Are We Underestimating Stroke Risk With CHADS2?

Written By

Katie Kiser, Pharm.D., BCPS

Reviewed By

Dave Dixon, Pharm.D., BCPS, CDE, CPP
Gloria Grice, Pharm.D., BCPS


Lip GYH, Nieuwlaat R, Pisters R, Lane DA, Crijns HJGM. Refining Clinical Risk Stratification for Predicting Stroke and Thromboembolism in Atrial Fibrillation Using a Novel Risk Factor-Based Approach: The Euro Heart Survey on Atrial Fibrillation. Chest 2010; 137: 263-72.

For patients with atrial fibrillation, what’s the best scoring system to estimate stroke risk? Clinicians have long relied on the CHADS2 [congestive heart failure, hypertension, age ≥ 75 years, diabetes, and previous stroke or TIA] scoring system to make antithrombotic treatment decisions. The CHADS2 scoring system is validated and has good predictive value.1 Current treatment guidelines recommend the CHADS2 scoring system2 and it has been widely adopted by clinicians because of its simplicity and low cost. But the CHADS2 scoring system isn’t perfect and there has been continued work to develop a better scoring system over the past several years. Would an improved scoring system provide clinicians and patients a more accurate assessment of thromboembolic risk? Would it change antithrombotic treatment decisions? Would it decrease the use of unnecessary treatments?

One potential alternative to the CHADS2 is the recently developed CHA2DS2-VASc scoring system.3 The CHA2DS2-VASc scoring system includes all of the original CHADS2 risk factors plus vascular disease, age 65-74 years, and gender (female). Each risk factor receives one point except age ≥ 75 and previous stroke / TIA which both receive two points. The CHA2DS2-VASc scoring system was developed from an analysis of 1,084 participants in the Euro Heart Survey on Atrial Fibrillation study. While none of the subjects in this study were receiving heparin or vitamin k antagonist therapy, seventy-four percent were on antiplatelet therapy at baseline. The table below describes the percent of participants in each risk category and the number (%) who experienced a thromboembolic event (defined as composite of ischemic stroke, pulmonary embolism, or peripheral artery embolism).

Table 1 – Risk Classification and Thromboembolic Events Using CHADS2 vs. CHA2DS2-VASc Scoring System (Euro Heart Survey on Atrial Fibrillation Study)

Scoring System

Low Risk

Intermediate Risk

High Risk

CHADS2 score



2 - 6

% in category




TE events, No. (%)

3 (1.4)

7 (1.9)

15 (3.1)

CHA2DS2-VASc score



2 - 6

% in category




TE events, No. (%)

0 (0)

1 (0.6)

24 (3)

The CHA2DS2-VASc scoring system classified most of the subjects as high risk; where-as the CHADS2 placed more patients in the low or intermediate risk categories. Of the risk factors included in the CHA2DS2-VASc scoring system, female gender was the only factor that had significant predictive power (OR 2.53 [1.08-5.92]).

One potential caveat to this study is the definition of a thromboembolic event. Clinicians currently do not typically use an anticoagulant in the setting of atrial fibrillation to prevent pulmonary embolism. Including additional thromboembolic events in the analysis skews the results in favor of the CHA2DS2-VASc scoring system as the CHADS2 was not developed to predict pulmonary embolism. One could argue that the risk for these additional events SHOULD be considered.

Obviously, more patients would be candidates for anticoagulant therapy if we adopt the CHA2DS2-VASc scoring system. This study did not examine whether increased anticoagulant use would result in increased rates of major bleeding.

Does the CHA2DS2-VASc really work better in practice? It’s one thing to develop and validate a risk scoring system in a relatively small subset of patients in a specific clinical trial – but it’s important to validate the tool in a large, robust sample that represents the breadth of patients we encounter in practice. In a registry-based cohort study conducted in Denmark, researchers measured the accuracy of each of these two scoring systems (CHADS2 and CHA2DS2-VASc) to predict stroke and thromboembolism in patients admitted to the hospital with non-valvular atrial fibrillation who had not received vitamin k antagonist or heparin therapy prior to admission.4 There were 73,538 subjects who met these inclusion criteria. The table below shows the event rate (95% CI) at hospital admission or death due to thromboembolism (e.g., peripheral artery embolism, ischemic stroke, and pulmonary embolism) based on risk category by scoring system.

