Comparison of two prognostic models for acute pulmonary embolism: clinical vs. right ventricular dysfunction-guided approach

Background: Recently, some prognostic models for acute pulmonary embolism (PE) have been proposed. We investigated whether the Pulmonary Embolism Severity Index (PESI) and the European Society of Cardiology (ESC) prognostic approaches result in different prognoses. Methods: Consecutive adult patients with acute PE were included. According to the ESC guidelines, high-risk patients were identified by the presence of shock/hypotension, intermediate-risk patients by elevated troponin I or right ventricular dysfunction as assessed by echocardiography, and low-risk patients by the absence of any of the above. In the PESI model, 11 clinical variables, easily accessible at the bedside, were used to generate three risk classes. The main outcomes were all-cause and PE-related in-hospital mortality. Results: Forty-one patients (8%, 95% confidence interval [CI] 5.8-10.8) of 510 died. According to the ESC model, 40% were at low risk of short-term mortality, 54% at intermediate risk, and 6% at high risk. The distribution according to the PESI model was 31% (P < 0.05 vs. ESC), 49% and 20% (P < 0.05 vs. ESC), respectively. Mortality increased through the risk classes (P < 0.01), without significant differences between the models. The ESC model identified with higher accuracy than the PESI model both high-risk and low-risk patients (P < 0.05 for both). When patients with shock/hypotension were excluded, the PESI model stratified patients into classes with increasing PE-related mortality (0.7%, 4.3%, and 11.6%, P < 0.05). Troponin I and right ventricular dysfunction added incremental prognostic value to the PESI model, particularly in normotensive patients at intermediate risk. Conclusions: The ESC model showed higher accuracy than the PESI model in identifying high-risk and low-risk patients. In normotensive patients, the PESI model could guide clinical management as well as troponin I and echocardiography testing.