Predictive Value Of P-wave Maximum Duration And Dispersion Associate With N-terminal Brain Natriuretic Peptide In Patients With Paroxysmal Atrial Fibrillation

Objective:To analyze the prognostic effect of p-wave maximum duration?Pmax?,p-wave dispersion?Pd?and N-terminal brain natriuretic peptide?NT-proBNP?in paroxysmal atrial fibrillation?PAF?patients.Methods:Case control study method of grouping was applied,and patients with PAF?PAF group?who hospitalized in department of cardiology of The First Affiliated Hospital of Guangxi Medical University from 2013 January to 2016 June and patients with non atrial fibrillation?controlled group?who hospitalized in department of hypertension of The First Affiliated Hospital of Guangxi Medical University from 2015 January to 2016 Novemberwere were selected to collect their Pmax,Pd and NT-proBNP levels.Logistic regression combined with ROC?receiver operating characteristic curve?curve was used to evaluate the predictive value of each index.Results:A total of 96 patients with PAF and 106 patients with Non atrial fibrillation were selected and divided into PAF group and controled group.Pmax,Pd and NT-proBNP of PAF group were higher than controlled group significantly,the difference was statistically significant?P<0.05?.Applied the ROC curve analysis of Pmax,Pd and NT-proBNP in predictor of patients with PAF and the AUC were 0.776,0.899,0.780.When the Youden exponent was largest,and the corresponding diagnostic thresholdst were 100 ms,40ms and119.50pg/ml,the sensitivity was respectively 54.17%,85.42% and 79.17%.thespecificity was respectively 90.48%,82.54% and 74.60%.the positive predictive values were respectively 81.25%,78.85% and 70.37%.the negative predictive values were respectively 82.61%?82.54% and 74.60%.the diagnostic accuracy rates were respectively 72.28%,82.18% and 74.26%.the rateof missed diagnosis were respectively 45.83%,14.58% and 20.83%.the misdiagnosis rate were respectively 10.53%,19.30%,28.07%.By binary logistic analysis there was the correlation among the above indexes and PAF.The rstatistical diagnosis model was builded:Pre?predicted?=1/[1+e^{-?-5.628+0.008×Pmax+0.184×Pd+0.031×NT-proBNP?}].The diagnostic thresholdst was 0.415 when the Youden exponent was largest,and the corresponding values of AUC?0.932?,specificity?92.45%?,positive predictive value?88.64%?,negative predictive value?84.21%?,diagnostic accuracy rate?86.14%?and misdiagnosis rate?8.72%?in Pre were better than the single indexes',while the values of sensitivity?81.12%?and rate of missed diagnosis?18.75%?were only worse than the Pd's in the single indexes.Conclusion:Pmax,Pd,NT-proBNP were increased significantly in patients with PAF,and can be effective predictors of PAF;by applying binary logistic regression and ROC curve analysis to establish the statistical diagnosis model can more effectively predict the patients with PAF.