Predicting bleeding risk in rivaroxaban-treated patients with non-valvular atrial fibrillation based on nomograms and CHA2DS2-VASc,HAS-BLED scores

LIN Yuan; CHEN Jun; TANG Feng; HE Yan

doi:10.3969/j.issn.1672-5301.2025.05.004

Abstract：Objective The purpose of this study was to compare the predictive value of CHA2DS2-VASc and HAS-BLED scores in patients with nonvalvular atrial fibrillation(NVAF),and to build a prediction system in order to provide basic clinical research data.Methods In this study,patients with NVAF treated with rivaroxaban in Guizhou Medical University Hospital from January 2019 to December 2023 were retrospectively analyzed.Demographic data,coagulation indexes,and laboratory biochemical detection indexes of the patients were collected,and CHA2DS2-VASc and HAS-BLED scores were calculated.The predictive performance of single and combined applications was evaluated.The main predictors of bleeding events in patients treated with rivaroxaban were identified by single factor analysis.Then,the main independent predictors selected in the single factor analysis were screened through the multi-factor analysis.Finally,a prediction model of bleeding risk in patients with NVAF treated with rivaroxaban was constructed based on the nomogram model.Results A total of 363 patients were included in this study,and they were divided into bleeding group(n=63 cases)and non-bleeding group(n=300 cases).The mean age of patients in the bleeding group was significantly higher than that in the non-bleeding group[(68.43±12.26)years vs.(65.05±13.77)years,P=0.014],and the creatinine clearance was significantly lower than that in the non-bleeding group[(67.71±28.91)vs.(75.89±30.26),P=0.009].In the bleeding group,Scr[(93.44±5.04)umol/L vs.(85.06±6.17)μmol/L,P=0.01],estimate glomerular filtration rate(eGFR)[(80.53±6.11)ml·min-1·1.73m-2 vs.(65.43±4.18)ml·min-1·1.73m-2,P=0.003],eGFR＜45％(20.63％vs.7.47％,P=0.021)and N-terminal pro-brain natriuretic peptide(NT-proBNP)[(1308.34±405.65)pg/ml vs.(564.48±115.62)pg/ml,P＜0.001)were significantly higher than those in non-bleeding group.In terms of coagulation function,prothrombin time(PT)[(13.10±2.42)s vs.(11.20±1.89)s,P=0.002),activated partial thromboplastin time(APTT)[(28.92±3.44)s vs.(27.34±3.81)s,P=0.003]and international normalized ratio(INR)values[(1.35±0.15)vs.(1.06±0.21),P=0.001]were observed in the bleeding group.P=0.001)were significantly higher than those without bleeding.There was no significant difference in the use rate of antiplatelet drugs between the two groups(47.62％vs.38.67％).The sensitivity,specificity,positive predictive value of 38.09％and negative predictive value of 96.75％of CHA2DS2-VASc score in predicting NVAF hemorrhage treated with rivaroxaban were 88.89％,69.67％,38.09％and 96.75％respectively.The HAS-BLED score had a sensitivity of 90.48％,a specificity of 77.67％,a positive predictive value of 45.97％,and a negative predictive value of 97.48％in predicting NVAF hemorrhage treated with rivaroxaban.The sensitivity,specificity,positive predictive value,and negative predictive value of CHA2DS2-VASc and HAS-BLED score were 93.65％,84.00％,55.14％,and 98.43％,respectively,in predicting NVAF hemorrhage treated with rivaroxaban.ROC analysis results showed that the area under the curve of CHA2DS2-VASc was 0.578 and that of HAS-BLED was 0.537,and the area under the curve was 0.678 in the combined prediction.Multivariate analysis showed that age(P＜0.001),creatinine clearance(P=0.008),Scr(P=0.028),CHA2DS2-VASc score(P＜0.001)and HAS-BLED score(P=0.002)were independent risk factors for bleeding.The analysis of the nomogram showed that the bleeding risk score ranged from 0 to 100,corresponding to the bleeding risk rate of 0.1 to 1.0,and the model AUC was 0.814(95％CI 0.769-0.859).Conclusion Age,creatinine clearance,serum creatinine combined with HAS-BLED and CHA2DS2-VASc scoring system can well predict the risk of bleeding in patients with NVAF treated with oral rivaroxaban,with good validity reliability and potential clinical application value.