Rivaroxaban, a factor Xa inhibitor, is a new oral anticoagulant that has been developed as an alternative to vitamin K antagonists. However, its safety remains unclear. Reported randomized controlled trials comparing the safety of rivaroxaban with that of vitamin K antagonists (warfarin, acenocoumarol, phenprocoumon, and fluindione) were systematically searched. Inclusion was restricted to studies of ≥30 days’ treatment duration. Safety end points examined included major and clinically relevant nonmajor bleeding, as well as mortality. Data were pooled across randomized controlled trials using random-effects meta-analysis models. Five randomized controlled trials including 23,063 patients that met the inclusion criteria were identified. Patients received treatment for nonvalvular atrial fibrillation (n = 14,264), deep vein thrombosis (n = 3,967), or acute symptomatic pulmonary embolism (n = 4,832). Overall, rivaroxaban was not associated with the risk of a composite end point of major or clinically relevant nonmajor bleeding (relative risk 0.99, 95% confidence interval 0.93 to 1.06). However, rivaroxaban was associated with a significant decrease in fatal bleeding (relative risk 0.48, 95% confidence interval 0.31 to 0.74). In 2 studies reporting intracranial bleeding events, rivaroxaban was associated with decreased risk compared with vitamin K antagonists. It was not associated with decreased risk for all-cause mortality (relative risk 0.89, 95% confidence interval 0.73 to 1.09). In conclusion, with a decrease in fatal bleeding and no suggestion of an increase in all-cause mortality, rivaroxaban has a favorable safety profile with respect to bleeding.
For many decades, vitamin K antagonists (VKAs) were the only available therapy for long-term anticoagulation. However, VKAs require frequent monitoring and regular dose adjustments as a result of food and drug interactions. To overcome these limitations, new oral anticoagulants have been developed as alternatives to VKAs. These agents include direct thrombin inhibitors such as AZD0837 and dabigatran, as well as direct factor Xa inhibitors including rivaroxaban, apixaban, betrixaban, edoxaban, LY-517717, and ym-150. These new oral anticoagulants do not require monitoring and can be given as a fixed dose. Moreover, there are no dietary interactions and only minimal drug interactions with these new agents. Rivaroxaban has been the most extensively studied of these agents, with several trials investigating the efficacy and safety of rivaroxaban for different indications. However, the risk for bleeding and mortality associated with this drug remains unclear. We therefore performed a systematic review and meta-analysis to compare the safety of rivaroxaban and VKAs, with a particular focus on bleeding and mortality.
Our systematic review and meta-analysis was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. We systematically searched the MEDLINE, Embase, Cumulative Index to Nursing & Allied Health Literature, Cochrane, and Science Citation Index Expanded databases from inception until May 2012, as well as clinical trials registries, for randomized controlled trials (RCTs) comparing the safety of rivaroxaban with that of VKAs. No language restrictions were imposed. The following search terms, Medical Subject Headings, and keywords were used: “rivaroxaban,” “Xarelto,” “BAY 59-7939,” “oral anticoagulant,” “oral thrombin inhibitor,” “oral factor Xa inhibitor,” “vitamin K antagonist,” “warfarin,” and “acenocoumarol.” Embase and MEDLINE searches were limited to RCTs and human subjects. Clinical trial registries were searched independently for potentially relevant RCTs not identified in the electronic database search.
Studies were included if they (1) were RCTs comparing rivaroxaban with VKAs, (2) included adults aged ≥18 years, (3) had treatment durations ≥30 days, and (4) included experimental arms of participants receiving a daily dose of 20 mg (standard dosage) of rivaroxaban. For trials comparing various treatment doses of rivaroxaban with VKA, only the 20-mg arms were included in our study. Studies that allowed the use of a lower dose of rivaroxaban in patients with diminished renal function were included provided that these patients were only a subset of included patients. Letters to the editor, editorials, reviews, and abstracts from conference proceedings were excluded from our study.
Data from identified studies were independently extracted by 2 reviewers. Disagreements were resolved by consensus or by a third party. Data extracted from relevant studies included patient and study characteristics, as well as safety outcomes. Patient characteristics included age, gender, mean time in therapeutic range, discontinuation rate, and previous use of VKAs. The main outcomes of interest were clinically relevant nonmajor bleeding, major bleeding (any, associated with a decrease in hemoglobin ≥2 g/dl, critical, fatal, gastrointestinal (GI), intracranial, macroscopic hematuria, intraocular, intra-articular, and epistaxis), and mortality.
Risk for bias was assessed using the Cochrane Collaboration’s tool for assessing risk for bias. This tool takes into account the following domains: sequence generation; allocation concealment; blinding of participants, personnel, and outcome assessors; incomplete outcome data; selective outcome reporting; and other potential sources of bias. The risk for bias in each of the domains was classified as high, low, or unclear. Risk for bias was assessed in duplicate, with disagreements resolved by consensus.
