Clinical outcomes of rivaroxaban in NVAF stroke prevention and VTE treatment

Abstract

Ten sources examining rivaroxaban for stroke prevention in nonvalvular atrial fibrillation were reviewed, including three reports from the ROCKET AF trial, five observational studies, and one systematic review. One systematic review also included venous thromboembolism treatment. Rivaroxaban demonstrated non-inferiority to warfarin for preventing stroke or systemic embolism in the pivotal randomized trial (1.7% vs 2.2% per year, HR 0.79, 95% CI 0.66-0.96), with consistent findings across multiple large observational studies reporting 18-19% relative risk reductions. Rivaroxaban showed particular benefit in reducing severe stroke by 44-48% and consistently lowered intracranial hemorrhage by 30-75% and fatal bleeding by approximately 50% across studies. Major bleeding rates were generally similar between rivaroxaban and warfarin, though rivaroxaban increased gastrointestinal bleeding in the randomized trial. Large observational studies demonstrated 20% reductions in all-cause mortality and dramatic reductions in poststroke mortality (24-59%). Benefits were consistent across subgroups including patients with moderate renal impairment, elderly populations, and those at high bleeding risk. Evidence for venous thromboembolism treatment outcomes was limited to one systematic review showing generally neutral or positive effects compared to warfarin.

Methods

We analyzed 10 sources from an initial pool of 200, using 8 screening criteria. Each paper was reviewed for 8 key aspects that mattered most to the research question.

Records from Elicit search

n = 200

Papers screened using: Target Population, Rivaroxaban Intervention, Clinical Outcomes, Study Design, Valvular Status, Indication Relevance, Study Type Appropriateness, Rivaroxaban Analysis

n = 200

Papers screened out

n = 190

Papers included for extraction

n = 10

Paper search

We performed a semantic search across over 138 million academic papers from the Elicit search engine, which includes all of Semantic Scholar and OpenAlex.

We ran this query: "Clinical outcomes of rivaroxaban in NVAF stroke prevention and VTE treatment"

The search returned 200 total results from Elicit.

We retrieved 200 papers most relevant to the query for screening.

Screening

We screened in sources based on their abstracts that met these criteria:

• Target Population: Does the study include adult patients (≥18 years) with confirmed non-valvular atrial fibrillation requiring stroke prevention OR patients with acute or chronic venous thromboembolism requiring treatment?
• Rivaroxaban Intervention: Is rivaroxaban used as the primary intervention or as one of the comparison arms in this study?
• Clinical Outcomes: Does the study report at least one clinical outcome measure (efficacy endpoints such as stroke, systemic embolism, VTE recurrence, or safety endpoints such as bleeding events, mortality)?
• Study Design: Is this study a randomized controlled trial, prospective or retrospective cohort study, case-control study, systematic review, or meta-analysis?
• Valvular Status: Does the study include patients other than those with valvular atrial fibrillation or mechanical heart valves (i.e., does it include non-valvular populations)?
• Indication Relevance: Does the study examine rivaroxaban for non-valvular atrial fibrillation stroke prevention or venous thromboembolism treatment (rather than exclusively for other indications such as coronary artery disease or orthopedic surgery prophylaxis)?
• Study Type Appropriateness: Is this study something other than a preclinical study, in vitro study, pharmacokinetic/pharmacodynamic study without clinical outcomes, case report, or case series with fewer than 10 patients?
• Rivaroxaban Analysis: Can rivaroxaban outcomes be separately analyzed from other interventions in this study?

We considered all screening questions together and made a holistic judgement about whether to screen in each paper.

Data extraction

We asked a large language model to extract each data column below from each paper. We gave the model the extraction instructions shown below for each column.

• Study Population:

Extract study design and patient population details specifically for rivaroxaban in NVAF stroke prevention or VTE treatment, including:

• Study type (RCT, observational, registry, etc.)

