Application of randomized clinical trial data to actual practice: apixaban therapy for reduction of stroke risk in non-valvular atrial fibrillation patients

BACKGROUND: Clinical event rates may differ among patients treated in the real world (RW) compared to randomized controlled trials (RCTs). When translating the efficacy of new treatments to RW, the relative risk reductions (RRRs) from RCTs may produce different absolute risk reductions in RW. OBJECTIVE: To estimate the absolute effect of apixaban on stroke and major bleeding (MB) rates in a RW non-valvular atrial fibrillation (NVAF) population. METHODS: NVAF patients were selected during 2007-2010 from a population of U.S. commercial and Medicare health plans using the Medco claims database. Pharmacy claims were used to define warfarin exposure periods. Stroke and MB were identified using diagnosis codes. RW event rates were calculated during periods of warfarin exposure. The numbers of stroke and MB events estimated to be avoided in RW with apixaban versus warfarin were calculated by applying RRRs from the ARISTOTLE trial to RW rates from the Medco database. The Medco data did not contain information for patients receiving apixaban as it was not on the market at the time of analysis. RESULTS: Stroke and MB rates among RW NVAF patients during warfarin exposure were higher compared with event rates in patients treated with warfarin in ARISTOTLE (stroke: 5.29 vs. 1.51 per 100 person years (PYs); MB: 10.78 vs. 3.09 per 100 PYs). If RRRs from trials persist in RW, apixaban vs. warfarin would result in greater absolute risk reductions (ARRs) and a lower number needed to treat (NNT) in RW vs. ARISTOTLE (stroke: 91 vs. 313; MB: 30 vs. 105). CONCLUSION: The impact of apixaban, as an alternative to warfarin in RW may be greater than in RCTs. The NNT with apixaban versus warfarin in RW may be lower versus ARISTOTLE if RRRs from the trial persists in RW and if baseline stroke and MB rates among RW patients are higher compared to trial participants.