Oral anticoagulation in Atrial Fibrillation with CHA2DS2-VASc = 1: too much uncertainty?

Fernando Botto –


Patients with any form of Atrial Fibrillation (AF) having an estimated thromboembolic risk (stroke or systemic embolism) of 1% or more per year should be considered for Oral Anticoagulation (OAC) after a balance with the bleeding risk.

Currently, clinical guidelines recommend thromboembolic risk evaluation using the CHA2DS2-VASc score and bleeding risk using the HAS-BLED score. Both have emerged from “real world data”, in general from retrospective research designs using administrative databases or registries. This type of design usually have many limitations, such as bias in pharmaceutical agents indication, known and unknown confounders, heterogeneous definitions, populations with different baseline characteristics, variable follow-up and event rates, etc. Some may be controlled through statistical tools (multiple logistic regression, propensity scores, stratification, etc) although adjustment frequently is not enough.

CHA2DS2-VASc validations have showed a modest discrimination with original C-index or AUC ROC of 0.61 and later 0.67, while HAS-BLED original AUC was 0.72 and later it was 0.66 or even lower. We should consider that flipping a coin has a chance of 0.50 of landing heads up, therefore an AUC value of 0.70 or higher is required to determine a “good” discrimination. Undoubtedly, the clinical tools that are available challenge decision making, especially because the evidence quality and methodology is far from perfect, but we are prompt to use them.

Regarding CHA2DS2-VASc, its main strength is probably the high negative predictive value (NPV), since 0 points (without counting female sex) have demonstrated repeatedly <0.5% per year of thromboembolic risk in different cohorts. Therefore, this risk will always be less than bleeding risk, and OAC would not be indicated.

On the contrary, CHA2DS2-VASc = 1 (without counting female sex) positive predictive value (PPV) may be quite variable. Proposed predictors have not the same prognostic value in different studies. In general, to reach 1 point the most prevalent variable is usually age between 65 and 74 years followed by HTN (in subjects <65 years). Regarding its prognostic value for stroke, the hazard ratio fluctuates between 1.7 and 3 according to the observed variable and the individual weight within the score also fluctuates depending on the study considered. Additionally, the variability increases even more since it is not the same controlled or uncontrolled HTN, stable or unstable congestive heart failure (CHF), systolic or diastolic CHF, symptomatic or asymptomatic CHF, and type I or II diabetes.

All these uncertainties impact, especially, in CHA2DS2-VASc = 1, because it may stands for a low or a high risk in the “individual” patient. Classically, it was assigned a thromboembolic risk between 0.6% and 1.5% per year, however, recent studies in patients without antithrombotic agents have reported a stroke rate between 1.46% and 1.55% per year, respectively, in women and men from Sweden, that increases to 2.55% and 2.75% per year, respectively, in women and men from Taiwan, and that reaches an average of 6.6% per year in China.

There exists also methodological limitations in different studies evaluating the HAS-BLED score. Zero points means an approximate incidence of bleeding between 0.6 to 1.13% per year; 1 point, 1 to 1.5% per year; 2 points, 1.88 to 3.2% per year; and 3 points may oscillate from 3.74% to 21% per year or more.

In conclusion, my viewpoint about OAC indication in CHA2DS2-VASc < 2 (without counting female sex) might be:

  • CHA2DS2-VASc = 0 should not receive OAC (with any HAS-BLED score result)
  • CHA2DS2-VASc = 1 + HAS-BLED ≥ 3 should not receive OAC due to an estimated bleeding risk higher than thromboembolic risk
  • CHA2DS2-VASc = 1 + HAS-BLED = 0 should receive OAC due to an estimated thromboembolic risk higher than bleeding risk
  • CHA2DS2-VASc = 1 + HAS-BLED = 1 the thromboembolic and bleeding risks are balanced, therefore we should “personalize” OAC indication (see “other risk factors…” below). In these cases we should favored OAC indication, preferably a DOAC (see last paragraph)
  • CHA2DS2-VASc = 1 + HAS-BLED = 2 the thromboembolic and bleeding risks may result balanced, therefore we should “personalize” OAC indication, but here we must be cautious because bleeding risk might be higher. So, we have to analyze the presence of other risk factors not included in the proposed scores: smoking, obesity and very low weight vs athletic appearance, cognitive disorders, chronic renal dysfunction, rheumatoid arthritis, cancer and low treatment adherence. From complementary evaluations: increase in volume (≥73 ml) or size (≥ 47 mm) of left atrial, empty velocity of left appendage <20 cm/s, and elevated burden of AF (persistent or permanent AF vs paroxysmal have higher stroke risk). Finally, we should consider elevation of high sensitivity troponin and NT-proBNP.
  • In regards to the OAC agent selection, it would be ideal a DOAC because all the randomized clinical trials have demonstrated a lower incidence of intracranial bleeding than Warfarin and major bleeding were predominantly gastro-intestinal, with a lower impact on mortality. Anyway, in this uncertain scenario the physician must inform the patient adequately to allow a shared decision medicine, expressing all her/his preferences and expectations. Finally, in those cases which Warfarin was decided as the OAC, it will be only justified if TTR is ≥70%.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s