• Primary indication (NVAF stroke prevention vs VTE treatment vs both)

• Sample size for rivaroxaban arm

• Key patient characteristics (age, gender, comorbidities, stroke risk scores like CHADS2/CHA2DS2-VASc for NVAF studies)

• Inclusion/exclusion criteria relevant to the indication

• Geographic setting and study period

• Rivaroxaban Regimen:

Extract rivaroxaban treatment details and comparators, including:

• Rivaroxaban dose and frequency (e.g., 20mg daily, 15mg twice daily)

• Treatment duration or follow-up period

• Dose adjustments for renal impairment or other factors

• Comparator treatments (warfarin, other DOACs, placebo, no treatment)

• Comparator dosing and management protocols

• Treatment adherence or discontinuation rates if reported

• Primary Efficacy:

Extract primary efficacy outcomes for rivaroxaban in the specific indication (NVAF stroke prevention or VTE treatment), including:

• For NVAF: stroke (ischemic, hemorrhagic, or composite), systemic embolism, composite thromboembolic events

• For VTE: VTE recurrence, pulmonary embolism, deep vein thrombosis

• Event rates per 100 patient-years or as reported

• Hazard ratios, relative risks, or other effect measures with confidence intervals

• P-values for superiority, non-inferiority, or equivalence

• Time to event data if available

• Bleeding Safety:

Extract bleeding outcomes and safety events for rivaroxaban treatment, including:

• Major bleeding (as defined by study, typically ISTH criteria)

• Clinically relevant non-major bleeding

• Minor bleeding

• Intracranial hemorrhage specifically

• Fatal bleeding

• Gastrointestinal bleeding

• Event rates per 100 patient-years or as reported

• Hazard ratios or risk comparisons with confidence intervals

• Bleeding severity classifications and outcomes

• Mortality Outcomes:

Extract mortality data for patients receiving rivaroxaban for NVAF stroke prevention or VTE treatment, including:

• All-cause mortality rates

• Cardiovascular mortality

• Stroke-related mortality (for NVAF studies)

• VTE-related mortality (for VTE studies)

• Bleeding-related mortality

• Event rates per 100 patient-years or as reported

• Hazard ratios or relative risks with confidence intervals

• Time points assessed

• Secondary Outcomes:

Extract additional clinical outcomes beyond primary efficacy, bleeding, and mortality for rivaroxaban treatment, including:

• Myocardial infarction

• Transient ischemic attack

• Hospitalization rates

• Treatment discontinuation due to adverse events

• Other serious adverse events

• Net clinical benefit measures (if calculated)

• Quality of life measures

• Event rates and effect measures with confidence intervals where available

• Subgroup Findings:

Extract any subgroup analyses or population-specific findings for rivaroxaban outcomes in NVAF stroke prevention or VTE treatment, including:

• Age subgroups (elderly, very elderly)

• Gender differences

• Renal function subgroups

• Bleeding risk subgroups (HAS-BLED scores)

• Stroke risk subgroups (CHADS2/CHA2DS2-VASc scores for NVAF)

• Geographic or ethnic subgroups

• Prior stroke or VTE history subgroups

• Effect measures and statistical significance for each subgroup

• Interaction testing results

• Study Context:

Extract contextual factors that may affect interpretation and generalizability of rivaroxaban outcomes, including:

• Study duration and length of follow-up
• Clinical setting (hospital, outpatient, community-based)
• Healthcare system characteristics
• Concomitant medications or interventions
• Study limitations acknowledged by authors
• Funding source and potential conflicts of interest
• Time period of study conduct (important for evolving clinical practice)
• Loss to follow-up rates and handling of missing data

Results

Characteristics of Included Studies

Ten sources were included in this review, comprising three reports from the ROCKET AF randomized controlled trial, five observational studies, one systematic review of observational studies. All studies focused on rivaroxaban for NVAF stroke prevention, with one systematic review also including VTE treatment.