For each outcome of interest reported in ≥3 studies, data were pooled across RCTs using DerSimonian and Laird random-effects models to estimate relative risks (RR) and corresponding 95% confidence intervals (CIs). Data were pooled using count data when possible; if count data were not available, meta-analyses were conducted using hazard ratios on the log scale. For outcomes for which associations were present, we also calculated the number needed to treat and corresponding 95% CI. To examine the robustness of our results, we excluded a 12-week, phase 2 trial (Oral Direct Factor Xa Inhibitor BAY 59-7939 in Patients With Acute Symptomatic Deep-Vein Thrombosis [ODIXa-DVT]) in sensitivity analyses. We also conducted a sensitivity analysis that used a 0.5 continuity correction to include trials with no events in either arm. I 2 values were used to quantify the amount of between-study heterogeneity that was present. All analyses were conducted using Stata version 11 (StataCorp LP, College Station, Texas).
Our systematic search yielded 3,176 potentially eligible reports ( Figure 1 ). After the removal of duplicates, we screened the titles and abstracts of 2,621 records, 31 of which were subsequently reviewed. Five published reports met our inclusion criteria and were included in our study. The 5 RCTs included a total of 23,063 patients ( Table 1 ).
|Characteristic||ROCKET AF ∗||EINSTEIN-PE||EINSTEIN-Acute DVT||EINSTEIN-DVT||ODIXa-DVT|
|Total number of patients||7,131||7,133||2,419||2,413||1,731||1,718||135||137||120||126|
|Drug discontinuation rate||23.7%||22.2%||10.7%||12.3%||11.3%||14.2%||13%||11%||NR †||7.9%|
|Median length of follow-up (days)||707||180||180||84||114|
|Median length of treatment (days)||590||180||180||85||84|
|Number of patients lost to follow-up||18||14||8||10||15||18||0||1||NR †||0|
|Percentage of time during which international normalized ratio was in therapeutic range||NR||55%||NR||62.7%||NR||57.7%||NR||50.3%||NR||60.0%|
|Age (yrs) ‡||73 (65–78)||73 (65–78)||57.9 ± 7.3||57.5 ± 7.2||55.8 ± 16.4||56.4 ± 16.3||58.0 ¶||57.0 ¶||57.5 ± 15.9||58.4 ± 18.3|
|Previous use of VKAs §,⋮||62.3%||62.5%||0%||0%||0%||0%||0%||0%||0%||0%|
† Although the study indicates that 73 of the patients randomized to rivaroxaban discontinued treatment and that 1 was lost to follow-up, it does not specify the distribution of these numbers among the different rivaroxaban groups.
Sample sizes ranged from 246 to 14,264. Patients were predominantly men and received treatment for nonvalvular atrial fibrillation (n = 14,264), deep vein thrombosis (n = 3,967), or acute symptomatic pulmonary embolism (n = 4,832). The median treatment duration ranged from 84 to 590 days, and follow-up ranged from 84 to 707 days. Efficacy outcomes differed among studies; however, most studies included the safety outcomes of interest (i.e., bleeding and mortality). Safety analyses included all patients who received ≥1 dose of a study drug.
Quality assessment for the included trials was performed using the Cochrane tool for assessing the risk for bias. Overall, the studies had a low risk for bias. The EINSTEIN-Pulmonary Embolism (EINSTEIN-PE) EINSTEIN-Acute Deep Vein Thrombosis (EINSTEIN-Acute DVT), EINSTEIN-DVT Dose-Ranging, and ODIXa-DVT studies were open-label trials. These studies did, however, include independent adjudication committees that were blinded to treatment allocation during assessment of outcomes, which likely reduced the bias associated with an open-label design. These studies were therefore assigned ratings of “unclear” in the domain of blinding of participants, personnel, and outcome assessors. The EINSTEIN-PE, EINSTEIN-Acute DVT, and ODIXa-DVT studies did not include information regarding how attrition and exclusions were addressed and were therefore assigned ratings of “unclear” for the domain of incomplete outcome data. All of the included studies performed analyses for bleeding and mortality outcomes in the safety population rather than the intention-to-treat population and were therefore assigned ratings of “unclear” in the domain of other sources of bias. However, the proportion of patients excluded from these safety analyses represented <1% of the intention-to-treat population.
