Establishment and evaluation of a prognostic model for acute pulmonary embolism

Abstract

To identify the factors influencing the prognosis of patients with acute pulmonary embolism(PE) and to establish a prognostic model. The clinical data of 331 patients (141 males and 190 females, aged 9 to 87 years ) with acute PE in Fujian Hospital from January 2007 to September 2013 were analyzed. Univariate analysis and logistic regression analysis were used for selecting the independent prognostic factors for acute PE. Based on logistic regression analysis, a prognostic model for PE was established. Univariate analysis showed that statistically significant (all P<0.05) factors influencing the prognosis of PE were diabetes, tricuspid systolic murmur, body temperature, respiratory rate, heart rate, aspartate aminotransferase, triglycerides, abnormal ECG, mechanical ventilation, circulatory failure during hospitalization, risk stratification of PE, types of treatment, and use of low-molecular-weight heparin and Warfarin. Logistic regression analysis showed that recent (<1 month) operation or fracture, tricuspid systolic murmur, high triglyceride level, circulatory failure during hospitalization and mechanical ventilation were independent factors for poor prognosis of PE, while combined use of low-molecular-weight heparin and Warfarin was a protective factor for the prognosis of PE. The Fisher prognostic model equation was y=0.144+ 1.266x1+ 0.869x2+ 1.794x3-0.517x4+ 3.555x5+ 0.661x6. The accuracy of the Fisher discriminant function was 93.0%. Recent (<1 month) operation or fracture, tricuspid systolic murmur, high triglyceride level, shock during hospitalization and mechanical ventilation were signs of poor prognosis for PE, while combined use of low-molecular-weight heparin and Warfarin were beneficial for the prognosis. The discriminant function based on these data can be helpful for predicting the prognosis of patients with PE.