J. Coleman1, E. E. Moore2, A. Banerjee1, C. C. Silliman1, M. P. Chapman1, H. B. Moore1, A. Ghasabyan2, J. Chandler2, J. Samuels1, A. Sauaia1 1University Of Colorado,Surgery,Denver, CO, USA 2Denver Health,Surgery,Denver, CO, USA
Introduction: Several thrombelastography (TEG) functional assays have been developed to reduce time to available data to guide transfusion in trauma and hemorrhage. These assays involve addition of various agonists: native, kaolin (contact activation/intrinsic pathway) and rapid with kaolin plus tissue factor (extrinsic pathway). How this acceleration of TEG affects its efficiency and accuracy as a predictor of adverse outcomes is unknown. We hypothesize that citrated native TEG, without addition of activators, best predicts massive transfusion (MT).
Methods: The Trauma Activation Protocol (TAP) study is an ongoing prospective study of trauma activation patients. Whole blood samples were immediately collected upon presentation from April 2015 to March 2017 for concomitant citrated native (CN-TEG), citrated kaolin (CK-TEG) and citrated rapid (CR-TEG) TEGs. We assessed the predictive performance for MT (defined as >10 RBC units or death within 6 hours postinjury) via the area under the receiver-operating- characteristics curve (AUROC, derived with a logistic regression model) for ACT/R, maximum amplitude (MA) and angle, and via positive/negative predictive values (PV+, PV-) for fibrinolysis (LY30, stratified as >3% or >5% or >7%). We excluded 43 (12%) patients who died within 30 minutes postinjury. Data are expressed as median(IQR) or n(%).
Results: Overall, 339 TAP patients had data for all TEGs (age: 32 years, IQR:26-46, 51% blunt injuries, NISS: 21, IQR:9-34). RBC transfusion was required in 140 (41%) patients; 12% (n=41) qualified as MT (>10RBC units or death/6hrs). The figure depicts AUROCs for ACT (CR-TEG) or R (CN-TEG, CK-TEG), MA and angle. Compared to CR-TEG, CK-TEG performed better for ACT/R (p=0.06 and angle (p=0.09), while CN-TEG was better for MA (p=0.16). LY30>7% produced the best balance between PV+ (CR-TEG: 47%, CK-TEG: 58%, CN-TEG: 67%) and PV- (CR-TEG: 91%, CK-TEG: 92%, CN-TEG: 92%). The 95% confidence intervals of the AUROCs and of the PVs of the different assays overlapped considerably, suggesting the CR-TEG’s results were not inferior to the other functional assays. When all TEG measurements (ACT/R+MA+angle+LY30>7%) for each assay (CR-, CN-, CK-TEG) were in the model, again there was considerable overlap between predictive performances of the different assays and all AUROCs were >0.79.
Conclusion: There was a significant overlap in the performance of the different TEG assays, suggesting CR-TEG is a rapid and efficient method to guide hemostatic resuscitation in trauma patients.
