Abstract 13039: Combination and Hierarchical Use of Echocardiographic Parameters for Predicting Favorable Hemodynamic Response after Drainage of Pericardial Effusion
Background: Echocardiography has a major role in decision-making for pericardial drainage. The individual accuracy of echocardiographic parameters is variable and their additive value is unknown. We sought how best to combine them together or in a hierarchy.
Methods: After excluding patients with mechanical respiratory ventilation or arrhythmia at the time of echocardiography, we assessed 184 consecutive patients (age 59±16 yrs; 81 post-operative, 21 malignant and 82 others) who had moderate or greater pericardial effusion. Response to pericardial drainage (61/82 patients) was defined by reduction of heart rate after the procedure. A Classification and Regression Tree (CART) algorithm was used to combine parameters for prediction of response. We evaluated selection by; 1) combination of these variables and 2) application of decision tree.
Results: There were 40 patients who had pericardiocentesis, 38 pericardial window and 4 had both procedures. Out of 82 those who were drained, indices chosen by CART were IVC diameter ≥2.5cm, systolic blood pressure (sBP) <120mmHg, mitral inflow respiratory variation (MVAR) >28%, lateral pericardial effusion ≥1.6cm and anterior pericardial effusion ≥1.1cm. Presence of any of these parameters showed very high sensitivity (96%) and negative predictive value (85%) but low specificity (17%). The use of a decision tree (Figure) yielded AUC 0.92 and relatively balanced accuracy (sensitivity 61%, specificity 75%, PPV 67% and NPV 71%). CART showed patients with a large IVC, or low sBP with mitral flow variation to improve with drainage. Patients with preserved sBP and a small effusion had a low likelihood of response to drainage.
Conclusion: IVC diameter, sBP, MVAR, and anterior and lateral size of effusion are the most discriminative markers. Absence of any of these parameters identifies patients unlikely to benefit from drainage. Specificity can be improved by the use of the decision tree.
- © 2012 by American Heart Association, Inc.