In Reply In their Letter to the Editor, Scridon and Șerban raise important issues regarding major bleeding in patients with diabetes in the non–vitamin K antagonist oral anticoagulant (NOAC) vs warfarin trials of patients with atrial fibrillation.1- 5 We agree that apixaban,3 edoxaban,2 and the 110-mg dose of dabigatran1 significantly reduced bleeding in their respective overall trial populations compared with warfarin and that this was not seen with rivaroxaban4 nor with the 150-mg dose of dabigatran.6 While there was statistically significant heterogeneity in the ARISTOTLE trial3 between diabetes status and treatment for the end point of major bleeding (less bleeding with apixaban vs warfarin in those without diabetes; no difference in bleeding risk between treatments in the group with diabetes; interaction, P = .003), no such heterogeneity of effect on bleeding risk by diabetes status was seen with edoxaban in a similar analysis from the ENGAGE AF-TIMI 48 trial2 (bleeding was reduced in patients with and without diabetes). However, given the large number of subgroups analyzed raising the possibility of type 1 error and the lack of a prespecified hypothesis or mechanistic explanation for the heterogeneity of comparative bleeding effects by diabetes observed with apixaban in the ARISTOTLE trial,3 we believe that this observation should be interpreted cautiously as hypothesis generating. Furthermore, in a meta-analysis of data from the pivotal NOAC trials,6 the risks of major bleeding with NOAC vs warfarin in patients with vs without diabetes were not statistically different (interaction, P = .12).
Plitt A, McGuire DK, Giugliano RP. Major Bleeding in Patients With Diabetes and Atrial Fibrillation Treated With New Oral Anticoagulants—Reply. JAMA Cardiol. Published online June 28, 2017. doi:10.1001/jamacardio.2017.2034