The CHA2DS2-VASc Score: Are All Components Equal?

March 9, 2016

The components may be equal on paper, but they may not be treated equally in clinical practice, according to new research.

The introduction of the simple-to-remember CHA2DS2-VASc score has created a clinically handy, easy-to-use risk stratification system to assess thromboembolic risk in non-valvular atrial fibrillation (AF), thus eligibility to receive oral anticoagulation (OAC). In assessing clinical risk, each component of the scoring system is weighted equally (with the exception of prior stroke/TIA and age >75, which are weighted twice as much as the other components).  Practicing clinicians, however, recognize that there may be some heterogeneity in risk and also in the weight we each give personally to components of the score. For example, a young diabetic patient with a prior TIA/stroke and severely reduced LVEF and a 77-year- old with hypertension and non-obstructive coronary artery disease may have the same calculated CHA2DS2-VASc score but may each carry a different absolute risk.

A recent analysis, led by researchers at the University of Colorado, utilized a large outpatient registry (the National Cardiovascular Data Registry [NCDR]) to determine if there were any differences in the weight assigned to individual components of the CHA2DS2- VASc score with respect to provider decision to initiate oral anticoagulation (OAC). Of the 706,308 patients identified with non-valvular AF (mean age 74y, relatively equal male:female ratio, mostly white) that had an indication for OAC use (CHA2DS2-VASc ≥2) 80% had hypertension and 84% were age ≥65y; half had CAD (51%) with only a minority having CHF (26%) or DM (24%) and even fewer having PAD (9%) or prior stroke/TIA (4%); Yet, only 12.4% to 18% received warfarin and only 35.8% to 47.2% received a novel OAC, leaving nearly one-third who didn’t receive OAC despite meeting guideline-based indications for therapy.

In order to minimize confounding, the authors meticulously adjusted their multivariable logistic regression models for demographic factors, clinical factors, as well as modified HASBLED scores and accounted for clustering by hospital location and by provider. The results indicated that, in fully adjusted models, lower OAC use as associated with female gender (OR 0.79) and vascular disease (OR 0.91) whereas higher OAC use was associated with older age, (>65 y [OR 3.35-3.36]), HTN (OR 2.57), DM (OR 1.17), CHF (OR 1.31), and stroke/TIA (OR 1.32). 

These results may indeed reflect some of the ongoing gender disparities seen in cardiovascular care, with guideline-based therapy being prescribed less often to women with heart disease, despite their higher risk for thromboembolic events compared with men. The finding of lower use of OAC in the presence of vascular disease, on the other hand, remains a mystery. Full methodologic details of the study are not available to determine the individual clinical factors that were adjusted for but these findings may reflect clinician hesitation to use OAC in conjunction with antiplatelet therapies (ie, ASA, clopidogrel, prasugrel, ticagrelor), which many patients with vascular disease are likely to be taking.

There are no published data that indicate a greater or lesser contribution to thromboembolic risk by any individual component of the CHA2DS2-VASc score, but these findings suggest that there is indeed heterogeneity at work in clinical decisions to initiate anticoagulation in guideline-eligible patients with AF. Study authors note that the results may reflect physician bias, patient preference, or under-recognition of risk factors. Additional research is required to better understand how each component of the score may influence prescribing practices and the ultimate impact these differences may have on clinical care.  


Thompson LE, Maddox TM, Lei L, et al. Impact of CHA2DS2-VASc risk factors on anticoagulant prescription in patients with atrial fibrillation: insights from the NCDR PINNACLE Registry. Circulation: Cardiovascular Quality and Outcomes. 2016; 9: A22.