Can Risk Score Model Predict Stroke Recurrence in Stroke Survivors With Atrial Fibrillation?

atrial fibrillation diagnosis
atrial fibrillation diagnosis
Researchers sought to develop and validate a comprehensive risk prediction model for stroke recurrence in patients with acute ischemic stroke and atrial fibrillation.

Conventional risk stratification tools and a newly developed risk score are not suitable for predicting the risk for recurrent stroke among stroke survivors with atrial fibrillation (AF), according to study results published in PLoS One.

Several risk stratification models are widely used to identify patients at high risk for recurrent ischemic stroke among stroke survivors with AF, including the CHADS2 score, CHA2DS2-VASc score, and ATRIA score. In light of the limitations of these tools, a new risk model was developed to predict stroke recurrence in stroke survivors with AF.

Using data from multicenter registries in South Korea (Clinical Research Collaboration for Stroke in Korea) and Japan (Stroke Acute Management with Urgent Risk factor Assessment and Improvement), the researchers identified patients with ischemic stroke and nonvalvular AF.

The developmental dataset included 4483 patients from South Korea and 1165 patients from Japan (mean age, 74 years; 53.1% male) with acute ischemic stroke with nonvalvular AF, who were hospitalized between 2011 and 2014. The external validation dataset included 3668 patients who were hospitalized between 2015 and 2018 and were registered in the Clinical Research Collaboration for Stroke in Korea. The clinical profiles of the external validation were generally comparable to those of Korean patients in the developmental dataset.

Functional outcomes were modified Rankin Scale scores at 3 months and at 1 year after the index stroke. Recurrent stroke, myocardial infarction, and death for up to 1 year were collected as event outcomes.

Of the whole population, 338 (6.0%) patients were diagnosed with recurrent stroke, including 252 (5.6%) patients from South Korea and 86 (7.4%) patients from Japan. Mortality rates at 1 year were 16% in the whole population, including 774 (17.3%) patients from South Korea and 129 (11.1%) patients from Japan.

The prediction model, incorporating the appropriately transformed variables and significant interaction terms, underwent internal validation through 999 bootstrap samples with further calibration and revision through the external validation dataset.

Neither the conventional risk scores – CHADS2, CHA2DS2-VASc, and ATRIA scores – nor the newly developed model showed a consistent dose-dependent relationship. The new risk score did not show better discriminative ability for predicting stroke recurrence among stroke survivors with AF, compared with the conventional risk prediction tools, as the discrimination ability of the risk score was modest, with c-statistics of 0.68 (95% CI, 0.66-0.71).

The use of data on Korean and Japanese patients may limit the generalizability of the findings. Furthermore, another potential limitation may be secondary to the significant increase in the proportion of direct oral anticoagulants use from the developmental set to the external validation.

“The newly developed model showed only modest utility in discriminating the risk of recurrence, similar to the conventional risk scores (ATRIA, CHADS2, and CHA2DS2-VASc scores). Detailed individual information, including brain imaging, serum biomarkers, and cardiac function, may be needed to build a more robust and precise risk prediction model,” concluded the researchers.

Disclosure: This research was supported by Bristol-Myers Squibb Korea. Please see the original reference for a full list of disclosures


Kim BJ, Lee KJ, Park EL, et al. Prediction of recurrent stroke among ischemic stroke patients with atrial fibrillation: Development and validation of a risk score model. PLoS One. Published online October 8, 2021. doi: 10.1371/journal.pone.0258377