Patient populations varied in age (mean 71.5-79 years), stroke risk (mean CHADS₂ scores 2.0-3.5, CHA₂DS₂-VASc scores 3.0-3.9), and comorbidities. Rivaroxaban dosing followed standard regimens of 20 mg daily with dose reduction to 15 mg daily for moderate renal impairment (creatinine clearance 30-49 mL/min). Treatment duration ranged from a median of 590 days in the ROCKET AF trial to a mean of 329 days in XANTUS and 22.7 months in the Spanish registry.

Primary Efficacy Outcomes

Stroke and Systemic Embolism

In the pivotal ROCKET AF trial, rivaroxaban demonstrated non-inferiority to warfarin for stroke or systemic embolism prevention, with event rates of 1.7% per year versus 2.2% per year (HR 0.79, 95% CI 0.66-0.96, p<0.001 for non-inferiority). The intention-to-treat analysis showed similar results (2.1% vs 2.4% per year, HR 0.88, 95% CI 0.74-1.03). This finding was reproduced in a large US observational study that reported stroke rates of 0.97 per 100 patient-years for rivaroxaban versus 1.32 for warfarin (HR 0.82, 95% CI 0.76-0.88).

Real-world registries reported lower absolute event rates than the RCT. The XANTUS study observed stroke rates of 0.7 events per 100 patient-years, while the Spanish registry reported 1.8 events per 100 patient-years. The systematic review of US observational studies found that 68.8% showed positive effects favoring rivaroxaban and 31.2% showed neutral outcomes for stroke and systemic embolism.

In a Chinese cohort of high bleeding risk patients, rivaroxaban was non-inferior to warfarin for preventing stroke and systemic embolism (13.2% vs 19.2%, HR 0.681, 95% CI 0.512-0.906, p<0.001 for non-inferiority).

Stroke Severity

Two large US observational studies provided novel insights into stroke severity outcomes. Alberts et al. demonstrated that rivaroxaban reduced severe stroke risk by 48% (NIHSS 16-42; HR 0.52, 95% CI 0.33-0.82) and minor stroke risk by 19% (NIHSS 1 to <5; HR 0.81, 95% CI 0.68-0.96), with no significant difference for moderate stroke. These findings were corroborated by Milentijevic et al., who reported similar risk reductions for severe stroke (HR 0.44, 95% CI 0.22-0.91), moderate stroke (HR 0.88, 95% CI 0.78-0.99), and minor stroke (HR 0.83, 95% CI 0.74-0.93).

Bleeding Safety Outcomes

Major Bleeding

Major bleeding rates were generally similar between rivaroxaban and warfarin across studies. In ROCKET AF, major and clinically relevant non-major bleeding occurred in 14.9% per year with rivaroxaban versus 14.5% per year with warfarin (HR 1.03, 95% CI 0.96-1.11, p=0.44). This pattern was consistent in the renal impairment subgroup (17.82 vs 18.28 per 100 patient-years, p=0.76).

Real-world registries showed somewhat lower bleeding rates. XANTUS reported major bleeding at 2.1 events per 100 patient-years, while the Spanish registry observed 3.2 events per 100 patient-years. In the Chinese high bleeding risk cohort, rivaroxaban showed superior safety with major bleeding rates of 6.23% versus 11.98% for warfarin (HR 0.469, 95% CI 0.314-0.702, p<0.001).

The systematic review of US observational studies found a mixed picture: 57.7% of studies showed neutral effects, 38.5% showed negative effects (higher bleeding with rivaroxaban), and 3.8% showed positive effects for major bleeding.

Intracranial Hemorrhage

A consistent finding across studies was reduced intracranial hemorrhage with rivaroxaban. ROCKET AF demonstrated a significant reduction (0.5% vs 0.7% per year, p=0.02), with similar rates in the renal impairment subgroup (0.71 vs 0.88 per 100 patient-years, p=0.54). XANTUS observed intracranial bleeding at 0.4 events per 100 patient-years, while the Spanish registry reported the same rate. In the Chinese cohort, rivaroxaban reduced intracranial hemorrhage by 75% (HR 0.249, 95% CI 0.139-0.448, p<0.001).

