Physicians' Academy for Cardiovascular Education

Modified CHADS2 shows best predictive performance for stroke across spectrum of kidney function

Validation of risk scores for ischaemic stroke in atrial fibrillation across the spectrumof kidney function

Literature - De Jong Y et al. - Eur Heart J. 2021 Mar 26;ehab059. doi: 10.1093/eurheartj/ehab059.

Introduction and methods

The estimated prevalence of CKD is 10-15% in de general population and is increasing [1]. CKD is associated with an increased risk of ischemic stroke (IS) [2]. The increasing prevalence of CKD can thereby partly explain an increase in prevalence of IS [3,4]. For personalized anticoagulation therapy, risks scores for IS are important to estimate the risk of IS vs the risk of treatment-related bleeding. However, the predictive performance of commonly used risk scores for IS in patients with CKD is unclear. This study validated six risk scores for IS in patients with AF across the spectrum of kidney function.

This study used data from 36004 subjects with new-onset AF from the Stockholm CREAtinine Measurements (SCREAM) project [5]. The study outcome was hospitalization for IS or IS as main cause of death. The following risk scores were validated: AFI [6], CHADS2 [7], Modified CHADS2 [8], CHA2DS2-VASc [9], ATRIA [10], and GARFIELD-AF [11]. The predictive performance of these risk scores was evaluated by discrimination and calibration ability across three categories kidney function: Normal kidney function (eGFR >60mL/min/1.73 m²), mild CKD (eGFR 30–60mL/min/1.73 m²), and advanced CKD (eGFR <30mL/min/1.73 m²). Discrimination was assessed by c-statistic. The c-statistic lies between 0.5 and 1.0 and reflects the ability of the risk score to distinguish between patients with and without the outcome. A c-statistic <0.7 is considered poor to moderate, 0.8 good, and >0.9 excellent. Calibration ability reflects the agreement between the predicted and actual observed probabilities of the outcome.

Main results


Most studied risk scores showed moderate to good discrimination in AF patients with normal eGFR, but discrimination decreased in patients with mild or advanced CKD. Calibration was largely independent of eGFR. The Modified CHADS2 score showed good performance for discrimination and calibration in all three kidney function categories. The authors of the article therefore suggest that the Modified CHADS2 score is the preferred risk score in clinical practice.


Show references

Find this article online at Eur Heart J.

Share this page with your colleagues and friends: