Assessment of ventricular function in critically ill patients: Limitations of pulmonary artery catheterization

Objective: To determine the accuracy of conventional hemodynamic assessment using pulmonary artery catheter-derived data in critically ill patients. Design: Cohort study. Setting: Kaiser Permanente and Veterans Affairs Medical Centers. Participants: Twenty-five consecutive patients who had undergone elective aortocoronary bypass surgery. Measurements and Main Results: In the intensive care unit,conventional assessment (CA) was performed hourly by clinicians using conventional (radial artery and pulmonary artery) hemodynamic measurements from which left ventricular (LV) function and intracardiac volume were estimated. Simultaneously, transesophageal echocardiography (TEE) data were recorded on videotape, blinded to the clinicians, and quantitatively analyzed off-line. TEE-determined LV function was classified as either normal (ejection fraction greater than or equal to 40%) or abnormal (ejection fraction <40%) and intracardiac volume as normal (end-diastolic area = 8 to 22 cm(2)), low (end-diastolic area <8 cm(2)), or high (end-diastolic area >22 cm(2)). Conclusion: Evaluable data included 130 of 150 (87%) observations of simultaneously collected CA and TEE data, averaging 5.6 +/- 4.4 observations per patient. The overall predictive probability for conventional clinical assessment of normal ventricular function was 98% (118/121), whereas for abnormal ventricular function it was 0% (0/9). For CA of volume, the overall predictive probabilities for hypovolemia, normovolemia, and hypervolemia were 50% (3/6), 60% (69/115), and 22% (2/9). Although conventional clinical assessment of normal LV function in the intensive care unit correlates well with echocardiographic assessment, both LV dysfunction and extremes of preload (hypovolemia or hypervolemia) are assessed poorly by clinicians using conventional clinical monitoring with pulmonary artery catheterization. Copyright (C) 1999 by W.B. Saunders Company.