Continuous cardiac output measurement: pulse contour analysis vs thermodilution technique in cardiac surgical patients

Transcription

1 British Journal of Anaesthesia 82 (4): (1999) Continuous cardiac output measurement: pulse contour analysis vs thermodilution technique in cardiac surgical patients G. Rödig*, C. Prasser, C. Keyl, A. Liebold and J. Hobbhahn Department of Anaesthesia, University Hospital, Franz-Josef-Strauss Allee 11, D Regensburg, Germany *To whom correspondence should be addressed We have analysed the clinical agreement between two methods of continuous cardiac output measurement: pulse contour analysis (PCCO) and a continuous thermodilution technique (CCO), were both compared with the intermittent bolus thermodilution technique (BCO). Measurements were performed in 26 cardiac surgical patients (groups 1 and 2, 13 patients each, with an ejection fraction 45% and 45%, respectively) at 12 selected times. During operation, mean differences (bias) between PCCO BCO and CCO BCO did not differ in either group. However, phenylephrine-induced increases in systemic vascular resistance (SVR) by approximately 60% resulted in significant differences. Significantly higher absolute bias values of PCCO BCO compared with CCO BCO were also found early after operation in the ICU. Thus PCCO and CCO provided comparable measurements during coronary bypass surgery. After marked changes in SVR, further calibration of the PCCO device is necessary. Br J Anaesth 1999; 82: Keywords: measurement techniques, cardiac output; measurement techniques, pulse contour analysis; measurement techniques, thermodilution; surgery, cardiovascular Accepted for publication: October 19, 1998 Continuous measurement of cardiac output (CO) is preferable to intermittent measurement, especially during rapid changes in cardiovascular function in critically ill patients. A nearly continuous measurement of CO, using a thermodilution technique generating thermal pulses by a heating filament attached to a pulmonary artery catheter, produces a clinically acceptable level of accuracy compared with the intermittent bolus technique. 1 8 A new system of pulse contour analysis is available which is less invasive than continuous CO monitoring and may cost less. The method of calculating stroke volume from the contour of the arterial pressure curve dates from 1899 and was resurrected by advancements in computer technology in the 1980s. 9 Arterial pulse pressure waveform analysis involves measuring the area under the systolic portion of the arterial pulse wave from the end of diastole to the end of the ejection phase, together with an individual calibration factor to account for individual vascular impedance. Cardiac output, determined by arterial thermodilution, is used for calibration of this device which uses an enhanced version of the algorithm of Wesseling and colleagues for analysis of the arterial pulse pressure curve We have compared two methods of continuous CO measurement: pulse contour analysis (PiCCO, Pulsion Medical Systems, Munich, Germany) and a continuous thermodilution technique (Opti-Q SvO 2 / CCO, Abbott Critical Care Systems, Moutain View, CA, USA), with conventional intermittent thermodilution measurement in cardiac surgical patients with both good and poor ventricular performance. Patients and methods After obtaining approval from the Ethics Committee and written informed consent, we studied 26 patients (22 males), aged yr (mean 62 yr), undergoing elective coronary bypass surgery. Thirteen patients with an ejection fraction of 45% were allocated to group 1 and 13 patients with an ejection fraction 45% to group 2. Patients with valvular heart disease, intracardiac shunts or peripheral vascular disease were