Does rotational thromboelastometry (ROTEM) improve prediction of bleeding after cardiac surgery?

BACKGROUND: Coagulopathy and massive bleeding are severe complications of cardiac surgery, particularly in procedures requiring prolonged cardiopulmonary bypass (CPB). There is huge variability in transfusion practices across hospitals and providers in cross-sectional studies. This variability may indicate unguided decision-making, perhaps attributable to lack of reliable, predictive laboratory testing of coagulopathy to guide transfusion practice. Rotational thromboelastometry (ROTEM) measures multiple coagulation parameters and may provide value from its ease of use, rapid results, and measurement of several steps in the coagulation pathway. Yet, the predictive value and utility of ROTEM remains unclear. In this study, we investigated ROTEM's predictive value for chest tube drainage after cardiac surgery.
METHODS: Three hundred twenty-one patients undergoing cardiac surgery involving CPB were enrolled. Patient data were obtained from medical records, including chest tube output (CTO) from post-CPB through the first 8 postoperative hours. Perioperative and postoperative blood samples were collected for ROTEM analysis. Three measures of CTO were used as the primary end points for assessing coagulopathy: (i) continuous CTO; (ii) CTO dichotomized at 600 mL (75th percentile); and (iii) CTO dichotomized at 910 mL (90th percentile). Clinical and hematological variables, excluding ROTEM data, that were significantly correlated (P < 0.05) with continuous CTO were included in a stepwise regression model (model 1). An additional model that contained ROTEM variables in addition to the variables from model 1 was created (model 2). Significance in subsequent analyses was declared at P < 0.0167 to account for the 3 CTO end points. Net reclassification index was used to assess overall value of ROTEM data.
RESULTS: For continuous CTO, ROTEM variables improved the model's predictive ability (P < 0.0001). For CTO dichotomized at 600 mL (75th percentile), ROTEM did not improve the area under the receiver operating characteristic curve (AUC) (P = 0.03). Similarly, for CTO dichotomized at 910 mL (90th percentile), ROTEM did not improve the AUC (P = 0.23). Net reclassification index similarly indicated that ROTEM results did not improve overall classification of patients (P = 0.12 for CTO ≥600 mL; P = 0.08 for CTO ≥910 mL).
CONCLUSIONS: These results suggest that ROTEM data do not substantially improve a model's ability to predict chest tube drainage, beyond frequently used clinical and laboratory parameters. Although several ROTEM parameters were individually associated with CTO, they did not significantly improve goodness of fit when added to statistical models comprising only clinical and routine laboratory parameters. ROTEM does not seem to improve prediction of chest tube drainage after cardiac surgery involving CPB, although its use in guiding transfusion during cardiac surgery remains to be determined.