The Value of Thromboelastography for Bleeding Risk Prediction in Hematologic Diseases

Background: This study aimed to explore the correlations between thromboelastography (TEG) parameters with platelet (PLT) count and fibrinogen and to evaluate the value of the maximal amplitude (MA) for bleeding risk prediction. Methods: A total of 1,559 patients with hematologic diseases underwent PLT counting and TEG tests, and 1,201 of these patients underwent conventional coagulation tests. Patients were divided into a bleeding group and a nonbleeding group according to their clinical records. Results: Patients in the bleeding group had lower PLT counts, a-angle values, MA values and higher K values (all P < 0.05) than patients in the nonbleeding group. Low PLT counts (<= 30 x 10(9)/L) were found in 265 patients and bleeding episodes occurred in 109 patients (41.13%). A total of 99 patients had both low MA values and bleeding episodes in this subgroup. A total of 124 of the 265 patients (46.79%) had hematological malignancies. In the 2 different types of diseases, there was a similar tendency in bleeding risk prediction according to the receiver operating characteristic curves. The curves using both the PLT counts and MA values show a higher sensitivity and a slightly lower specificity than those of the PLT count or MA alone. Conclusions: There are some correlations between the TEG parameters and the traditional hemostatic parameters. The combination of the PLT counts and MA values had greater predictive value for bleeding risk in hematological diseases when the PLT counts were at a low level (<= 30 x 10(9)/L).