Do you consider bleeding risk in elderly, frail patients with atrial fibrillation to be similar for all NOACs?

I believe that apixaban carries a lower risk of bleeding, with particular reference to GI bleed, when compared to rivaroxaban and dabigatran. This is true in the population of AF patients at large and most probably in frail patients as well.

From lowest to highest risk of intracranial and extracranial bleeding based on anecdotal experience:

apixaban < rivaroxaban < dabigatran < warfarin

I've never seen a patient on edoxaban.

The prior evidence would suggest that not all NOACS have the same bleeding risk. More specifically, Apixaban appears to have a lower risk of GI bleeding than Rivaroxaban. While this is based on observational analyses, and head-to-head trials of non-generic medications are hard to come by, I have encountered a similar pattern in my practice. The FRAIL-AF trial was not designed to answer this question.

Apixiban, administered q 12 H, has the shortest half life, and would dissipate most rapidly if bleeding were to occur.