Effects of body temperature on accuracy of continuous cardiac output measurements

Intermittent measurement of cardiac output is routine in the critically ill surgical patient. A new catheter allows real-time continuous measurement of cardiac output. This study evaluated the impact of body temperature variation on the accuracy of these measurements compared to standard intermittent bolus thermodilution technique. This prospective study in a university hospital surgical intensive care unit included 20 consecutive trauma patients. Data were collected with pulmonary artery catheters, which allowed both continuous (COC) and bolus (COB) thermodilution measurements. The catheter was placed through either the subclavian or internal jugular vein. Measurements for COB were performed using a bolus (10 cm3) of ice-cold saline with a closed-injectate delivery system at end-expiration. Computer-generated curves were created on a bedside monitor, and the average of three measurements within 10% of one another was used as COB. COC was determined as the average of the displayed CO before and after thermodilution CO measurements. Body temperature was measured from the pulmonary artery catheter and was grouped as < or =36.5 degrees C, 36.6-38.4 degrees C, and > or =38.5 degrees C. COB and COC were compared for agreement by plotting the mean of the differences (COB - COC) between the methods. The differences were plotted against the average of each pair and analyzed with linear regression. One hundred seventy-eight paired measurements were made over a period of 1 to 3 days. CO ranged from 3.7 to 15.5 L/min. Eighty-one percent of measurements were at a temperature of 36.5-38.4 degrees C. Approximately 7% of measurements were at a temperature below 36.5 degrees C and 11.2% were in patients with a core temperature above 38.5 degrees C. Correlation between the two techniques was 0.96, 0.91, and 0.82 for temperatures of < or =36.5 degrees C, 36.6-38.4 degrees C, and > or = 38.5 degrees C, respectively. In conclusion, the COC measurements correlate well with COB in trauma patients with a core temperature < or =38.5 degrees C. The accuracy degraded at higher temperatures, which may be related to the smaller signal-to-noise ratio at elevated body temperatures.