Abstract 18538: The ATRIA Stroke Risk Score Predicts Ischemic Stroke Better than CHADS2 and CHA2DS2-VASc in a large Swedish Cohort of Patients with Atrial Fibrillation
Introduction: Atrial fibrillation (AF) is a major risk factor for acute ischemic stroke (AIS). Anticoagulation therapy (OAC) effectively prevents AIS, but increases bleeding risk. There is a need for better AIS risk prediction to optimize the anticoagulation decision in AF. The ATRIA stroke risk score (ATRIA) (table) was superior to CHADS2 and CHA2DS2-VASc in two large California community AF cohorts. We now report the performance of the 3 scores in a very large Swedish AF cohort.
Methods: The cohort consisted of all Swedish patients hospitalized with a diagnosis of AF from July 1, 2005 to December 31, 2008. Predictor variables and the outcome, AIS, were obtained from inpatient ICD-10 codes. Warfarin use was determined from National Pharmacy Database. Risk scores were assessed via c-index (C) and net reclassification index (NRI).
Results: The cohort included 158,370 AF patients off warfarin who contributed 340,332 person-years of follow-up, and 11,823 incident AIS, for an overall AIS rate of 3.47%/yr, higher than the 2%/yr seen in the California cohorts. Using the entire point score, ATRIA had a good C of 0.712 (0.708-0.716), significantly better than CHADS2, 0.694 (0.689-0.698), or CHA2DS2-VASc, 0.697 (0.693-0.702). Using published cut-points for Low/Moderate/High AIS risk, C deteriorated for all scores but ATRIA and CHADS2 were superior to CHA2DS2-VASc. NRI favored ATRIA; 0.16 (0.15-0.18) versus CHADS2; 0.22 (0.21-0.24) versus CHA2DS2-VASc. However, NRI decreased to near-zero when cut-points were altered to better fit the cohort’s stroke rates.
Conclusion: Findings in this large Swedish AF cohort validate those in the California AF cohorts, with the ATRIA score predicting stroke risk better than CHADS2 or CHA2DS2-VASc. However, relative performance of the categorical scores varied by population stroke risk. Knowledge about this population risk may be needed to optimize cut-points on the multipoint scores to achieve better net clinical benefit from OAC.
Author Disclosures: S. Aspberg: None. Y. Chang: None. D. Singer: Research Grant; Significant; Bristol-Myer Squibb, Johnson and Johnson. Consultant/Advisory Board; Modest; Bayer Healthcare, Boehringer Ingelheim, Daiichi Sankyo, Merck, Pfizer. Consultant/Advisory Board; Significant; Johnson and Johnson, Bristol-Myers Squibb.
- © 2014 by American Heart Association, Inc.