excluded. Preoperative ECG showed sinus rhythm without arrhythmia. Patients were anaesthetized with fentanyl, etomidate and pancuronium; fentanyl and isoflurane were used for maintenance of anaesthesia. Mechanical ventilation was set (FI O2 0.5, ventilatory frequency 10 bpm) to give an endtidal PCO 2 of kpa throughout the study. In all patients, a 4-French gauge arterial thermodilution catheter (PiCCO, Pulsion Medical Systems, Munich, Germany) was inserted via the femoral artery for clinical monitoring of arterial pressure, continuous CO (PCCO) measurements derived from arterial pressure and intermittent arterial thermodilution CO measurements. A pulmonary artery British Journal of Anaesthesia

2 Rödig et al. catheter (Opti-Q SvO 2 /CCO, 8F) was also inserted via an 8.5-French gauge introducer, into the right internal jugular vein for intermittent conventional pulmonary artery thermodilution CO and continuous CO (CCO) measurement. The arterial catheter was connected to a computer for pulse contour analysis (Pulsion Medical Systems, Munich, Germany). To calibrate this system, individual arterial input impedance to arterial pressure is calculated by simultaneously determining the area under the systolic portion of the arterial pulse wave (A sys ) and measuring arterial thermodilution CO. The mean of three subsequent measurements, randomized within the respiratory cycle, was used. These were performed by injection of 10 ml of iced saline solution via a central venous catheter with subsequent detection by the thermistor embedded into the wall of the arterial catheter. PCCO is computed by multiplying stroke volume by heart rate. A moving average of the preceding 30 s is calculated and displayed on the monitor. The passage of the PAC to the pulmonary artery, with its thermal filament connected to the cardiac computer (Q-vue CCO), was accomplished by monitoring the pressure waveform from the distal port of the catheter. This catheter has a lumen (DTPP lumen, distal thermal coil positioning port) at a distance of 1 cm from the tip of the thermal filament to check if its position is correct within the right ventricle by waveform analysis. Intracardiac and pulmonary artery pressures were monitored continuously to ensure correct placement of the catheter. The displayed CCO reflected either the average flow of the previous 3 min ( fast mode ) or of the previous s ( urgent mode ). Intermittent pulmonary artery thermodilution CO measurements (BCO) were performed by manual injection of iced saline solution. The average of three measurements, measured within a 10% range, distributed randomly over the respiratory cycle, was calculated. Furthermore, mixed venous blood (PAC tip) and rectal (Sirecust 1281 monitor, Siemens, Erlangen, Germany) temperatures were measured continuously and recorded at each time during cardiac surgery. A standard cardiopulmonary bypass (CPB) technique was used in all patients. Non-pulsatile perfusion at a pump flow rate of 2.4 litre min 1 m 2 was used during mild hypothermia (arterial temperatures C). Before disconnection from CPB, patients were rewarmed to a minimum rectal temperature of 36 C. PCCO, CCO and BCO measurements were compared after induction of anaesthesia, before surgical incision (T1), after the start of operation when dissection of the saphenous vein was performed (T2), after sternotomy during preparation of the internal mammary artery (T3), after pericardiotomy (T4), and 5 min (T5), 15 min (T6), 30 min (T7) and 45 min (T8) after weaning from CPB. Measurements were performed during stable surgical stimulation where the surgeons paused if necessary. After admission to the ICU, further sets of measurements were obtained during the subsequent 2 h (T9 T12). Sedation was not