Introduction
Current guidelines recommend non-vitamin‑K oral anticoagulants (NOACs) as the first-choice therapy for stroke prevention in patients with atrial fibrillation (AF) [
1‐
4]. The currently available NOACs have proved to be either comparable (dabigatran 110 mg, rivaroxaban and edoxaban) or superior (dabigatran 150 mg and apixaban) to warfarin in preventing stroke or systemic embolism and have demonstrated either a similar (dabigatran 150 mg and rivaroxaban) or superior (dabigatran 110 mg, apixaban and edoxaban) safety profile, with reductions in major and minor bleeding rates [
5‐
9]. All four agents have been associated with a statistically significant reduction in the rates of intracranial haemorrhage compared with warfarin. Despite these promising results, with the introduction of the NOACs in the Netherlands for stroke prevention in AF patients, questions were raised concerning the safety, efficacy and cost-effectiveness of these new drugs in daily practice [
10,
11]. The main argument was that the use of drugs in a clinical trial setting differs from that in real-world populations, as randomised controlled trials generally have stricter inclusion criteria and structured monitoring schemes with a shorter follow-up interval. When evaluating the safety and efficacy outcome events associated with NOACs, real-world data are important to accrue more heterogeneous patient populations with respect to co-morbidities and co-medication use. Therefore, the questions regarding the safety and efficacy of NOACs in patients with AF treated in real-world clinical settings in the Netherlands will remain until real-world data are presented. As required by the Dutch government, we referred all patients to our tertiary NOAC clinic with the purpose of monitoring the safety and efficacy of the introduction of NOACs.
The objectives of this study were to describe demographics, to evaluate the rates of bleeding, mortality and thromboembolic events, to evaluate the rates of reported adverse drug events, and to evaluate treatment persistence in patients with AF treated with NOACs in daily practice in a large tertiary hospital in the Netherlands with a tertiary NOAC clinic.
Methods
The St. Antonius Hospital NOAC registry is an observational, single-centre, prospective study in patients with AF treated with a NOAC for stroke prevention.
Because of the non-interventional character of the study the Ethical Committee (EC) confirmed that the Medical Research Involving Human Subjects Act (WMO) does not apply to this study and therefore official ethical approval of this study by the EC is not required. Declaration approval of the Institutional Board Review (LMTE) was obtained. A local study protocol is available. The study was conducted in accordance with the Declaration of Helsinki.
The NOAC clinic was initiated in 2013 with the purpose of monitoring the introduction of NOACs in our centre. All patients visited the outpatient clinic, where they consulted a physician with extensive knowledge about all NOACs and clinical trials. The patients were referred by either a cardiologist, neurologist, internal medicine doctor, or a general practitioner. During the consultation, the patient received information regarding AF and anticoagulation in general. Furthermore, patients received tailored education on and explanation about the use of a NOAC, the (modifiable) bleeding risks and the required laboratory assessments. Risk factors for bleeding were checked and reasons to reduce the NOAC dose were assessed. Also, all patients received an anticoagulation card [
3]. This card contains information regarding the anticoagulation therapy, as well as other drugs patients are using. Every patient received a direct telephone number and/or e‑mail address of a dedicated physician in order that that physician could be easily reached by a patient. When needed, a follow-up visit at the NOAC clinic was planned.
Follow-up was done through the regular outpatient clinic of the cardiology department, or by the general practitioner, whichever was appropriate. The general practitioner and/or referring physician received documentation about the initiation of NOAC therapy.
Study population and follow-up
We included all consecutive patients who visited or consulted the NOAC clinic. Because of the observational design, no exclusion criteria apply. Patients could be either (N)OAC naïve or (N)OAC experienced. Decisions with regard to NOAC choice, continuation or changes were at the discretion of the treating physician.
Baseline characteristics at the NOAC start date were collected. Follow-up data were collected by reviewing electronic patient files and follow-up was performed from the start of NOAC treatment until the last (outpatient) visit in our hospital or until a patient died, with a final follow-up date of 24 May 2018. No study-specific visits were performed. We did not assess medication adherence. Mortality was checked at the Dutch civil registry.
Study outcomes
The primary outcome was the occurrence of any in-hospital bleeding or bleeding reported at the outpatient NOAC clinic or regular outpatient cardiology clinic during long-term follow-up. The co-primary outcomes were the rates of stroke/transient ischaemic attack (TIA) and mortality at long-term follow-up. Secondary outcomes were the rates of pulmonary embolism, deep venous thrombosis, myocardial infarction, NOAC therapy discontinuation and adverse drug reactions. Furthermore, we collected the reasons for NOAC therapy discontinuation; whether and to which drug patients switched; the number of patients who changed to another type or dose of NOAC. The predefined definitions of endpoints are listed in the Electronic Supplementary Material (Appendix 1).
Additionally, we performed an on-treatment event analysis. In all patients who permanently discontinued their NOAC therapy, follow-up was censored 3 days after discontinuation. Thus, only events that occurred during or within 3 days after discontinuation of treatment were evaluated.
Statistical analysis
Descriptive analysis of the data was performed using summary statistics for categorical and quantitative (continuous) data. Continuous data are reported as means with standard deviation or medians with interquartile range. Categorical data are expressed as percentages. Distributions of categorical data were examined by the Χ2-test or Fisher’s exact test, as appropriate. Continuous data were compared using Student’s t-test or the Mann-Whitney U test, as appropriate. The analyses were performed using SPSS software for Windows, version 24 (IBM Corporation, Armonk, NY, USA).