The different trials included in our systematic review reported several bleeding and mortality outcomes ( Table 2 ). Across all studies, the proportion of patients with the composite end point of major and clinically relevant nonmajor bleeding events was not significantly different between rivaroxaban (5.9% to 20.7%) and VKAs (6.3% to 20.3%). In 2 of the 3 studies in which critical bleeding was reported, rivaroxaban was associated with a decreased risk (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]: RR 0.68, 95% CI 0.53 to 0.89; EINSTEIN-PE: RR 0.27, 95% CI 0.12 to 0.62). In the third study, there was no significant difference between the groups. Rivaroxaban was also associated with a decreased risk for fatal bleeding in the ROCKET AF, EINSTEIN-PE, and EINSTEIN-Acute DVT studies. GI bleeding events were reported only in the ROCKET AF study and occurred predominantly in the rivaroxaban group ( Table 2 ). Rivaroxaban was found to be associated with a significant decrease in intracranial bleeding in 2 studies (ROCKET AF: RR 0.66, 95% CI 0.47 to 0.92; EINSTEIN-PE: RR 0.25, 95% CI 0.07 to 0.88; Table 3 ). None of the studies reported a significant difference in major bleeding events, bleeding associated with a decrease in hemoglobin ≥2 g/dl, or clinically relevant nonmajor bleeding. There was also no suggestion of an increased incidence of mortality in any of the studies, although the results of the ROCKET AF trial suggest that rivaroxaban may be associated with a decreased risk in all-cause mortality (RR 0.83, 95% CI 0.69 to 1.00).
|Outcome||ROCKET AF||EINSTEIN-PE||EINSTEIN-Acute DVT||EINSTEIN-DVT||ODIXa-DVT|
|RIV (n = 7,111)||VKA (n = 7,125)||RIV (n = 2,412)||VKA (n = 2,405)||RIV (n = 1,718)||VKA (n = 1,711)||RIV (n = 135)||VKA (n = 137)||RIV (n = 117)||VKA (n = 126)|
|Major/clinically relevant nonmajor bleeding events||1,475 (20.7%)||1,449 (20.3%)||254 (10.5%)||287 (11.9%)||140 (8.1%)||139 (8.1%)||8 (5.9%)||12 (8.8%)||11 (9.4%)||8 (6.3%)|
|Any||395 (5.6%)||386 (5.4%)||26 (1.1%)||52 (2.2%)||14 (0.8%)||20 (1.2%)||1 (0.7%)||2 (1.5%)||2 (1.7%)||0 (0.0%)|
|Associated with decrease in hemoglobin of ≥2 g/dl||305 (4.3%)||254 (3.6%)||17 (0.7%)||26 (1.1%)||10 (0.6%)||12 (0.7%)||NR||NR||2 (1.7%)||0 (0.0%)|
|Critical ∗||91 (1.3%)||133 (1.9%)||7 (0.3%)||26 (1.1%)||3 (0.2%)||3 (0.2%)||NR||NR||NR||NR|
|Fatal||27 (0.4%)||55 (0.8%)||2 (<0.1%)||3 (0.1%)||1 (<0.1%)||5 (0.3%)||NR||NR||0 (0.0%)||0 (0.0%)|
|GI||224 (3.2%)||154 (2.2%)||NR||NR||NR||NR||NR||NR||NR||NR|
|Intracranial||55 (0.8%)||84 (1.2%)||3 (0.1%)||12 (0.5%)||NR||NR||NR||NR||NR||NR|
|Macroscopic hematuria||26 (0.37%)||21 (0.29%)||NR||NR||NR||NR||NR||NR||NR||NR|
|Intraocular||17 (0.2%)||24 (0.3%)||2 (<0.1%)||2 (<0.1%)||NR||NR||NR||NR||NR||NR|
|Intra-articular||16 (0.2%)||21 (0.3%)||0 (0.0%)||3 (0.1%)||NR||NR||NR||NR||NR||NR|
|Epistaxis||13 (0.2%)||14 (0.2%)||NR||NR||NR||NR||NR||NR||NR||NR|
|Clinically relevant nonmajor bleeding||1,185 (16.7%)||1,151 (16.2%)||228 (9.5%)||235 (9.8%)||126 (7.3%)||119 (7.0%)||7 (5.2%)||10 (7.3%)||NR||NR|
|All-cause mortality||208 (3.0%)||250 (3.5%)||58 (2.4%)||50 (2.1%)||38 (2.2%)||49 (2.9%)||4 (3.0%)||5 (3.6%)||4 (3.4%)||1 (0.8%)|
|Outcome||ROCKET AF||EINSTEIN-PE||EINSTEIN-Acute DVT||EINSTEIN-DVT||ODIXa-DVT||Pooled †|
|RR||95% CI||RR||95% CI||RR||95% CI||RR||95% CI||RR||95% CI||RR||95% CI||I 2|
|Major/clinically relevant nonmajor bleeding events||1.02||0.95–1.09||0.88||0.75–1.04||1.00||0.80–1.25||0.68||0.29–1.60||1.44||0.60–3.47||0.99||0.93–1.06||2.9|
|Associated with decrease in hemoglobin of ≥2 g/dl||1.20||1.02–1.41||0.65||0.36–1.20||0.83||0.36–1.91||NR||NR||5.25||0.26–108.20||0.99||0.66–1.49||41.9|
|Clinically relevant nonmajor bleeding||1.03||0.96–1.11||0.97||0.81–1.15||1.05||0.83–1.34||0.71||0.28–1.81||NR||NR||1.02||0.96–1.09||0.0|