A Retrospective Analysis of Thromboelastometry and Conventional Laboratory Values in Nonsurgical and Noncardiac Surgery Patients: Not What One Might Expect From Cardiac Surgery Data

BACKGROUND: Thromboelastometry is often used to guide coagulation management, but validation outside cardiac surgery is limited. We evaluated correlations between thromboelastometry parameters and conventional laboratory tests in noncardiac patients and assessed a cardiac surgery-derived predictive app. METHODS: A retrospective single-center cohort of 257 patients with simultaneous thromboelastometry and laboratory measurements formed the study population. Correlations were assessed with the Pearson r and agreement with Bland-Altman analysis. Diagnostic performance for hypofibrinogenemia (≤1.5 g/L) and thrombocytopenia (≤100 × 10⁹/L) was evaluated using receiver operating characteristic (ROC) analyses with area under the ROC (AUROC) and 95% confidence intervals. RESULTS: Fibrinogen correlated strongly with FIBTEM (the extrinsic activation pathway, whereby platelets are blocked in FIBTEM and the resulting clot consists only of fibrin formation and polymerization) A10 (specific point of time of 10 minutes) and FIBTEM MCF (maximal clot firmness) (r = 0.82). Platelet count correlated with EXTEM (the extrinsic activation pathway) A10 (r = 0.58), INTEM (the intrinsic pathway is activated) A10 (r = 0.61), and EXTEM-FIBTEM A10 (r = 0.57). AUROC for hypofibrinogenemia was 0.92 (FIBTEM A10), 0.89 (FIBTEM MCF), and 0.89 (EXTEM alpha). AUROC for thrombocytopenia was 0.95 (EXTEM A10), 0.96 (INTEM A10), and 0.97 (EXTEM-FIBTEM A10). Predicted versus observed values correlated at r = 0.82 for fibrinogen and r = 0.58 to 0.61 for platelets. Bland-Altman analyses showed systematic underestimation by the app with wide limits of agreement (bias +0.8 g/L for fibrinogen; +62, +50, and +50 × 10⁹/L for platelets). CONCLUSIONS: FIBTEM A10 (fibrinogen) and EXTEM/INTEM A10 (platelets) correlate well with conventional laboratory measures in a heterogenous noncardiac cohort, supporting thromboelastometry as a point-of-care tool. However, the cardiac-derived predictive model shows substantial bias and variability, precluding reliable patient-level estimates. The use cannot be recommended in noncardiac patients.