Differentiating between “no proof of difference” and “proof of no difference” for new oral anticoagulants