given in the postoperative period, according to the standard procedures of this institution. At each time, simultaneous readings of CO values, displayed on the pulse contour computer and on the Q-vue CCO-computer, were compared with BCO measurements obtained immediately after switching the Q-vue CCO-monitor from the continuous fast mode to the bolus mode. If PCCO changed by more than 15% during the pulmonary thermodilution period, CO values were rejected and comparison of measurements was repeated. Initial calibration of the pulse contour analysis system was performed before initiation of the test sequence at T1. This system was recalibrated immediately after weaning from CPB (before T5) and after admission to the ICU (before T9). In patients requiring intervention because of low systemic vascular resistance (MAP 55 mm Hg, CI 2.5 litre min 1 m 2 ), at the end of operation and before admission to the ICU, an i.v. infusion of phenylephrine was given in a standardized manner to restore arterial pressure to MAP values 80 mm Hg within approximately 20 min. PCCO (recalibrated), CCO (urgent mode) and BCO measurements were obtained before starting administration of phenylephrine, and after 20 min of administration, when an increase in arterial pressure to 80 mm Hg had been achieved. Statistical analysis was performed using the method described by Bland and Altman. 12 Bias between the methods was calculated as the mean difference between PCCO and BCO, and between CCO and BCO. The upper and lower limits of agreement were calculated as bias 2SD, and defined the range in which 95% of the differences between the methods were expected to lie. The precision of the bias analysis and limits of agreement were assessed using 95% confidence intervals (CI). The relative error, defined as 100 (BCO PCCO)/[(BCO PCCO)/2] and 100 (BCO CCO)/[(BCO CCO)/2], was calculated. To compensate for the variability of the samples, absolute values of these differences were also calculated. Student s t test for paired samples was used to compare bias, relative error at each time for groups 1 and 2, and inter-group comparisons. The Wilcoxon paired sample test was used to compare absolute differences which were not normally distributed. The level of significance was corrected according to Bonferroni to compensate for the effect of multiple comparisons. Bias between PCCO and BCO, and CCO and BCO, respectively, before and after 20 min of administration of phenylephrine, was analysed using the paired Student s t test. P 0.05 was considered significant. Results There were 308 comparative measurements performed between PCCO and BCO, and 312 measurements between CCO and BCO in 26 patients. In one patient, PCCO was not available from T5 to T8 because of prominent diastolic waves recorded in the iliac artery which the pulse contour analysis monitor was unable to differentiate from the systolic peak. The range of PCCO values was litre min 1 526

3 Pulse contour vs thermodilution cardiac output Table 1 Pulse contour cardiac output (PCCO) and continuous cardiac output (CCO) measurements (mean (SD)) at sample times T1 T12 in group 1 (ejection fraction 45%) and group 2 (ejection fraction 45%) Group 1 Group 2 PCCO CCO PCCO CCO (litre min 1 ) (litre min 1 ) (litre min 1 ) (litre min 1 ) T1 4.5 (1.8) 4.1 (1.6) 3.9 (1.4) 3.4 (1.1) T2 4.4 (1.6) 4.7 (1.3) 4.3 (1.4) 3.8 (1.1) T3 4.2 (1.9) 4.4 (1.5) 4.3 (1.0) 4.2 (1.1) T4 4.3 (1.7) 4.6 (1.5) 4.2 (0.9) 3.9 (0.8) T5 6.8 (2.4) 6.5 (2.2) 6.7 (1.5) 6.6 (1.6) T6 6.2 (2.1) 6.4 (1.9) 6.6 (1.3) 6.9 (1.6) T7 6.1 (2.4) 6.6 (1.6) 6.5 (1.0) 6.9 (1.7) T8 6.0 (2.0) 6.5 (2.0) 6.6 (1.0) 6.4 (1.3) T9 6.7 (1.5) 6.6 (1.8) 6.4 (1.2 ) 6.8 (1.4) T (1.5) 6.6 (1.6) 7.0 (2.2) 6.9 (1.4) T (2.3) 7.0 (1.4) 7.0 (1.2) 6.9 (1.4) T (2.1) 