What is the stroke risk threshold for anticoagulation to provide net benefit in atrial fibrillation? A Markov decision model analysis

Abstract Background Randomized trials have clearly demonstrated that oral anticoagulation is effective in preventing strokes in atrial fibrillation (AF) patients at high stroke risk. However, in lower-risk patients, the relative reduction in stroke risk with anticoagulation has appeared to be smaller. Current knowledge of the net benefit of anticoagulation in these low- to moderate-risk patients relies mainly on observational studies, which are inherently confounded when assessing treatment effects. The exact tipping point in the non-anticoagulated stroke risk, above which stroke reduction with anticoagulation outweighs the increased bleeding risks, is unknown. Purpose We aimed to estimate the threshold of the non-anticoagulated stroke risk in AF patients, above which non-vitamin K antagonist oral anticoagulant (NOAC) therapy offers a net benefit in terms of quality-adjusted life years (QALYs). Methods A Markov decision model was developed to determine the net impact of NOAC therapy on the QALYs of AF patients across a range of non-anticoagulated stroke risks, using a hypothetical base-case patient aged 70 years. The model incorporated trial-based data on the effects of NOACs on stroke and bleeding outcomes at different stroke risk levels, along with prior evidence on their impact on mortality and quality of life. The model considered that the initial non-anticoagulated bleeding and mortality rates are higher in patients with higher stroke risk. Additionally, the effect of anticoagulation on the severity of stroke and bleeding events was incorporated into the analysis (Table 1). The model considered ischemic stroke, hemorrhagic stroke, other intracranial and extracranial major bleeding events, as well as mortality. Cumulative QALYs over a 20-year simulation were calculated for 10,000 iterations across the annual stroke risk range of 0% to 10%, with and without NOAC therapy. Results The decision to initiate NOAC therapy resulted in higher cumulative QALYs than withholding treatment when the annual non-anticoagulated stroke risk exceeded 0.65% (Figure 1). Average incremental QALYs per patient in the 20-year simulation with NOAC therapy at non-anticoagulated annual stroke risk levels of 1%, 2%, and 5% were 0.11, 0.20, and 0.38, respectively. At a stroke risk level of 0.5%, NOAC therapy resulted in a reduction of 0.01 QALYs per patient over 20 years. When the untreated stroke risk levels were correlated with the CHA2DS2-VA score in patients with clinical AF, our model indicated that NOAC therapy provides a net benefit for patients with a CHA2DS2-VA score of 1 point or higher. Conclusions This Markov decision modeling study estimated that NOAC therapy leads to a net benefit in patients with AF whose non-anticoagulated annual stroke risk exceeds 0.65%. This suggests that NOAC therapy could be offered to moderate-risk patients with clinical AF and a CHA2DS2-VA score of 1.Markov model input parameters