Limited real-world evidence exists on comparative effectiveness and safety of NOACs vs warf by NOAC dosage.

A retrospective observational study of NVAF patients (pts) newly initiating apixaban, dabigatran, rivaroxaban, or warfarin from 01 January 2013–30 September 2015 was conducted using CMS Medicare data and four other US commercial claims databases, covering>180 million beneficiaries annually (∼56% of US population). After propensity score matching in each database between each standard NOAC dose and warfarin (5mg BID api-warf, 150mg BID dabi-warf, 20mg QD riva-warf), as well as each lower NOAC dose and warf (2.5mg BID api-warf, 75mg BID dabi-warf, and 10 or 15mg QD riva-warf), the resulting pts records were pooled. Cox models were used to estimate S/SE and MB hazard ratios. S/SE included ischemic stroke, hemorrhagic stroke, and SE; MB included gastrointestinal (GI) bleeding, intracranial hemorrhage (ICH), and other MB.

Pts initiating lower and standard NOAC doses had different baseline characteristics, such as age and renal disease. Pts data were assessed for a mean of 7–8 months. Standard- and lower-dose api pts were each associated with lower rates of S/SE and MB vs warf. Standard and lower-dose dabi pts each had similar rates of S/SE compared to warf. Standard-dose dabi pts had a lower rate of MB and lower-dose dabi pts had a similar rate of MB vs warf. Standard-dose riva pts were associated with a lower S/SE rate, and lower-dose riva pts had a similar rate of S/SE compared to warf. Standard- and lower-dose riva pts were each associated with higher MB rates compared to warf. All doses of NOACs were associated with lower rates of ICH compared to warf.

In this large observational study, api was the only NOAC associated with lower rates of S/SE and MB for both doses compared to warf. Dose selection criteria cannot be ascertained from current data sources. Future studies of pts who were appropriately dosed should be warranted.