Assessing hemodynamic status in critically ill patients: Do physicians use clinical information optimally?

Abstract

Estimates of hemodynamic status in critically ill patients are frequently inaccurate. This may be because clinical information correlates poorly with hemodynamic status or because physicians do not use clinical information optimally. To distinguish between these two possibilities we used discriminant analysis to identify the clinical variables which were most predictive of the pulmonary capillary wedge pressure and cardiac index in 69 right heart catheterizations. When discriminant functions were derived for all 69 catheterizations, they predicted the range (low, normal, or high) of wedge pressure and cardiac index 66.7% and 71.0% of the time, respectively. When discriminant functions were derived for specific clinical subgroups they predicted wedge pressure 78% to 94% of the time and cardiac index 73% to 92% of the time. Physicians, however predicted the range of pulmonary capillary wedge pressure and cardiac index correctly only 49.7% and 42.5% of the time, respectively. Thus, optimal use of the clinical information by discriminant analysis resulted in significant improvement in the prediction of hemodynamic status in these critically ill patients when compared to physician judgments. Physician estimates of hemodynamic status in these patients were inaccurate apparently because physicians used the available clinical information suboptimally. Improved use of clinical information could allow more accurate clinical judgements, reducing the need for invasive hemodynamic measurements in critically ill patients.