Consequences of inadequate direct oral anticoagulant (DOAC) dosing

Yao X, et al. Non-vitamin K antagonist oral anticoagulant dosing in patients with atrial fibrillation and renal dysfunction. J Am Coll Cardiol 2017;69:2779-90.

Bottom line:

  • Many patients with non-valvular AF receiving a DOAC receive inappropriate doses.

  • In patients with an indication to reduced the DOAC dose, "overdosing" (using regular doses) is associated with a 2-fold higher risk of major bleeding vs using appropriate reduced doses.

  • In patients taking apixaban without an indication to reduce the dose, "underdosing" (using reduced doses) apixaban is associated with a ~5-fold increased risk of stroke/systemic embolism.


Design summary

  • Cohort study using administrative claims database & linked lab data
  • Included: US patients enrolled with Medicare Advantage with non-valvular AF treated with apixaban, dabigatran or rivaroxaban between Oct 2010-Sept 2015 who had serum creatinine results available within 1 year
  • Excluded: Valvular heart disease, other indication for DOAC, eGFR <15 (calculated using CKD-EPI)
  • Account for bias & confounding: Propensity score matching to balance baseline characteristics, statistical adjustment with Cox proportional hazards regression, numerous sensitivity analyses (such as changing criteria for apixaban dose reduction to age >80 + SCr >132 umol/L, dabigatran dose reduction criteria to eGFR <50), different matching criteria, subgroup analyses based on baseline criteria & individual DOACs, analysis of ICH as individual outcome, & analysis of 'falsification outcomes')
    • Note on the point of evaluating falsification outcomes: Falsification outcomes are outcomes that should not be associated with the presence/absence of an exposure. This is used to identify residual confounding after matching & statistical adjustment (i.e. if there's an association with an outcome that the exposure shouldn't affect, it implies residual confounding).

Results for PICO 1: Normal renal function (no need for DOAC dose reduction), n=13,392

  • Overall, 16.5% of patients underdosed
  • Increased risk of stroke/systemic embolism with reduced dose ("underdosed") vs normal dose of apixaban
    •  2.6% vs 0.5% per year, hazard ratio (HR) 4.87 (1.30-18.26)
  • No statistically significant difference in stroke/systemic embolism with reduced vs normal doses of dabigatran or rivaroxaban, though confidence intervals are very wide
  • No statistically significant difference in major bleeding with reduced vs normal doses of any DOAC

PICO 2: Renal impairment (indication for DOAC dose reduction), n=1,473

  • Overall, 48.5% of patients were overdosed
  • Definition of indication for dose reduction: Apixaban (SCr >132 umol/L), dabigatran (eGFR <30), rivaroxaban (eGFR <50)
  • Increased risk of major bleeding with normal ("overdosing") vs reduced dose of any DOAC
    • 11.3% vs 5.1% per year, HR 2.19 (1.07-4.46)
  • No difference of stroke/systemic embolism with normal vs reduced dose of any DOAC
    • 2.32% vs 1.85%, HR 1.66 (0.40-6.88)


  • Due to a limited sample size & low outcome event rates, many analyses were underpowered, particularly when evaluating DOACs individually.
  • Due to limitations with available data, study definitions for criteria to reduce DOAC dose were not entirely consistent with drug label (e.g. weight for apixaban)
  • The analyses only partially accounted for known drug interactions (e.g. amiodarone, digoxin), & could not account for unknown or novel drug interactions (e.g. P-glycoprotein interaction between dabigatran & simvastatin).