6.8 (1.4) 7.0 (1.3) 7.4 (1.4) in group 1 and litre min 1 in group 2. CCO values were litre min 1 in group 1 and litre min 1 in group 2. Mean CO values, measured by PCCO and CCO for the different sample times in both groups, are presented in Table 1. There were no significant differences between CO values, assessed by PCCO and CCO, at any time in either group, nor in PCCO and CCO measurements, respectively, between groups. Differences between PCCO and BCO, and between CCO and BCO, were plotted against the mean of the two methods in both groups (Fig. 1). The results of the analysis of agreement, assessed by bias, and the distribution of the observed differences, indicated by bias 2SD as upper and lower limits of agreement, including 95% confidence intervals at each time, are presented in Tables 2 and 3. Differences between PCCO and BCO values were not significantly different from differences between CCO and BCO at any time in either group. Analysis of the absolute values of the differences showed significantly higher biases of PCCO BCO compared with CCO BCO at T10 and T12 in group 1 and at T12 in group 2. The relative error was within 15% for 64% of comparisons between PCCO and BCO in group 1, and for 76% in group 2, compared with 86% in group 1 and 80% in group 2 of the differences between CCO and BCO. The absolute values of the relative error were significantly higher for PCCO and BCO compared with CCO and BCO measurements at T10 and T12 in group 1 and at T12 in group 2. Ten patients received phenylephrine to restore arterial pressure as a result of low systemic vascular resistance (SVR). Phenylephrine increased SVR by 62 (SD 29)% (from 732 (190) to 1153 (264) dyn s cm 5 m 2 ) after 20 min of administration. CO, assessed by the bolus technique, did not change significantly 20 min after administration of phenylephrine (5.7 (1.1) vs 6.1 (0.9) litre min 1 ). PCCO BCO bias did not differ from CCO BCO bias before the start of administration of phenylephrine. Phenylephrineinduced increases in SVR resulted in significantly higher PCCO BCO bias compared with CCO BCO after 20 min (see Fig. 2). From T1 to T4, mean pulmonary artery blood temperature (35.5 (SD 0.6) to 34.9 (0.5) C in group 1 and 35.7 (0.4) to 35.0 (0.6) C in group 2) and rectal temperature (35.8 (0.8) to 35.3 (0.7) C in group 1 and 35.7 (0.9) to 35.1 (1.0) C in group 2) decreased in both groups. After CPB, mean blood temperature (37.1 (0.4) to 36.6 (0.4) C in group 1 and 37.1 (0.7) to 36.6 (0.5) C in group 2) and rectal temperature (37.0 (0.9) to 36.4 (0.9) C in group 1 and 36.8 (0.7) to 36.3 (0.9) C in group 2) decreased until the end of operation. After admission to the ICU, only pulmonary blood temperatures were recorded. They increased during the first 2 h (from T9 to T12) in the ICU (36.4 (0.7) to 36.9 (0.8) C in group 1 and 36.5 (0.7) to 36.9 (1.0) C in group 2). The 26 patients were operated on in a uniform manner. Mean duration of CPB was 105 (SD 32) min in group 1 and 105 (20) min in group 2. Aortic cross-clamp time was 65 (18) min and 64 (15) min, respectively. A mean of 3.5 bypasses were constructed. In all patients the left internal mammary artery was dissected in order for it to be used as one of the bypass vessels. Weaning from CPB was achieved in all patients by administration of dopamine 3 7 g kg 1 min 1. Seven patients in group 2 required additional infusions of epinephrine to increase inotropic support. All patients had sinus rhythm after weaning from CPB. Volume infusions were administered in accordance with individual patient needs via a peripheral venous cannula. No adverse effects of either the PiCCO catheter or the Opti-Q SvO 2 /CCO catheter were observed. Discussion Previous studies have reported a small mean difference (bias) and limits of agreement (bias 2SD), considered clinically acceptable, between CCO, based on the pulsed warm thermodilution technique, and BCO measurements. 