Stefano Spolveri, Filippo Pieralli, Giovanni Antonio Porciello, Simone Meini, Valerio Verdiani, Daniele Baldoni, Alessandra Petrioli, Carlotta Casati, Massimo Alessandri, Carlo Passaglia, Maurizio Manini, Luca Masotti, Michele Voglino, Stefano Tatini, Lucia Raimondi, Plinio Fabiani, Alessandro Pampana, Chiara Angotti, Stefano Arrigucci, Stefano Giuntoli, Grazia Panigada, Salvatore Lenti, Sara Bucherelli, Alberto Camaiti, Alberto Fortini, Roberto Cappelli, Roberto Andreini, Lucia Ciucciarelli, Alessandro Tafi, Gianni Lorenzini, Marco Cei, Filippo Risaliti, Alessandro Morettini, Laila Teghini, Adriano Cioppi, Massimo Di Natale, Michele Piacentini, Anna Maria Romagnoli, Raffaele Laureano, Carlo Nozzoli, Guidantonio Rinaldi, Nicola Mumoli, Veronica De Crescenzo, Maria Chiara Bertieri, Irene Cascinelli, Emilio Santoro, Giuseppa Levantino, Luciano Ralli, Alessandro De Palma, Claudia Rosi, Anna Frullini, Giancarlo Tintori, Paola Lambelet, Giancarlo Landini, Rino Migliacci, Maddalena Grazzini, Rossella Nassi, Roberta Mastriforti, Barbara Cimolato, Carlo Palermo, Francesco Corradi, Valentina Carli, Mario Felici, Alba Dainelli, and Stefano Fascetti
Background: Prognostic stratification is of utmost importance for management of acute Pulmonary Embolism (PE) in clinical practice. Many prognostic models have been proposed, but which is the best prognosticator in real life remains unclear. The aim of our study was to compare and combine the predictive values of the hemodynamics/biomarkers based prognostic model proposed by European Society of Cardiology (ESC) in 2008 and simplified PESI score (sPESI). Methods: Data records of 452 patients discharged for acute PE from Internal Medicine wards of Tuscany (Italy) were analysed. The ESC model and sPESI were retrospectively calculated and compared by using Areas under Receiver Operating Characteristics (ROC) Curves (AUCs) and finally the combination of the two models was tested in hemodinamically stable patients. All cause and PE-related in-hospital mortality and fatal or major bleedings were the analyzed endpoints Results: All cause in-hospital mortality was 25% (16.6% PE related) in high risk, 8.7% (4.7%) in intermediate risk and 3.8% (1.2%) in low risk patients according to ESC model. All cause in-hospital mortality was 10.95% (5.75% PE related) in patients with sPESI score ≥1 and 0% (0%) in sPESI score 0. Predictive performance of sPESI was not significantly different compared with 2008 ESC model both for all cause (AUC sPESI 0.711, 95% CI: 0.661-0.758 versus ESC 0.619, 95% CI: 0.567-0.670, difference between AUCs 0.0916, p=0.084) and for PE-related mortality (AUC sPESI 0.764, 95% CI: 0.717-0.808 versus ESC 0.650, 95% CI: 0.598-0.700, difference between AUCs 0.114, p=0.11). Fatal or major bleedings occurred in 4.30% of high risk, 1.60% of intermediate risk and 2.50% of low risk patients according to 2008 ESC model, whereas these occurred in 1.80% of high risk and 1.45% of low risk patients according to sPESI, respectively. Predictive performance for fatal or major bleeding between two models was not significantly different (AUC sPESI 0.658, 95% CI: 0.606-0.707 versus ESC 0.512, 95% CI: 0.459-0.565, difference between AUCs 0.145, p=0.34). In hemodynamically stable patients, the combined endpoint in-hospital PE-related mortality and/or fatal or major bleeding (adverse events) occurred in 0% of patients with low risk ESC model and sPESI score 0, whilst it occurred in 5.5% of patients with low-risk ESC model but sPESI ≥1. In intermediate risk patients according to ESC model, adverse events occurred in 3.6% of patients with sPESI score 0 and 6.65% of patients with sPESI score ≥1. Conclusions: In real world, predictive performance of sPESI and the hemodynamic/biomarkers-based ESC model as prognosticator of in-hospital mortality and bleedings is similar. Combination of sPESI 0 with low risk ESC model may identify patients with very low risk of adverse events and candidate for early hospital discharge or home treatment.