Variations in Published Stroke Rates Confound Net Anticoagulation Benefit Analyses
Variations in published stroke rates for patients with AF not receiving anticoagulants result in differences in the calculated net clinical benefit of anticoagulation.
Results varied widely when researchers used published stroke rates from several different studies to calculate the net clinical benefit of anticoagulation in patients with atrial fibrillation (AF). According to a study published in the Annals of Internal Medicine, standardized guidelines to obtain precise and reproducible stroke rates should be a focus in future AF trials.
Authors of this study sought to determine whether the variation observed in published AF-associated stroke rates would have an effect on the net clinical benefit of anticoagulation in adults with atrial fibrillation.
The study sample included 33,434 adults with incident AF receiving oral anticoagulants. Applying the Markov state-transition model, the investigators analyzed stroke rates from different published studies, including the Anticoagulation and Risk Factors in Atrial Fibrillation (ATRIA) study, the Danish National Patient Registry, the Swedish Atrial Fibrillation cohort, and the Stroke Prevention Using Oral Thrombin Inhibitor in Atrial Fibrillation (SPORTIF) study.
Stroke rate outcomes were measured in quality-adjusted life years (QALYs), and clinical benefit of oral anticoagulation therapy was determined by the difference between the reduction in ischemic stroke risk and the risk of increased bleeding. The CHA2DS2-VASc (congestive heart failure, hypertension, age, diabetes, stroke, and vascular disease) score was used to calculate the difference in risk.
Of 33,434 patients, 81% had a CHA2DS2-VASc score of 2 or more. The benefit of warfarin anticoagulation in the patient population varied almost 4-fold depending on which stroke rates were used: benefit was lowest when using stroke rates from the ATRIA study (6290 QALYs; 95% CI, ± 2.3%; P <.001), and greatest when using the Danish National Patient Registry (24,110 QALYs; 95% CI, ± 1.9%; P <.001).
The optimal CHA2DS2-VASc threshold for maximum population benefit also varied among study stroke rates. Using stroke rates from the ATRIA study showed treating patients with a score of at least 3 would yield the greatest benefit; from the Swedish Atrial Fibrillation cohort, a score of at least 2 yielded the greatest benefit; from the SPORTIF study, a score of at least 1 yielded the greatest benefit; and according to the Danish National Patient Registry, treating all patients (a score of 0 or higher) yielded the greatest population benefit.
Using non-vitamin K antagonist oral anticoagulants (associated with lower risk of intracranial hemorrhage) instead of warfarin anticoagulation decreased optimal CHA2DS2-VASc score thresholds; however, they still varied widely between studies.
One limitation of the study was a high population of adults living in California, so results may not be generalizable to other populations.
The authors calculated the net clinical benefit of warfarin anticoagulation in patients with AF using published stroke rates from different atrial fibrillation studies; the results varied depending on the stroke rate used. While non-vitamin K antagonist oral anticoagulants were associated with greater benefits and lower CHA2DS2-VASc thresholds than warfarin, variation was still significant using the different reported stroke rates.
Future studies should apply standardized and rigorous methods to obtain accurate and reproducible stroke rates.
Shah SJ, Eckman MH, Aspberg S, Go AS, Singer DE. Effect of variation in published stroke rates on the net clinical benefit of anticoagulation for atrial fibrillation [published online September 25, 2018]. Ann Intern Med. doi:10.7326/M17-2762