1 8 The BCO technique was used as the reference because it is a widespread clinical means of measuring CO despite its poor reproducibility and a variability of 15% under clinical situations Our results comparing these two methods in cardiac surgical patients with good and reduced ventricular performance were similar to others, 1 7 including early after mild hypothermic bypass. 8 However, factors such as uneven rewarming after CPB, indicated by opposite changes in blood and rectal temperatures (not the case in our patients, who were rewarmed to a minimum rectal temperature of 36 C at the end of CPB) or delivery of large volumes of cold fluids may affect the reliability of the continuous thermodilution CO measurement Arterial pulse contour analysis may be a less invasive alternative requiring an arterial thermodilution catheter and a central venous line for indicator injection. 20 The Wesseling algorithm, 10 which involves measuring the area under the systolic portion of the arterial pulse wave (A sys ) from the end of diastole to the end of the ejection phase divided by 527

4 Rödig et al. Fig 1 Bland and Altman plot for comparison between PCCO and BCO, and between CCO and BCO in groups 1 and 2 for all measurements. Solid lines represent bias 2SD. Table 2 Mean difference between PCCO BCO and between CCO BCO (bias), and lower and upper limits of agreement (bias 2SD), together with 95% confidence intervals (in parentheses) at sample times T1 T12 in group 1 (ejection fraction 45%) PCCO BCO (litre min 1 ) CCO BCO (litre min 1 ) Bias Lower limit Upper limit Bias Lower limit Upper limit T ( ) 2.27 ( 2.60 to 1.90) 2.89 ( ) 0.02 ( ) 1.59 ( 1.83 to 1.35) 1.55 ( ) T ( ) 1.81 ( 2.05 to 1.57) 1.45 ( ) 0.29 ( ) 0.96 ( 1.13 to 0.79) 1.54 ( ) T ( ) 1.86 ( 2.10 to 1.62) 1.46 ( ) 0.01 ( ) 1.65 ( 1.89 to 1.41) 1.63 ( ) T ( ) 1.62 ( 1.84 to 1.40) 1.24 ( ) 0.18 ( ) 0.70 ( 0.85 to 0.55) 1.06 ( ) T ( ) 2.06 ( 2.34 to 1.78) 1.98 ( ) 0.30 ( ) 1.80 ( 2.00 to 1.58) 1.20 ( ) T ( ) 1.85 ( 2.13 to 1.57) 2.23 ( ) 0.07 ( ) 2.03 ( 2.34 to 1.72) 2.20 ( ) T ( ) 3.25 ( 3.69 to 2.81) 2.63 ( ) 0.13 ( ) 1.85 ( 2.13 to 1.57) 2.11 ( ) T ( ) 3.21 ( 3.67 to 2.75) 3.11 ( ) 0.29 ( ) 1.12 ( 1.32 to 0.92) 1.70 ( ) T ( ) 1.31 ( 1.57 to 1.05) 2.15 ( ) 0.31 ( ) 1.26 ( 1.50 to 1.02) 1.88 ( ) T ( ) 2.37 ( 2.89 to 1.85) 4.77 ( ) 0.02 ( ) 0.78 ( 0.89 to 0.67) 0.82 ( ) T ( ) 2.94 ( 3.46 to 2.42) 4.42 ( ) 0.14 ( ) 1.37 ( 1.59 to 1.15) 1.32 ( ) T ( ) 2.94 ( 3.42 to 2.46) 3.52 ( ) 0.28 ( ) 0.88 ( 0.99 to ( ) aortic impedance (Z ao ) to provide a measure of stroke volume, has proved superior to other approaches in calculating CO from the contour of the arterial pulse wave. 21 In an enhanced version, Wesseling and colleagues 11 based the pressure flow relationship on a three-element model of arterial input impedance to arterial pressure to compute aortic flow from pressure. The algorithm uses the three major properties of the aorta and arterial system (i.e. aortic characteristic impedance, windkessel compliance of the arterial system and peripheral vascular resistance) to com- pute a flow pulsation from an arterial pressure pulsation. Wesseling and colleagues reported a bias of 0.09 litre min 1 and precision (SD) of 0.36 in eight cardiac surgical patients. 11 Others, using the above approach, found biases of 0.06 (0.93) litre min 1, (1.56) litre min 1 21 and 0.09 (0.85) litre min 1 23 in intensive care patients. The results of PCCO measurement improved to a bias of litre min 1 and a precision of 0.56 litre min 1 if additional calibration was used. 23 A radial artery was used for pressure monitoring in all patients, except for five patients where a femoral 528

