Prognostic value of right ventricular dysfunction and derivation of a prognostic model for patients with acute pulmonary thromboembolism

Abstract

Acute pulmonary thromboembolism (PTE) patients with right ventricular dysfunction (RVD) may benefit from thrombolytic therapy but may end up with worse prognosis. RVD was assessed in prognosis to which a model on it was constructed to decide the indexes correlated to the best prognosis. This prospective study included 520 consecutive acute PTE patients from 41 hospitals in China between June 2002 and February 2005. All the patients were evaluated by transthoracic echocardiography (TTE), CT pulmonary angiography (CTPA), laboratory tests, and blood gas analysis. Physicians were asked to record all the clinical manifestations. Data from Univariate analysis demonstrated the parameters correlated with an 14-day clinically adverse outcomes. Multiple logistic regression analysis was used to decide the independent predictors and to construct a prognostic model. Mean age of the patients was 57.4 +/- 14.1 years and 323 of them (62.1%) were male. The 14-day mortality in normotensive patients with RVD was higher' (2.0% vs. 0.4%, P < 0.01) than those without. ROC curve showed the best cut-off values of RVED/LVED and SPAP for 14-day prognosis were 0.67 and 60 mm Hg, respectively. SPAP independently predicted 3-month clinical outcomes (P < 0.01). Results from Univariate analysis demonstrated that 24 parameters were correlated with an adverse 14-day clinical outcomes, which include palpation, syncope, panic, cyanosis, respiratory rate > or = 30/min, pulse > or = 110/min, jugular vein, accentuation of P2, murmurs in tricuspid area, time interval from onset, creatine phosphokinase (CPK), lactate dehydrogenase (LDH), alveolar-arterial PO2 difference (PA-aO2), white blood cell (WBC) < 4.0 x 10(9)/L or >10.0 x 10(9)/L, platelet, thrombus on TTE, RV/LV > 1.1, TRPG >30 mm Hg, IVCmin <8 mm, RVD, RVED/LVED > 0.6, SPAP >60 mm Hg, RVWM, PTE range larger than two lobes or seven segments on CTPA. Furthermore, a multiple logistic regression model implied 8 predictors including RVD, RVED/LVED > 0.6, SPAP >60 mm Hg, pulse > or = 110/min, accentuation of P2, Syncope, CPK, WBC < 4.0 x 109/L or > 10.0 x 10(9)/L be independent predictors of an 14-day clinically adverse outcome (P < 0.01). This model seemed to fit well (P < 0.001). We chose a cut-off value as P > or = 0.2 and compared the model to the original derivation samples. Data showed that the sensitivity (true positive rate) was 81.82%, specificity was 92.11%, false positive rate was 18.18%, coincidence was 91.14%, and the concordance rate was 80.96%. RVD seemed a nice discriminator for poor prognosis in normotensive patients. Early detection of RVD (especially RVED/LVED > 0.67 and/or SPAP >60 mm Hg) was beneficial for identifying patients at high-risk and the multiple logistic regression model (P < 0.001) could be well fitted.