Fatal Bleeding

Fatal bleeding occurred less frequently with rivaroxaban across studies. ROCKET AF reported 0.2% versus 0.5% per year (p=0.003), with similar findings in the renal impairment subgroup (0.28% vs 0.74% per 100 patient-years, p=0.047). XANTUS observed fatal bleeding at 0.2 events per 100 patient-years. The Chinese study confirmed this advantage with fatal bleeding rates of 2.67% for rivaroxaban versus 5.45% for warfarin (HR 0.530, 95% CI 0.287-0.979, p=0.043).

Gastrointestinal Bleeding

Rivaroxaban was associated with higher gastrointestinal bleeding rates in ROCKET AF (3.2% vs 2.2% per year, p<0.001). However, XANTUS reported lower rates at 0.9 events per 100 patient-years.

Mortality Outcomes

All-cause mortality was lower with rivaroxaban in several studies. ROCKET AF showed a trend toward reduced mortality (1.9% vs 2.2% per year, HR 0.85, 95% CI 0.70-1.02, p=0.07), though not statistically significant. The US observational study by Milentijevic et al. demonstrated a significant 20% reduction in all-cause mortality (HR 0.80, 95% CI 0.74-0.86). Alberts et al. found similar all-cause mortality reduction (2.26 vs 3.55 per 100 patient-years, HR 0.80, 95% CI 0.74-0.86).

Poststroke mortality was markedly reduced with rivaroxaban. Alberts et al. reported 24% lower poststroke mortality (0.31 vs 0.44 per 100 patient-years, HR 0.76, 95% CI 0.61-0.95) and a dramatic 59% reduction in 30-day poststroke mortality (0.09 vs 0.20 per 100 patient-years, HR 0.41, 95% CI 0.28-0.60). Milentijevic et al. confirmed these findings with similar effect sizes.

XANTUS reported all-cause mortality at 1.9 events per 100 patient-years, while the Spanish registry observed 5.5 events per 100 patient-years. The Chinese study found no significant difference in all-cause mortality (3.56% vs 4.79%, HR 0.760, 95% CI 0.435-1.329, p=0.336).

Secondary Clinical Outcomes

Myocardial infarction rates showed no significant difference between rivaroxaban and warfarin in ROCKET AF (0.9% vs 1.1% per year, HR 0.81, 95% CI 0.63-1.06, p=0.12). The Spanish registry reported 1.0 events per 100 patient-years, while the Chinese study found similar rates between groups (4.07% vs 5.66%, HR 1.940, 95% CI 0.495-1.069, p=0.254).

Treatment discontinuation rates in ROCKET AF were 23.7% for rivaroxaban and 22.2% for warfarin. XANTUS reported a discontinuation rate of 20.1% at the end of the observation period, with 7.9% of patients discontinuing due to adverse events.

Subgroup Analyses

Renal Function

Patients with moderate renal impairment (creatinine clearance 30-49 mL/min) represented 20.7% of the ROCKET AF population and were older (mean age 79 years vs 73 years) with higher event rates regardless of treatment. In this subgroup, rivaroxaban maintained non-inferiority for stroke or systemic embolism (2.32 vs 2.77 per 100 patient-years, HR 0.84, 95% CI 0.57-1.23), with similar bleeding rates but significantly lower fatal bleeding (0.28% vs 0.74% per 100 patient-years, p=0.047).

A large US observational study stratified by renal function found that rivaroxaban users with diagnosed renal dysfunction had 45% lower stroke rates (HR 0.55, p=0.0004) and 38% lower thromboembolic event rates (HR 0.62, p<0.0001) compared to warfarin. Alberts et al. reported that patients with prior renal disease experienced 21% stroke risk reduction and 31% mortality risk reduction with rivaroxaban, compared to 18% and 14% reductions in those without renal disease.

