Thromboembolic and bleeding events with rivaroxaban in clinical practice in Spain: impact of inappropriate doses (the EMIR study)

EMIR (Estudio observacional para la identificación de los factores de riesgo asociados a eventos cardiovasculares mayores en pacientes con fibrilación auricular no valvular tratados con un anticoagulante oral directo [Rivaroxaban] [“Observational study to identify risk factors associated with major cardiovascular events in patients with NVAF treated with a DOAC [rivaroxaban]”) was a postauthorization, observational and multicenter study aimed at ascertaining the predictors of major cardiovascular events in NVAF patients treated with rivaroxaban for at least 6 months by different specialists in Spain. Patients were followed up for 2.5 years. In this study, the appropriateness of the dosage of rivaroxaban and the related outcomes (thromboembolic events and major bleeding) were analyzed after 2 years of follow-up. Overdosage was defined as creatinine clearance (Cockcroft–Gault) <50 ml/min with rivaroxaban 20 mg and underdosage as creatinine clearance ≥50 ml/min with rivaroxaban 15 mg.

The study population comprised NVAF patients aged ≥18 years of both sexes who had been treated with rivaroxaban for at least 6 months before being enrolled. The study population excluded patients participating in a clinical trial, patients starting treatment with rivaroxaban after the inclusion period and patients with severe cognitive impairment, chronic infectious disease, systemic autoimmune disease, active cancer or significant liver insufficiency. All patients gave their written consent prior to enrollment. The study was approved by each participating Institutional Review Board.

Baseline data were recorded using a specific electronic case report form and included biodemographic data (age, sex, level of dependency, type of atrial fibrillation [AF]), physical examination (weight, BMI, heart rate, blood pressure), thromboembolic data (CHADS₂ [congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, prior stroke or TIA or thromboembolism (doubled)] and CHA₂DS₂-VASc [congestive heart failure, hypertension, age ≥75 years [doubled], diabetes mellitus, prior stroke or TIA or thromboembolism [doubled], vascular disease, age 65–74 years, sex category]), bleeding data (HAS-BLED [hypertension, abnormal renal/liver function, stroke, bleeding history or predisposition, labile INR [international normalized ratio], elderly, drugs/alcohol]), cardiovascular risk factors (hypertension, dyslipidemia, diabetes, smoking status), concomitant structural or vascular disease (heart failure, ischemic heart disease, renal insufficiency, prior cerebrovascular disease, peripheral artery disease, aortic plaque, venous thromboembolic disease and prior systemic embolism). Conditions that increase the risk of bleeding (e.g., labile INR in patients taking VKAs before starting rivaroxaban, alcohol intake, falls) were also recorded. Dependency was classified as autonomous (no dependency), partial dependency to daily activities or complete dependency to daily activities.

Previous antithrombotic treatment before starting rivaroxaban therapy, current dose of rivaroxaban, and the appropriateness of this dose (correct dose, overdosage or underdosage) were analyzed. For the purpose of this analysis, overdosage was considered to have occurred when a patient with a Cockcroft–Gault creatinine clearance <50 ml/min was taking rivaroxaban 20 mg and underdosage when a patient with a Cockcroft–Gault creatinine clearance ≥50 ml/min was taking rivaroxaban 15 mg. Factors associated with prescribing inadequate doses of rivaroxaban, underdosage and overdosage were also analyzed.

Thromboembolic events (stroke, transient ischemic attack, systemic embolism or myocardial infarction), and major bleeding (following International Society of Thrombosis and Haemostasis definition) [16]. In addition, during this period, all-cause death and cardiovascular death that included sudden cardiac death and death from heart failure, myocardial infarction, arrhythmia or percutaneous or surgical coronary revascularization were analyzed. Outcomes were analyzed according to the appropriateness of the rivaroxaban dosage.

A total of 1503 patients were initially enrolled in the study. After excluding 82 patients because of lack of data or inconsistent data, 1421 patients were finally analyzed.

