Article Text

Download PDFPDF
Systematic review and meta-analysis
Non-vitamin-K oral anticoagulants reduce mortality, stroke and intracranial haemorrhage when compared with warfarin in randomised trials of patients with non-valvular atrial fibrillation
  1. Benjamin A Steinberg
  1. Electrophysiology Section, Duke University Medical Center, Duke Clinical Research Institute, Durham, North Carolina, USA
  1. Correspondence to : Dr Benjamin A Steinberg, Duke Clinical Research Institute, PO Box 17969, Durham, NC 27715, USA; benjamin.steinberg{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Commentary on: OpenUrlCrossRefPubMedWeb of Science


Historically, the standard medication for stroke prevention in atrial fibrillation (AF) has been a vitamin-K antagonist (warfarin). However, several non-vitamin-K oral anticoagulants (NOACs) have been developed and shown to be at least as effective as dose-adjusted warfarin in their respective phase-3 clinical trials.1–4 These include the direct thrombin inhibitor dabigatran and the factor Xa inhibitors rivaroxaban, apixaban and edoxaban. Yet, despite these large trials, there remains limited power to examine certain end points and treatment effects in subgroups.


In a prespecified meta-analysis, the authors pooled trial-level data from the four existing phase-3 randomised clinical trials comparing a NOAC with dose-adjusted warfarin for stroke prevention in AF. These included the RE-LY, ROCKET AF, ARISTOTLE and ENGAGE AF-TIMI 48 trials.

Importantly, for the two trials that randomised to more than one dose level of the NOAC (RE-LY and ENGAGE AF-TIMI 48), the authors included the higher dose in their primary analysis.

The investigators assessed efficacy and safety outcomes in the overall cohort, and in important subgroups by age, sex, history of diabetes mellitus, creatinine clearance, CHADS2 score, and prior use and quality of warfarin treatment. Random effects models were used to calculate pooled estimates, while heterogeneity of effects was measured by the I2 statistic. Results are reported as relative risks (RR).


A total of 42 411 patients were assigned to a NOAC versus 29 272 assigned to warfarin. Patients assigned to NOACs were found to have consistently lower risk of stroke or systemic embolism (RR=0.81, p<0.0001), intracranial haemorrhage (RR=0.48, p<0.0001) and all-cause mortality (RR=0.90, p=0.0003).

While there were no significant differences for the end points of ischaemic stroke (RR=0.92, p=0.1) or myocardial infarction (RR=0.97, p=0.77), patients assigned to NOACs did have significantly more gastrointestinal bleeding (RR=1.25, p=0.043). Overall, major bleeding tended to be lower for NOAC patients (RR=0.86, p=0.06); however, heterogeneity was highest for the end points of major bleeding (I2=83%), gastrointestinal bleeding (I2=74%) and myocardial infarction (I2=43%).

Across prespecified subgroups, rates of stroke or systemic embolism and major bleeding consistently favoured NOAC patients. Finally, the authors note that in the sensitivity analyses of low-dose NOACs, a dramatic reduction in intracranial haemorrhage was observed at the expense of increased risk of ischaemic stroke.


These data provide support for the use of NOACs as a group for stroke prevention in patients with non-valvular AF. Some guidelines have already stated explicit preference for NOACs over warfarin, where appropriate. This meta-analysis also provides reassurance for the use of NOACs in several subgroups of patients that were difficult to assess in any single study. However, there are limitations to the conclusions that can be drawn from these data.

First, this analysis cannot provide data on the use of these drugs in patients who were excluded from the trials in question—most importantly patients with valvular AF (usually due to rheumatic disease) or mechanical valve replacements and patients with severe renal dysfunction. Furthermore, clinicians should not accept this meta-analysis as conclusive evidence that ‘all NOACs are created equal’. There remain significant differences, and this was manifest in the original trials and the heterogeneity statistics in the current analysis. For example, differences for the end point of major bleeding between NOAC and warfarin varied among the individual trials, and this yielded significant heterogeneity (I2=84%) in meta-analysis. It is not clear that NOACs as a group have a consistent effect. Differences in drugs, trial designs, patient populations, end point definitions and outcomes cannot be fully addressed through meta-analysis. Importantly, the package labels and clinical guidelines distinguish among the NOACs for prevention of stroke in non-valvular AF, and there are no prospective, head-to-head comparisons between NOACs.

The analysis by Ruff and colleagues provides an important contribution to our knowledge of NOACs and their overall treatment effects in the clinical trial AF population. Subsequent studies of community use of NOACs will provide vital additional data on their comparative safety and effectiveness.


View Abstract


  • Competing interests BAS was funded by NIH T-32 training grant #5 T32 HL 7101-38, and reports significant research support from Janssen Pharmaceuticals, Inc.

  • Provenance and peer review Commissioned; internally peer reviewed.