Response to: Current opinion of the ESC Working Group on Cardiovascular Pharmacotherapy and ESC Council on Stroke

We thank Nielsen et al.1 for their critical appraisal of the current opinion statement of the ESC Working Group on Cardiovascular Pharmacotherapy and ESC Council on Stroke. The authors highlighted an important aspect in their analysis of data obtained between 1998 and 2014 by Danish nationwide registry: patients with atrial fibrillation (AF) presenting with a CHA2DS2-VASc score of 1 and HAS-BLED score equal or above 2 hold a higher thromboembolic risk as their HAS-BLED <2 counterparts. However, those findings seem intuitive, considering that a large scale of patient characteristics that proved to be associated with the risk of bleeding (e.g. kidney function, age, or hypertension) also strongly influence the patients’ individual risk of ischaemic events.2–6 As a consequence, there should be very viewed patients with a high HAS-BLED score but a low CHA2DS2-VASc score. Indeed, an elevated HAS-BLED score mirrors a valid surrogate parameter for a sicker AF patient population with an elevated risk for both stoke and systemic embolism, but a disproportionately higher risk for major bleeding events. This issue has already been widely discussed within the present current opinion statement of the ESC Working Group on Cardiovascular Pharmacotherapy and ESC Council on Stroke.6 In the analysis presented by Nielsen et al. it would be of interest to know the individual risk factors and their association with thromboembolic events and bleeding. As outlined, caregivers need to consider that the risk of bleeding should be carefully weighed against the individual ischaemic risk to achieve a net clinical benefit of oral anticoagulation (OAC) with regard to stroke prevention. When considering OAC in this intermediate-risk AF patient with a HAS-BLED score ≥2 we need to bear in mind that any increase of the patients’ risk of major bleeding ultimately leads to imbalance in the treatment benefit associated with OAC. For example, it seems obvious, that a 66 years old, physically active male patient with AF and concomitant poor liver synthesis function has a higher risk for intracerebral bleeding as compared to stroke prevention in case of OAC. Therefore, a general recommendation of OAC therapy in these individuals would weaken the net clinical benefit due to an increased risk of bleeding events—as outlined within the current consensus statement. Another important aspect outlined by Nielsen et al. covers the interpretation of presented event rates in AF patients with a CHA2DS2-VASc score of 1. Within the current opinion statement of the ESC Working Group on Cardiovascular Pharmacotherapy and ESC Council on Stroke, event rates for both CHA2DS2-VASc score and HAS-BLED score categories were summarized and presented in a uniform manner as event rates per year to allow a proper comparability of both ischaemic and bleeding rates that allows to draw conclusions on the treatment threshold.2–4 Considering annual bleeding rates of 1.88–3.20% per year in patients presenting with a HAS-BLED score of 2, the annual risk of bleeding outweighs the thromboembolic risk associated with a CHA2DS2-VASc score of 1, which amounts only 0.6–1.3% per year. Therefore, the ESC Working Group on Cardiovascular Pharmacotherapy and ESC Council on Stroke feel that OAC should not be considered in intermediate thromboembolic risk patients with a HAS-BLED score ≥2 since the risk of bleeding clearly outweighs the thromboembolic risk in this patient population (as validated in several highly powered independent cohorts).7–10 We agree with the authors, that the interpretation and comparison of event rates between several investigations and analysis available in literature need to be undertaken with caution. Nielsen et al. would have the unique chance to overcome the limitation of varying risk by assessing the net benefit in one cohort. While the authors of the letter focus on the influence of competing risk on the calculation of risk they do not mention several alternative explanations for varying risk between registries such as the imprecise definition of risk factors or the definition and weighing of endpoints. When interpreting data of population-based healthcare-system associated registries as presented by Nielsen et al. one also needs to consider the inherent limitations. It remains unclear how the authors adequately calculated the HAS-BLED score for their study population, since both a labile international normalized ration and uncontrolled hypertension with a systolic blood pressure <160 mmHg are not assessable with this study design. Therefore, the presented event rates and absolute risk calculations might be misleading when considering the issues mentioned both above and within the present consensus statement. Surprisingly, in the presented analysis the risk of thromboembolic events in patients with a CHA2DS2-Vasc score of 1 exceeds 1% per year in almost all subgroup independently of the kind of analysis. This may point to a relevant number of unidentified risk factors. Even though randomized trials on OAC in patients with low CHA2DS2-VASc scores would provide a solid evidence base for clinical practice guidelines, their conduction appears most improbable when considering the low number of eligible patients suitable for study participation and the associated event rates. We agree with the authors that the currently available level of evidence remains therefore low, but we need to rely on those—mainly highly powered—observational investigations to guide caregivers through this challenging approach of identifying AF patient with increased ischaemic risk and doing no harm by considering unjustified OAC. In conclusion, the basis for the decision of OAC is a careful individual assessment of thromboembolic and bleeding risk as outlined in our current opinion. Conflict of interest: P.S. reports grants from Daiichi Sankyo, Boehringer-Ingelheim, outside the submitted work; A.