Age

XANTUS demonstrated that major bleeding increased markedly with age: 0.9 events per 100 patient-years in patients <65 years, 1.7 in those 65-75 years, and 3.2 in those >75 years. The Spanish registry confirmed higher stroke, major bleeding, and death rates in patients >75 years compared to younger patients.

Treatment Duration

Longer treatment duration (>9 months) was associated with more pronounced benefits in the Alberts study, including greater reductions in severe stroke (HR 0.15, 95% CI 0.05-0.47), poststroke mortality (HR 0.49, 95% CI 0.33-0.75), and 30-day poststroke mortality (HR 0.32, 95% CI 0.17-0.62).

High Bleeding Risk

The Chinese study specifically enrolled patients with HAS-BLED scores ≥3, demonstrating that rivaroxaban remained non-inferior for efficacy (HR 0.681, p<0.001) and superior for bleeding safety (HR 0.469, p<0.001) even in this high-risk population.

Synthesis

The evidence consistently demonstrates rivaroxaban’s non-inferiority to warfarin for stroke prevention in NVAF across diverse populations and settings. The ROCKET AF trial established this foundation in a multinational RCT, subsequently confirmed by multiple large observational studies in the US, Europe, and Asia.

The apparent discrepancy in absolute stroke rates between the RCT (1.7-2.2% per year) and observational registries (0.7-1.8 events per 100 patient-years) reflects differences in patient selection and risk profiles rather than conflicting efficacy findings. The ROCKET AF population had higher baseline stroke risk (mean CHADS₂ score 3.5) compared to XANTUS (mean CHADS₂ 2.0, CHA₂DS₂-VASc 3.4), explaining the higher event rates. Additionally, the RCT’s strict follow-up protocols captured more events than real-world registries prone to under-ascertainment.

Rivaroxaban’s safety profile shows a nuanced pattern requiring careful interpretation. While major bleeding rates were similar overall, rivaroxaban consistently reduced intracranial hemorrhage by approximately 30-75% and fatal bleeding by approximately 50%. However, gastrointestinal bleeding was higher with rivaroxaban in the RCT, though real-world registries reported lower rates. This suggests that patient selection, concomitant medications (particularly antiplatelet agents), and management strategies influence bleeding location and severity. The systematic review’s finding that 38.5% of observational studies showed higher major bleeding with rivaroxaban likely reflects heterogeneity in bleeding definitions, detection methods, and population characteristics rather than fundamental differences in drug safety.

The mortality benefit observed in large US observational studies but not definitively in ROCKET AF warrants explanation. The observational studies had longer follow-up periods (mean 27-28 months) versus ROCKET AF’s median 707 days, allowing more time for mortality benefits to emerge. Additionally, the dramatic reduction in severe stroke—which was not a prespecified endpoint in ROCKET AF—mechanistically explains the lower poststroke mortality, as severe strokes carry substantially higher case fatality rates. The RCT’s lack of statistical significance for mortality (HR 0.85, p=0.07) likely reflects insufficient power for this secondary endpoint rather than absence of effect.

Patient characteristics modify treatment effects in predictable ways. Patients with renal impairment experience higher baseline event rates but maintain rivaroxaban’s efficacy advantage, with particularly pronounced benefits in the diagnosis-based analysis (45% stroke reduction) compared to the creatinine clearance-based analysis. This discrepancy likely reflects the larger sample size in the diagnosis-based analysis (39,872 vs 874 rivaroxaban users), providing greater statistical power to detect differences. Elderly patients demonstrate higher bleeding rates but also higher stroke rates, suggesting that the net clinical benefit remains favorable despite increased bleeding risk. Longer treatment duration enhances benefits, consistent with cumulative risk reduction over time.

Geographic variation in outcomes exists but does not invalidate overall conclusions. The Chinese study’s larger effect sizes for both efficacy (32% stroke reduction) and safety (53% major bleeding reduction) compared to Western studies may reflect ethnic differences in bleeding risk, healthcare system factors affecting warfarin management, or the selected high bleeding risk population. The US observational studies’ consistent findings across different databases support generalizability within the American healthcare context.