Table 1 shows the baseline clinical characteristics of the study population. Mean age was 74.2 ± 9.7 years, 55.5% of patients were men, and total or partial dependency was reported in 9.9% of patients. The mean CHA₂DS₂-VASc score was 3.5 ± 1.6 and the mean HAS-BLED score was 1.6 ± 1.0. Major comorbidities were common, such as heart failure in 22.7%, ischemic heart disease in 16.5%, renal insufficiency in 15.8% and prior cerebrovascular disease in 12.5%.

A total of 1096 patients (77.1%) were taking rivaroxaban 20 mg and 325 (22.9%) were taking rivaroxaban 15 mg. At baseline, 238 patients (16.7%) were taking an inadequate dose of rivaroxaban, namely, underdosage in 138 (9.7%) and overdosage in 100 (7.0%), according to the Cockcroft–Gault equation. If estimated glomerular filtration rate is taken into consideration, the percentage of patients underdosed and overdosed would have been, respectively, 14.8 and 3.7% according to the MDRD-4 formula and 12.1 and 5.4% according to CKD-EPI. After 2 years of follow-up, mean creatinine clearance (Cockcroft–Gault) increased slightly from 76.0 ± 30.5 ml/min at baseline to 77.0 ± 33.8 ml/min (p = 0.014). The number of overdosed patients decreased from 100 (7.0%) to 47 (3.3%) during the study period.

At baseline, compared with patients taking the proper dose of rivaroxaban, underdosed patients were older (78.4 ± 8.7 vs 73.0 ± 9.6 years; p < 0.001) and less autonomous (77.9 vs 91.9%; p < 0.001) and had a higher thromboembolic risk (CHA₂DS₂-VASc score: 4.0 ± 1.7 vs 3.4 ± 1.5; p < 0.001), a higher bleeding risk (HAS-BLED score: 2.0 ± 1.0 vs 1.5 ± 1.0; p < 0.001) and more frequent renal insufficiency (31.9 vs 12.9%; p < 0.001). Compared with patients taking the proper dose of rivaroxaban, overdosed patients were older (82.3 ± 5.6 vs 73.0 ± 9.6 years; <0.001) and less autonomous (82.7 vs 91.9%; p = 0.003) and had lower weight (67.4 ± 9.7 vs 80.7 ± 15.9 Kg; p < 0.001), a higher thromboembolic risk (CHA₂DS₂-VASc score: 4.2 ± 1.3 vs 3.4 ± 1.5; p < 0.001), and a higher bleeding risk (HAS-BLED score: 1.9 ± 0.9 vs 1.5 ± 1.0; p < 0.001) and more frequent renal insufficiency (28.0 vs 12.9%; p < 0.001). However, diabetes was less frequent in this group (16.0 vs 27.5%; p = 0.013; Table 1).

Factors associated with the prescription of an inadequate dose of rivaroxaban are reported in Table 2. In the bivariate analysis, older age, female sex, renal insufficiency, prior cerebrovascular disease and partial or total dependency were associated with inappropriate dosage. Of note, labile INR did not have any impact on dosage. However, in the multivariate analysis, the only remaining independent factors were age (OR 1.09; 95% CI 1.07–1.11; p < 0.001) and renal insufficiency (OR 2.00; 95% CI 1.41–2.84; p < 0.001). Variables associated with underdosage and overdosage are reported in Table 3. In the bivariate analysis, older age, permanent AF (versus paroxysmal AF), dependency (versus autonomous) and anemia were associated with underdosage. However, in the multivariate analysis, the only remaining independent factors were age (OR 1.04; 95% CI 1.02–1.07; p < 0.001) and dependency (OR 2.13; 95% CI 1.43–3.18; p = 0.003). In the bivariate analysis, old age, female sex, absence of Type 2 diabetes, dependency and falls in the previous year were associated with overdosage. However, the only remaining independent factors were age (OR 1.13; 95% CI 1.10–1.16; p < 0.001) and absence of Type 2 diabetes (OR 0.47; 95% CI 0.26–0.85; p = 0.013).