Table 2 – Thromboembolic Event Rate Based on Risk Classification Using CHADS2 vs. CHA2DS2-VASc Scoring System (Danish Validation Study)

Risk category (score)

1 year Follow-up

5 year Follow-up

10 year Follow-up


Low Risk (0)

1.67 (1.47 to 1.89)

1.28 (1.19 to 1.38)

1.24 (1.16 to 1.33)

Intermediate Risk(1)

4.75 (4.45 to 5.07)

3.70 (3.55 to 3.86)

3.56 (3.42 to 3.7)

High Risk (2-6)

12.2 (11.8 to 12.7)

8.3 (8.0 to 8.5)

8.0 (7.7 to 8.1)


Low Risk(0)

0.78 (0.58 to 1.04)

0.69 (0.59 to 0.81)

0.66 (0.57 to 0.76)

Intermediate Risk(1)

2.01 (1.70 to 2.36)

1.51 (1.37 to 1.67)

1.45 (1.32 to 1.58)

High Risk(2-9)

8.8 (8.5 to 9.1)

6.0 (5.9 to 6.1)

5.7 (5.6 to 5.8)

Again, the CHA2DS2-VASc placed far more subjects in the high risk category (80.3%) when compared to CHADS2 (45.5%). Fewer patients were considered intermediate (11.2% vs. 32.3%) or low risk (8.7% vs. 22.3%) with CHA2DS2-VASc vs. CHADS2. The two scoring systems performed similarly in their ability to predict who would have an embolic event; however, the CHA2DS2-VASc was more accurate in terms of placing patients in the low, intermediate, and high risk categories. In other words, the CHA2DS2-VASc was better at identifying patients who were truly at low risk and those at high risk (generally considered to be an annual risk of thromboembolism > 6%). Interestingly, it appears that the risk of thromboembolism did not increase over the 10 years follow-up period as would have been anticipated (due to increasing age). Again, some would argue that it is difficult to draw conclusions regarding this data due to the inclusion of pulmonary embolism in the endpoint. The authors state that the results were similar when pulmonary embolism was excluded. However, they do not report a detailed table to support this statement in the manuscript.

So, should we change the way that we risk stratify patients with atrial fibrillation? I think not … at least not yet. We need a prospective study to show the potential harms (and benefits) of placing more patients on anticoagulation therapies. With the advent of new oral anticoagulants, we now have alternatives to vitamin k antagonist therapy. These new therapies provide some POTENTIAL advantages including safety, efficacy, monitoring, and patient satisfaction. Thus, a new classification system may help us to identify MORE patients who would benefit from anticoagulation therapies. Time will tell.

While I favor continued use of the CHADS2 scoring system, I think the CHA2DS2-VASc scoring system may be useful in patients who are intermediate risk based on their CHADS2 score. The risk of thromboembolism in those with a CHADS2 score of 1 is not inconsequential – perhaps as high as 5% in the first year! In this circumstance, I think it would be appropriate to use the CHA2DS2-VASc scoring system to help determine whether antiplatelet or anticoagulant therapy would be more appropriate. If the patient is re-classified as high risk using the CHA2DS2-VASc scoring system, I’d feel comfortable recommending anticoagulation therapy. If the score remained in the intermediate category, I think antiplatelet therapy is a reasonable alternative.

Have you adopted the use of CHA2DS2-VASc scoring system in your practice? Should the next CHEST guidelines recommend using CHA2DS2-VASc over the CHADS2? Or should we develop a stratification system that tells us which patients should be on dabigatran and which should be on warfarin? What are your thoughts?



<p>I agree that for patients who are already determined high risk with CHADS<sub>2</sub>, then the CHA2DS2-VASc really will not provide any additional information that will change clinical decisions. One of our residents informally looked at a few of our patients in our anticoag clinic, and noticed it was the intermediate risk patients from CHADS2 that sometimes changed risk categories and were higher risk with CHA2DS2-VASc.</p> <p>Personally,&nbsp;CHADS2 is quicker and easier to do - and if a patient is high risk...I'll stop there. But, if the patient is low or intermediate risk with CHADS2 - I think I'll calculate CHA2DS2-VASc and then use my clinical judgement....</p>

<p>I have not begun using CHA2DS2-VASc, primarily because our clinic patients have already been stratified into&nbsp;a higher risk group by virtue of being placed on warfarin and they would not change to a lower risk&nbsp;by use of the new tool.&nbsp; The main place this could be useful is when considering bridge therapy risk stratification.&nbsp; However, since this is generally a short duration and the&nbsp;"daily"&nbsp;risk of stroke is not&nbsp;dramatically different between the&nbsp;two stratification schemes I will stick to CHADS2 for now.&nbsp; I am curious to see how newer therapies with potentially less bleeding risk will change risk stratification and thus antithrombotic therapy as the original stratifications have more to do with balancing bleeding and thromboembolic risk.</p>