The predictive value of thrombelastography system on thromboembolic and bleeding events in Chinese aged people-receiver operating characteristic analysis

ZHENG Qi-wen,BAI Jie,CHEN Da-fang

DOI: https://doi.org/10.16386/j.cjpccd.issn.1004-6194.2013.05.019

2013-01-01

Abstract:Objective To evaluate the value of thrombelastography(TEG) system in predicting thromboembolic and bleeding events in Chinese aged people using the receiver operating characteristic(ROC) curve, to determine the best cutoff of TEG indices for predicting. Methods The subjects were from the patients(more than 65 years old) who received TEG test in Chinese PLA General Hospital between January 2007 and December 2010. Patients with hematological system disease would be excluded.Medical records of subjects were reviewed to obtain clinical data including age, gender, details of anti-platelet or anticoagulant agent usage, thromboembolic disease history(stay in bed) and results of TEG testing. Clinical outcomes(thromboembolic and bleeding events) would be collected during the several follow-ups. ROC curves and the area under the curve(AUC) were used to evaluate and compare the accuracy and effects for the prediction. Results Four hundred and three subjects involved in this follow-up study, including 373 males and 30 females. 25 cases(6.2%) with bleeding and 78 cases(19.4%) with tromboembolic events were found during follow-up. The predictive value of TEG existing cut points for thromboembolic events was low. The best cutoff of TEG indices for predicting thromboembolic events were listed as followed: R =7.0 min, K =1.5 min, MA =63.5 mm and ANGLE=67.1°. After combination the four indices, the AUC was increased to 0.602 8, with sensitivity of 85.9% and specificity of34.7%. The best cutoff of TEG indices for predicting bleeding events were listed as followed: R=7.8 min, K=2.3 min, MA=50.5 mm and ANGLE=53.7°. After combination R and MA, the AUC was increased to 0.658 3, with sensitivity of 60.0% and specificity of71.7%. The results of multiple Cox regression showed that R and MA had the predict value in the outcome prediction. Conclusion The new cutoffs and the combination of four indices show better predictive value. The establishment of an effective laboratory method and an accurate assessment system of indices will be helpful for the identification and prediction of thromboembolic and bleeding events. It would make a great contribution to preventing aged people from bleeding and thromboembolic disease.