Clinical outcomes after 2 years of follow-up are presented in Table 4. Cumulative time was 2537.56 years. Annual rates for death, thromboembolic events (stroke, transient ischemic attack, systemic embolism or myocardial infarction), and major bleeding were 2.68, 0.71 and 0.95 events per 100 patient-years, respectively in total population. The equivalent values for underdosed and overdosed patients were 5.76, 0.89 and 1.33 events per 100 patient-years and 3.43, 1.14 and 2.29, respectively. There was a trend toward higher all-cause mortality among underdosed patients (unadjusted HR 2.51; 95% CI 1.36–4.63; adjusted HR 1.39; 95% CI 0.75–2.58; Supplementary Table 1). No significant differences were found in cardiovascular mortality among underdosed patients (0.44 vs 0.75 events per 100 patient-years; p = 0.99). More bleedings were recorded in overdosed patients, although the difference was not significant (2.29 vs 0.80 events per 100 patient-years; p = 0.14; Table 4).

The main findings of this study are as follows: approximately 17% of patients treated with rivaroxaban for NVAF received off-label doses, more commonly the lower dose; advanced age and high dependency explained most of the underdosing prescriptions; after a 2-year follow-up period, rates of thromboembolic or bleeding complications were lower than expected based on data from other prospective real-world registries.

The ROCKET-AF trial, which was performed in a population with a high thromboembolic risk, showed that rivaroxaban was at least as effective as warfarin for the prevention of stroke or systemic embolism with the same risk of major bleeding, but with a lower risk of fatal and intracranial bleeding [17]. Compared with the rivaroxaban arm of the ROCKET-AF trial, both the number of comorbidities and the thromboembolic risk were lower (CHADS₂ score 3.5 vs 2.0), even though the patients in our study were slightly older [17]. The XANTUS study was the first real-world, prospective and observational study of patients treated with rivaroxaban for prevention of stroke in AF, with a thromboembolic risk that was similar to that reported in our study [18]. Other studies of NVAF patients treated with rivaroxaban in Spain have revealed a similar clinical profile [19–22]. Therefore, these data suggest that the EMIR population is highly representative.

In the EMIR study, the proportion of patients taking 20 and 15 mg of rivaroxaban (77 and 23%, respectively) was similar to that of the ROCKET-AF trial (20 mg dose: 79.3%) and the XANTUS real-world study (79 and 21%, respectively) [17,18]. We found that approximately 83% of patients were taking the correct dose of rivaroxaban according to approved recommendations, 10% of patients were underdosed and 7% were overdosed. Remarkably, during the study, the proportion of overdosed patients decreased to only 3%. As a result, our data showed that in most cases, rivaroxaban is adequately prescribed in clinical practice, and that only a small proportion of patients are underdosed. However, there were relevant differences in the numbers according to the method used to estimate renal function and this could have an impact on the appropriateness of dosage in clinical practice.

Various studies have analyzed the frequency of inappropriate dosing of rivaroxaban. In the XANTUS study, 15% of patients with a creatinine clearance ≥50 ml/min received rivaroxaban 15 mg, whereas 36% of patients with a creatinine clearance <50 ml/min received rivaroxaban 20 mg [18]. In a preplanned pooled analysis of the XANTUS, XANAP (Asia) and XANTUS-EL (Latin America and EMEA Region) registries, 18.3% of patients with a creatinine clearance ≥50 ml/min received rivaroxaban 15 mg [23].

Although other authors have found that a higher risk of bleeding may be associated with the use of low doses of DOACs [9], in our study, advanced age and dependency, but not bleeding risk, were associated with underdosage. In fact, we found thromboembolic and bleeding risk to be similar in underdosed and overdosed patients. Many authors report that underdosage is more common in elderly patients, particularly in those with creatinine clearance consistent with the dose reduction criteria [7,24,25]. Of note, some studies, but not all, have shown that dependency and frailty, that are commonly observed in elderly patients, may be associated with underdosage [26,27]. However, dosage of DOACs should be performed according to the summary of product characteristics [2]. This is very relevant, since some studies have shown that inadequate prescription may be associated with worse outcomes and that this is more evident when adjustment of the DOACs prescribed is more complex [10–12,15,26,28,29]. Thus, recent data from the GARFIELD-AF registry have shown a higher risk of all-cause mortality – mainly cardiovascular – for underdosing [29]. We recorded numerically higher rates of death and thromboembolic and bleeding outcomes in underdosed patients, although these numbers did not reach statistical significance, likely due to the low number of events during follow-up. On the other hand, frailty is frequent in patients with AF and is associated with adverse clinical outcomes, partially due to lower rates of adequate anticoagulation in this population [30,31]. As a result, the dosage of rivaroxaban should also be based on renal function in this population.