Discussion
In this real-world registry of AF patients treated with NOACs we provide a complete description of all patients that visited the NOAC clinic in a high-volume centre in the Netherlands. We also provide the adverse event rates and the treatment details.
This tertiary NOAC clinic was launched with the purpose of monitoring the safety and efficacy of the use of the recently introduced anticoagulation drugs. We extensively provided support and education to every patient starting a NOAC and coordinated their integrated care. We aimed to waive the concerns of non-adherence by comprehensive education and intense monitoring.
We compared our data with available data from the four pivotal randomised control trials (RCTs) of NOACs in AF: RE-LY, ROCKET-AF, ARISTOTLE and ENGAGE AF-TIMI 48 [
5‐
8]. Mean age and proportion of male patients were similar. The baseline stroke risk of patients in this NOAC registry is somewhat lower than that of pivotal NOAC randomised trials with a mean CHADS
2 score ranging from 2.1 to 3.5. The number of patients having experienced myocardial infarction was comparable. On the other hand, in our cohort the numbers of patients with a history of prior stroke, diabetes mellitus, hypertension and heart failure are noticeably lower.
In our registry, the any bleeding rate per 100 patient-years was 6.0 for the total population. This rate was found to be lower than in the pivotal randomised trials and available real-world data. The Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation (ARISTOTLE) study found an annual incidence of any bleeding of 18.1% for apixaban [
7]. Furthermore, the Effective Anticoagulation with Factor Xa Next Generation in Atrial Fibrillation—Thrombolysis in Myocardial Infarction 48 (ENGAGE AF-TIMI 48) trial found an annual incidence of any overt bleeding of 14.15% for high-dose edoxaban [
8]. The Rivaroxaban Once Daily Oral Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation (ROCKET-AF) and Randomised Evaluation of Long-Term Anticoagulant Therapy (RE-LY) trials did not provide incidences of any bleeding [
5,
6]. However, the rate of major bleeding per 100 patient-years in the present registry was 1.8, and this rate is in the range of the phase III trial data. Namely, RE-LY reports annual incidences of 2.71% and 3.11% for dabigatran 110 mg b.i. d. and dabigatran 150 mg b.i. d. respectively, ARISTOTLE reports an annual incidence of 0.96%, ENGAGE an annual incidence of 2.75% and ROCKET-AF 3.6 major bleedings per 100 patient-years [
5‐
8]. Major bleeding rates of 2.1–3.0 per 100 patient-years in patients using rivaroxaban and 2.8 per 100 patient-years in patients using apixaban were reported in the Dresden NOAC registry and the Xarelto for Prevention of Stroke in Patients with Atrial Fibrillation (XANTUS) registry [
12‐
15]. A possible explanation for the relatively low any bleeding rate in our study might be the observational character and the absence of obligatory study-specific visits. This could have caused underreporting of any bleeding, mainly minor bleeding. Furthermore, in our cohort ASA was discontinued in all patients after initiation of NOAC therapy as opposed to the pivotal NOAC trials where this was left to the discretion of the physician. This difference might be another explanation for the lower any bleeding rate in our study.
The presented rate of stroke in our registry is consistent with the data of the pivotal NOAC trials. We report an incidence of 1.2 strokes per 100 patient-years, whereas the RCTs report rates ranging from 0.92% to 1.49% per year. Conversely, all-cause mortality in our registry was 6.4 patients per 100 patient-years, whereas the RCTs report all-cause mortality ranging from 1.9% to 4.0% per year. The XANTUS registry reports incidences of 0.7 strokes and 1.9 deaths per 100 patient-years, respectively. In the Dresden NOAC registry incidence of stroke and mortality is not reported. Korenstra et al. reported a rate of 0.6 strokes per 100 patient-years and a mortality rate of 2.0% per year [
16].
Drug persistence is a major concern in stroke prevention because anticoagulant discontinuation potentially leaves patients unprotected against the risk of stroke. In our study, we showed that 16.5% of patients permanently discontinued their NOAC therapy, 8.8% being cases of premature discontinuation. In the pivotal NOAC trials, permanent discontinuation ranged between 21% and 25% with a mean follow-up of 1.8–2 years. Furthermore, in the Dresden NOAC registry, 223 patients (18.5%) permanently discontinued rivaroxaban during follow-up (554 days) [
17]. These figures are in accordance with our results. Unfortunately, we did not register adherence to the prescribed NOAC therapy.
Our study is subject to limitations. First, some events may have been underestimated. Because of the observational nature of the study, some thrombotic and bleeding events might not be reported. In contrast, the mortality data were not underestimated, since mortality was checked at the Dutch civil registry. Another limitation is that we did not study medication adherence in this cohort. Also, there might be selection bias and confounding in this study because of its observational nature, as always with real-world evidence.
One of the strengths of this study is the completeness of the data. Since this registry is not a claim registry, we did not depend on the completeness and quality of the diagnosis coding in a claim database. For example, information on whether the proper NOAC dose was prescribed based on clinical and laboratory data is not frequently recorded in claims databases. Finally, we included patients who were using all four available NOACs. Other available registries in the Netherlands only reported on the use of a single NOAC [
16,
18].
In conclusion, in this real-world registry of AF patients using NOACs, we showed that NOACs are safe and efficacious. The observed rates of major bleeding and stroke were similar, the all-cause mortality was higher, and the any bleeding and discontinuation rates were lower than those in the randomised trials. This study, including a real-world cohort of consecutive patients visiting a tertiary NOAC clinic, confirms the results of the RCTs in everyday clinical practice.