The growing use of direct oral anticoagulants, in particular among older subjects, raises questions about the limits of the evidence-based medicine. The phase III studies that have validated the efficacy and the safety profile of these molecules (dabigatran, rivaroxaban, apixaban, edoxaban) in their both indications, the venous thromboembolic disease and the non-valvular atrial fibrillation raise concerns in four major fields: the financial support of pharmaceutical companies, the links of interest for many authors with the industry, the study design (exclusively non-inferiority studies), and the poor representativeness of the older subjects included. All these points are discussed, using data of sub-groups studies, post-marketing studies and recent meta-analysis. The lack of data for the very old subjects, with frailty or comorbidities, remains the main concern from these phase III studies.