Although less common, overdosage is also recorded with rivaroxaban. In our study, the proportion of overdosed patients decreased from 7 to 3% over time. Due to pharmacovigilance safety reasons, the investigators were advised in case of inappropriate dosage according to the summary of product characteristics (Cockcroft–Gault creatinine clearance). Then, the investigators could modify the dosage according to their clinical criteria, but this only occurred in case of overdosing during the second year of follow-up. While study results vary, most are similar to those we report [19,21,32–34]. Of note, after 2 years of treatment with rivaroxaban, renal function remained stable. Compared with patients prescribed warfarin, patients treated with DOACs may experience a less pronounced decline in renal function [35], thus indicating an added value in the management of patients with NVAF compared with VKAs [36].

After 2 years of follow-up, rates for death, thromboembolic events and major bleeding were 2.68, 0.71 and 0.95 events per 100 patient-years, respectively. The equivalent values were 1.9, 1.7 and 3.6 per 100 patient-years, respectively, in the ROCKET-AF trial (rivaroxaban-arm) and 1.9, 1.8 and 2.1 per 100 patient-years, respectively in the XANTUS study [17,18]. Therefore, in clinical practice, thromboembolic and bleeding events are less frequent than in the pivotal clinical trial, likely because of differences in the patients’ clinical profile. Of note, bleeding outcomes occur more frequently during the first weeks after initiating treatment, and in the EMIR study, patients were on continuous rivaroxaban therapy, thus potentially leading these events to be underrepresented.

In addition, real-life studies have shown lower rates of thromboembolic complications of rivaroxaban compared with warfarin [14,37]. These data strongly suggest that rivaroxaban can be safely used in clinical practice.

Our study confirms that dosing of rivaroxaban for prevention of thromboembolic events in patients with NVAF is appropriate in most patients in current clinical practice in Spain. Older age and renal insufficiency are the main predictors of inadequate prescription. After a 2-year follow-up period, rates of death and thromboembolic and major bleeding events are low. There was a trend toward a higher risk of death among underdosed patients. As a result, adequate prescription of rivaroxaban according to renal function should be strongly encouraged.

M Sanmartín Fernandez has received speaker and advisory fees from the following companies in the past 3 years: Bayer, Boehringer-Ingelheim, BMS and Pfizer. F Marín has received consultancy/lecturing fees from Bayer, Boehringer Ingelheim, Pfizer, Bristol Myers Squibb and Daiichi Sankyo. C Rafols is an employee of the Medical Department, Bayer Hispania SL. Bayer Hispania SL is the sponsor of the EMIR study. F Arribas has received consultancy/lecture fees and research funding from Daiichi Sankyo, Pulse Dynamics, Medtronic, Boston Scientific, Bayer, Bristol Myers Squibb, Arrhythmia Network Technology SL/BAROSTIM, Abbott, Novartis, Janssen-Cilag, Biosensors, Edwards Lifesciences. He has also received financial compensation from Bayer for participating in the EMIR registry. V Barrios has received consultancy/lecture fees from Bayer, BMS/Pfizer, Boehringer Ingelheim and Daiichi Sankyo. He has also received financial compensation from Bayer for participating in the EMIR registry. J Cosín-Sales has received financial compensation from Bayer for participating in the EMIR registry. M Anguita Sánchez has received financial compensation from Bayer for participating in the EMIR registry. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Editorial and writing assistance was provided by Content Ed Net (Madrid, Spain) with funding from Bayer Hispania.

The Supplementary Material details the names of the EMIR Study Investigators.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

The study was approved by each participating Institutional Review Board. All patients gave their written consent prior to enrollment.