58 CHAPTER 4 TO THE EDITOR We read the recent publication by Shah and colleagues with great interest.1 Using data from 4 population-based studies and a rigorous modelling approach, the authors demonstrated the uncertainty of defining the optimal threshold for the CHA2DS2-VASc score above which anticoagulant treatment for stroke prevention in atrial fibrillation (AF) should be initiated. The optimal CHA2DS2-VASc threshold varied considerably, ranging from 0 points (i.e. anticoagulation for all patients with AF) to 3 points, depending on the predicted stroke risk for each score. In our recent meta-analysis of all 19 studies validating the CHA2DS2-VASc rule in patients with AF who were not receiving anticoagulation, we found, similar to Shah and colleagues, that it is not possible to adequately decide about anticoagulation in individual patients with AF on the basis of the CHA2DS2-VASc rule. 2 To illustrate this, we calculated 95% prediction intervals (PI) indicating the range of expected stroke rates for any patient who presents with AF similar to those included in the meta-analysis. For a CHA2DS2-VASc score of 2, the 95% PI for the annual risk for stroke ranged from 0.4% to 3.3% in community-dwelling patients with AF. Of interest, variation was more pronounced in studies recruiting patients with AF from the hospital setting: For a CHA2DS2-VASc score of 2, the 95% PI for the annual risk for stroke ranged from 0.03% to 7.8%. Disease burden in AF is insufficiently captured by the CHA2DS2-VASc rule. In addition to comorbidity, its severity (such as the type of heart failure or the extent of hypertension or diabetes control) and other already known predictors of stroke in AF (such as renal insufficiency) should be considered. Therefore, we believe the difference between settings of care as observed in our study further strengthens the work by Shah and colleagues. More accurate stroke prediction is needed for safe and effective anticoagulation in patients with AF. Therefore, future research should focus on model revision. Until then, we agree with Shah and colleagues that guidelines on anticoagulation in AF should include the uncertainty around the current thresholds for anticoagulation and the associated predicted stroke risks per CHA2DS2-VASc score.
RkJQdWJsaXNoZXIy MTk4NDMw