Policy, Payment & Practice -> Clinical Quality Measures and Outcomes D-PO03 - Poster Session III (ID 48) Poster

D-PO03-117 - Clinic Model-based Point Scoring System For Predicting The Risk Of Strokes In Atrial Fibrillation Patients In The Era Of Catheter Ablation (ID 330)

 Y. Chen: Nothing relevant to disclose.


Background: The stroke risk scoring system for atrial fibrillation (AF) patients can vary considerably based on whether receiving catheter ablation (CA).
Objective: This study aimed to develop a clinical scoring system to predict the risk of stroke based on status of AF CA.
Methods: This study evaluated the stroke rates in AF patients by linking the Taiwan National Health Insurance Research Database. First, 147,225 AF patients were identified as validation AF cohort, with 1897 (1.3%) receiving AF CA. Second, 787 AF patients undergoing AF CA were age- sex-, underlying disease-matched to the same number of AF patients without CA as a propensity-score (PS) matched cohort. Multivariate Cox model-derived coefficients were used to construct a simple point-based clinic model using PS-matched cohort. The novel model (AF-CA-stroke score) was validated using a validation AF cohort.
Results: The AF-CA-stroke score was calculated based on the age (point=6), ablation status (point=4), prior history of a stroke (point=4), chronic kidney disease (point=2), diabetes mellitus (point=1), and congestive heart failure (point=1)(Table 1). The risk function for predicting the 1-, 5-, and 10-year absolute stroke risks was reported (Figure). The estimated area under the receive operating characteristic curve for AF-CA-stroke scores in PS-matched cohort was 0.844 (Table 2). A validation study showed that AF-CA-stroke scores had a significantly higher discrimination ability than CHADS2/CHA2DS2-VASc scores (P<0.001; Table 2).
Conclusion: The newly constructed clinic model-based point scoring system was useful for identifying the stroke risk using clinical factors and the AF CA status in AF patients.