Prediction of right ventricular dysfunction in patients with pulmonary embolism

Purpose

Right ventricular dysfunction (RVD) is associated with adverse events in patients with pulmonary embolism (PE). An early diagnosis of RVD is thus necessary and may lead to a more aggressive approach. We investigated clinical, electrocardiographic and chest X-ray variables in patients with confirmed pulmonary embolism and proposed a model for prediction of RVD.

Methods

A multicenter cohort included 625 patients, from January 1998 to May 2003, admitted with diagnosis of pulmonary embolism confirmed by: pulmonary angiography, helical computer tomography, magnetic resonance, echocardiography or lung scan. From 550 patients who had a 2D echocardiogram, 191 (34%) met RVD criteria (ventricular dilatation or hypokinesia). We investigated 28 clinical (risk factors, signs and symptoms), electrocardiographic and chest X-ray findings in those patients for correlation with RVD. After univariate analysis, we selected variables (P < 0.20) for logistic regression. C-statistics were determined and the independent variables were applied for building a model for RVD prediction.

Results

In univariate analysis, gender, recent surgery (<30 days), chronic cor pulmonale, chest pain, tachycardia (>100 beats/min), syncope, tachypnea (>20 breaths/min), arterial hypotension (systolic BP <90 mmHg), cyanosis, right axis deviation, S1Q3T3 pattern, right bundle block (RBB) and T-wave inversion (V1–V4 leads) were selected for multivariate analysis. In logistic regression, recent surgery (OR = 0.37, P = 0.004), tachypnea (OR = 1.89, P = 0.001), hypotension (OR = 2.00, P = 0.002), S1Q3T3 pattern (OR = 3.33, P < 0.001), RBB (OR = 2.22, P < 0.001) and T-wave inversion (V1–V4 leads) (OR = 2.80, P < 0.01) were independent variables for prediction of RVD. The C-statistic was 0.71 and the resulting model presented a linear trend (P < 0.01).

Conclusion

In this proposed model, simple bedside information such as recent surgery, tachypnea, hypotension, S1Q3T3 pattern, RBB and T-wave inversion can be used to predict RVD in patients with PE in clinical practice.