have focused on acute ischaemic stroke or haemorrhage as outcome parameter but have neglected covert brain infarction, also known as ‘silent’ stroke. With covert brain infarction, imaging studies show lesions typical for ischaemic stroke but patients do not have a clinical stroke history.3 A meta-analysis reported covert brain infarction as detected by magnetic resonance imaging in 40% of AF patients.4 These numbers are worrisome as covert brain infarction is associated with dementia and epileptic seizures.5,6 Brain infarction is known to be the most frequent cause of acquired epilepsy in elderlies,7 with 8% of stroke patients developing epilepsy over 5 years after the ischaemic insult.8 Thus, covert brain infarction may be one potential explanation why AF is associated with a higher epilepsy rate.9,10 However, until now, it is still unclear whether anticoagulant drugs would reduce covert brain infarction or its clinical manifestations.11,12

In a previous pharmacovigilance study, we applied methods for signal detection in claims data to explore potential side effects of DOAC) treatment.13 The aim of the current study was to further examine a signal suggesting an association between DOAC dispensations and international classification of diseases (ICD) codes of epilepsy or seizures using a study design that specifically addresses the research question and minimizes potential sources of bias (e.g. temporal relationship between covariates and exposure, confounding). Following patients with a defined indication for anticoagulation from the time point of treatment initiation, we applied an active comparator design to assess the risk of epilepsy from DOACs vs. VKA. In addition, we aimed to explore potential underlying mechanisms of the observed association between DOACs and epilepsy.

Methods
Data source
This study was performed on data from the German Pharmacoepidemiological Research Database (GePaRD). The GePaRD is based on claims data from four statutory health insurance providers in Germany and currently includes more than 25 million subjects who have been insured with one of the participating providers since 2004 or later. In addition to demographic data, GePaRD contains information on all reimbursable drug dispensations and all reimbursable outpatient (i.e. from general practitioners and specialists) and inpatient services and diagnoses. Per data year, there is information on ∼20% of the general population covering all geographical regions of Germany. The GePaRD has been used for various drug utilization and safety studies, including studies on oral anticoagulants.14 All involved health insurance providers as well as the German Federal Office for Social Security and the Senator for Health, Women and Consumer Protection in Bremen as their responsible authorities approved the use of GePaRD data for this study according to current legal conditions. According to the Ethics Committee of the University of Bremen, studies based on GePaRD are exempt from institutional review board review.

Study design and cohort of patients with atrial fibrillation
The study was designed as an active comparator case-control study nested in a cohort of patients with AF initiating DOAC or phenprocoumon (PPC) treatment after a period of 365 days with no use of the respective anticoagulant drug (‘new users’). The PPC is the VKA predominantly prescribed in Germany. The study period started on 1 January 2011 and ended on 31 December 2017. To be included into the study, individuals had to have (i) at least one dispensation of one of the index drugs (PPC, rivaroxaban, apixaban, dabigatran, and edoxaban) during the study period with no dispensation of any index drug in the 365 days before cohort entry, (ii) at least 365 days of continuous active insurance before the cohort entry, and (iii) at least one outpatient or hospital diagnosis of atrial fibrillation (ICD-10-GM I48, including sub-codes) in the 365 days before cohort entry. Patients with a diagnosis of epilepsy/seizures or cancer at any time before cohort entry and women who were pregnant at cohort entry were excluded (Figure 1).



Flowchart of the study cohort according to the defined inclusion and exclusion criteria. The diagnosis of epilepsy at any time before cohort entry (exclusion criteria) was based on the International Classification of Diseases, 10th revision, German modification (ICD-10-GM) codes listed in Supplementary material online, Table S7.

Incident cases of epilepsy or seizures were identified by searching for the first ICD-10-GM codes G40 (epilepsy, including sub-codes) or R56.8 (other and unspecified convulsions) as either outpatient diagnosis (classified as certain) or main hospital diagnosis. We defined the index date as the date of diagnosis (outpatient diagnoses) or the admission date (hospital diagnoses).

Up to 10 controls were matched to each case with epilepsy/seizures by sex, age at index date (±1 year), and statutory health insurance provider, using risk set sampling (with replacement) with time in cohort as the time axis. Eligible patients hospitalized for any reason at the index date of the case were excluded from the set of potential controls. Cases were eligible to be selected as a control before their index day.15 We categorized patients as current users if the supply of the respective drug overlapped the index date. The supply of index drugs was estimated by dividing the dispensed amount of active ingredient by the defined daily dose of the respective index drug.

Based on their potential to confound the association between oral anticoagulation and epilepsy or seizures, we selected numerous variables, including components of the CHA2DS2-VASc risk score for stroke, the HAS-BLED score for bleeding, additional risk factors of stroke and epilepsy, and data on other chronic diseases. Confounder variables were obtained from in- and outpatient diagnoses and procedures. See Supplementary material online, Table S1 for a detailed list of all potential confounders.