Novel composite score to predict atrial Fibrillation in acute stroke patients: AF predicting score in acute stroke.

OBJECTIVE

Identification of high risk population for atrial fibrillation among acute stroke patients is a center of attention. The objective of the present study was to construct a model that can predict the presence of atrial fibrillation in ischemic stroke patients and to validate the model.

METHODS

From a prospectively collected hospital-based stroke registry participated by two hospital, we selected data of patients who were admitted within 24 h after the onset of symptoms. Using a dataset of 1355 acute ischemic stroke patients, a model to predict the presence of atrial fibrillation was constructed and the probability of the presence of atrial fibrillation (AF-probability) was calculated. The patients were classified into low-risk, moderate-risk, and high-risk groups according to AF-probability. The performance of the model to predict atrial fibrillation among acute stroke patients was investigated and validated.

RESULTS

Seven factors were selected as constituents of the model including age, left atrial size, free fatty acid level, triglyceride level, susceptibility vessel sign, hemorrhagic transformation, and cortical involvement. The performance of the model was excellent, with a C-statistic of 0.908 (95% confidence interval 0.887-0.930). According to risk group, true positivity for atrial fibrillation was 4.3%, 36.5%, 91.2% in the low-risk, moderate-risk, and high-risk groups, respectively. The internal and external validation test showed stable consistency of the model.

CONCLUSIONS

The model constructed in this study could stratify stroke patients according to their risk of AF and may be helpful for selecting candidates who need extensive cardiac monitoring.