Abstract
Fatal pulmonary embolism is the third most common cause of death after major trauma. We hypothesised that the Trauma Embolic Scoring System (TESS) would have adequate calibration and discrimination in a group of severely injured trauma patients in predicting venous thromboembolism (VTE), and could be used to predict fatal and non-fatal symptomatic pulmonary embolism. Calibration and discrimination of the TESS were assessed by the slope and intercept of the calibration curve and the area under the receiver operating characteristic curve, respectively. Of the 357 patients included in the study, 74 patients (21%) developed symptomatic VTE after a median period of 14 days following injury. The TESS predicted risks of VTE were higher among patients who developed VTE than those who did not (14 versus 9%, P=0.001) and had a moderate ability to discriminate between patients who developed VTE and those who did not (area under the receiver operating characteristic curve 0.71, 95% confidence interval 0.65 to 0.77). The slope and intercept of the calibration curve were 2.76 and 0.34, respectively, suggesting that the predicted risks of VTE were not sufficiently extreme and overall, underestimated the observed risks of VTE. Using 5% predicted risk of VTE as an arbitrary cut-point, TESS had a high sensitivity and negative predictive value (both ≥0.97) in excluding fatal and non-fatal pulmonary embolism. The TESS had a reasonable ability to discriminate between patients who developed VTE and those who did not and may be useful to select different strategies to prevent VTE in severely injured patients.
