The value of urine albumin in predicting thromboembolic events for patients with non-valvular atrial fibrillation

Background: Accurate risk stratification is important in the management of patients with non-valvular atrial fibrillation (NVAF). However, one cohort study demonstrated an annual ischemic stroke rate of 1.61% in the group of patients classified in "the true low risk" according to CHA(2)DS(2)-VASc. We aimed to find out more indicators and evaluate their abilities in predicting thromboembolic events (TE). Methods: We assigned 58 patients with TE to the thrombosis group, and 157 patients without TE to the non-thrombosis group. The clinical parameters of these patients were subjected to univariate analysis and unconditioned logistic regression analysis for screening the risk factor, which was urine albumin (UA) according to the result. The receiver operating characteristic (ROC) curve was used to determine the optimal cut-off point of the UA. Then we formed the CHA(2)DS(2)-VASc-UA(2) score and made a comparison with CHA(2)DS(2)-VASc score. Results: Mean UA of the thrombosis group was significantly higher than that of the non-thrombosis group (0.1 g/L vs 0.0 g/L, P < 0.01). The results of unconditioned logistic regression analysis showed that OR of UA was 40.98 (95% CI: 3.58-468.88, P < 0.01). The Area Under the Curve (AUC) of UA was 0.700 with an optimal cut-off point of 0.03 g/L. ROC curve analysis result showed that AUC of CHA(2)DS(2)-VASc-UA(2) score was larger than that of CHA(2)DS(2)-VASc score (0.873 vs 0.860, P < 0.01). Conclusion: UA >= 0.03 g/L is the independent predictive factor of TE for NVAF patients. And the CHA(2)DS(2)-VASc-UA(2) score might perform better in predicting TE compared with the CHA(2)DS(2)-VASc score. (C) 2016 Elsevier Ireland Ltd. All rights reserved.