2014 Volume 78 Issue 8 Pages 1843-1845
Although atrial fibrillation (AF) increases the risk of stroke, this risk is not homogeneous and depends on associated stroke risk factors, which have resulted in clinical scores to aid stroke risk stratification of AF patients. When vitamin K antagonists (VKAs) were the only option for oral anticoagulation, the focus was on the identification of “high risk” patients to be targeted for (inconvenient) VKA treatment. However, the landscape of thromboprophylaxis has changed markedly with the improvements in our understanding of how to use the VKAs (eg, warfarin) by ensuring high time in the therapeutic range (TTR, ≥70%) to provide the best efficacy and safety,1,2 as well as the availability of the non-VKA oral anticoagulants (NOACs, previously referred to as new or novel OACs).
Editorial p 1331
Article p 1481
A recent Editorial by Ikeda3 addressed the question of which of the various clinical scores (CHADS2, R2CHADS2 and CHA2DS2-VASc) should be used for risk stratification for ischemic stroke in patients with AF. Ikeda commented that “a simpler risk stratification may be better because AF is a common disease, and in many cases is managed by general physicians”. Also, he considers the CHA2DS2-VASc score as “very complicated”. Interestingly, the Editorial by Ikeda was a comment on an original article by Hoshino et al,4 which showed that the CHADS2, R2CHADS2 and CHA2DS2-VASc scores are associated with 3-month functional outcomes (based on a modified Rankin scale) of stroke in a small (and underpowered) heterogeneous cohort of patients with prior coronary artery disease (n=139; AF in only 28.8%). Indeed, 11.8% of the patients were being anticoagulated and analyses of functional outcomes were performed by logistic regression rather than time-to-event survival analysis. Thus, the paper by Hoshino et al did not even deal with stroke risk prediction per se in an AF cohort,4 contrary to the editorial focus of Ikeda.2
We are concerned that the Editorial by Ikeda is seriously misleading, and rather misses the point. The predictive values of all clinical-based risk scores such as CHADS2, R2CHADS2 and CHA2DS2-VASc for predicting “high risk” AF patients that develop ischemic stroke have generally been modest (c-statistics approximately 0.6–0.7), but this depends on the study setting and design. In one of the largest validation cohorts,5 CHA2DS2-VASc had a c-statistic of 0.850, which outperformed the CHADS2 score (c-statistic 0.722) in predicting “high-risk” patients who developed thromboembolism; importantly, CHA2DS2-VASc was also better than CHADS2 in defining those patients at low-risk of thromboembolism. The other risk score referred to by Ikeda, the R2CHADS2 score, has many limitations that have been well recognised, notwithstanding its initial derivation from a selected trial anticoagulated cohort that did not include the range of stroke risk and excluded those with severe renal impairment.6 Also, other “real world” studies have shown that adding extra points for “renal impairment” (as with the R2CHADS2 score) did not improve the predictive value of CHA2DS2-VASc.7–9
Nonetheless, various attempts to improve prediction of “high-risk” AF subjects using biomarkers have been proposed, but the (very) marginal improvements in c-statistics over clinical-based scores were at the cost of substantially much lower simplicity and applicability for everyday clinical practice.
If Ikeda assumes that “a simple score is better,” then classifying patients as “low-risk” using the CHADS2 score is simply putting many AF patients at risk of fatal and devastating strokes. In the Danish nationwide cohort study, based on >17,300 patients with a CHADS2 score=0, stroke rates were as high as 3.2% (with the upper boundary of the 95% CI as high as 6.4%) per year.10 Would anyone reasonably withhold anticoagulation therapy for a 74-year-old female patient with AF and prior peripheral artery disease? Such a hypothetical patient would have a CHADS2 score=0 (“low risk”), but a CHA2DS2-VASc score=3 (for which anticoagulation is recommended). A comparison of CHA2DS2-VASc, CHADS2 and the van Walraven scores found that only the CHA2DS2-VASc score was a significant predictor of the absence of thromboembolism for a cohort of lone AF patients followed up over 12 years.11
If CHA2DS2-VASc is “very complicated”, then Ikeda should be reassured that a recent survey of European clinical practice demonstrated uptake and use of this score in 97.7% of respondents whereas the HAS-BLED score was used by >78% for the assessment of bleeding-risk.12 The EORP-AF survey found that anticoagulation rates were approx 80%, especially for CHA2DS2-VASc scores of 2–7.13 Also, CHA2DS2-VASc has been validated in numerous independent cohorts, including those from the Far East14,15 and Japan.16 Among the Far Eastern cohorts the CHA2DS2-VASc score was a significant predictor of thromboembolism among general AF patients, while CHADS2 was not.14,15
Finally, the CHA2DS2-VASc score is now recommended by contemporary major international guidelines from the European Society of Cardiology, Asia Pacific Heart Rhythm Society, 2014 American Heart Association/American College of Cardiology/Heart Rhythm Society and 2014 National Institute for Health and Care Excellence. In a modeling analysis among European AF patients, appropriate use of the CHA2DS2-VASc score would translate into the prevention of over 60,000 major cardiovascular events and deaths each year among patients with a CHA2DS2-VASc score ≥2, resulting in a profound annual mathematical net clinical benefit on stroke and major bleeds.17 A similar effect has been noted in a modeling exercise among Chinese AF patients, given the improved net clinical benefit from NOAC use.18
In conclusion, we remain rather concerned by misleading or erroneous approaches to using stroke risk scores, despite clear recommendations advocated in guidelines. Defining patients as “low risk” using the CHADS2 score and not treating appropriately is simply exposing many AF patients to fatal and devastating strokes. As recommended in current major guidelines, the CHA2DS2-VASc score should be used to identify the “truly low-risk patients” with AF, as the first decision-making step. These “low-risk” patients with AF, who are “aged <65 and lone AF (irrespective of gender)” – essentially a CHA2DS2-VASc score=0 (males) or 1 (females) – do not need any antithrombotic therapy. Subsequently, patients with AF and ≥1 additional stroke risk factor can be offered effective stroke prevention, which is OAC, whether delivered as well-controlled adjusted-dose warfarin (TTR ≥70%) or one of the NOACs.1 A (very) simplified approach to decision-making for thromboprophylaxis in AF is provided in Figure. Thus, what we really need is improved implementation and guideline-adherence – and not misinformed opinion.
Simplified approach to decision-making for thromboprophylaxis in atrial fibrillation patients.
Conflicts of Interest: G.Y.H.L. has served as a consultant for Bayer Healthcare, Astellas, Merck, AstraZeneca, Sanofi Aventis, Bristol Myers Squibb/Pfizer and Boehringer Ingelheim, and has been on the speaker bureaus for Bayer, Bristol Myers Squibb/Pfizer, Boehringer Ingelheim and Sanofi Aventis and has received travel expenses from these companies. D.A.L. has received investigator-initiated educational grants from Bayer Healthcare and Boehringer Ingelheim. She has been on the speaker bureau for Boehringer Ingelheim and Bristol Myers Squibb/Pfizer and has received travel expenses from Boehringer Ingelheim.
Both G.Y.H.L. and D.A.L. are co-authors of the CHA2DS2-VASc score.