5 Pulse contour vs thermodilution cardiac output Table 3 Mean difference between PCCO BCO and between CCO BCO (bias), and lower and upper limits of agreement (bias 2SD), together with 95% confidence intervals (in parentheses) at sample times T1 T12 in group 2 (ejection fraction 45%) PCCO BCO (litre min 1 ) CCO BCO (litre min 1 ) Bias Lower limit Upper limit Bias Lower limit Upper limit T ( ) 1.16 ( 1.40 to 0.92) 2.04 ( ) 0.09 ( ) 1.49 ( 1.69 to 1.29) 1.31 ( ) T ( ) 1.62 ( 1.93 to 1.31) 2.58 ( ) 0.10 ( ) 1.35 ( 1.52 to 1.18) 1.15 ( ) T ( ) 0.99 ( 1.19 to 0.79) 1.67 ( ) 0.20 ( ) 0.96 ( 1.13 to 0.79) 1.36 ( ) T ( ) 0.85 ( 1.00 to 0.70) 1.31 ( ) 0.05 ( ) 1.27 ( 1.44 to 1.10) 1.17 ( ) T ( ) 2.34 ( 2.69 to 1.99) 2.52 ( ) 0.02 ( ) 1.41 ( 1.63 to 1.20) 1.45 ( ) T ( ) 2.50 ( 2.85 to 2.15) 2.18 ( ) 0.05 ( ) 1.75 ( 2.01 to 1.49) 1.85 ( ) T ( ) 2.01 ( 2.29 to 1.73) 1.83 ( ) 0.29 ( ) 1.95 ( 2.28 to 1.62) 2.53 ( ) T ( ) 1.87 ( 3.67 to 2.75) 3.11 ( ) 0.10 ( ) 2.22 ( 2.53 to 1.91) 2.02 ( ) T ( ) 1.86 ( 2.12 to 1.60) 1.82 ( ) 0.42 ( ) 0.73 ( 0.90 to 0.56) 1.57 ( ) T ( ) 2.66 ( 3.10 to 2.22) 3.32 ( ) 0.22 ( ) 1.29 ( 1.51 to 1.07) 1.73 ( ) T ( ) 2.88 ( 3.29 to 2.47) 2.68 ( ) 0.29 ( ) 2.19 ( 2.45 to 1.93) 1.61 ( ) T ( ) 3.54 ( 4.04 to 3.04) 3.26 ( ) 0.19 ( ) 0.76 ( 0.89 to 0.63) 1.14 ( ) Fig 2 PCCO BCO and CCO BCO bias ( SD) before and 20 min after the start of administration of phenylephrine in 10 patients. *P Broken line represents an ideal bias of zero. artery was used. 23 In all of the studies, concern was expressed as to whether greater degrees of hyper- or hypotension ( 25% 22 or 50% 23 ) would result in significant alterations in arterial input impedance to arterial pressure, requiring recalibration of the system. Aortic impedance and arterial compliance are non-linear, pressure-dependent properties of the arterial system. The third element, total systemic peripheral resistance (SVR), according to Wesseling s model of arterial impedance, is a time-varying property of the vascular bed. As actual aortic impedance, arterial compliance and peripheral vascular resistance for an individual patient are unknown, CO, determined by a separate method, is used as a calibration factor. Measurement of CO by arterial thermodilution, used for calibration of the device we used, is a reliable alternative to conventional CO measurement via a pulmonary artery cather. 24 An enhanced version of the Wesseling algorithm, which has not yet been published by the manufacturer, was applied to the pulse contour analysis system we used with a refined correction factor to reduce the effects of changing mean arterial pressure on arterial impedance. Comparison of PCCO, provided by this new device, using femoral arterial access, to conventional BCO measurement showed a bias of litre min 1 and precision of 1.10 in experimental animals. 25 In cardiac surgical patients, Rauch and colleagues found a bias of (1.15) litre min 1 between PCCO and BCO compared with a value of 0.4 (1.25) litre min 1 between CCO and BCO. 26 Our results on the agreement, assessed by bias and upper and lower limits of agreement between PCCO and conventional BCO determination, were comparable with others When PCCO, recalibrated after CPB, and the CCO measurement were compared with standard BCO determination, they did not differ significantly during coronary bypass surgery, either in patients with good ventricular performance or in those requiring an additional infusion of epinephrine after weaning from CPB. Phenylephrine-induced increases in systemic vascular resistance by approximately 60%, however, resulted in a significantly greater PCCO BCO bias compared with CCO BCO. After operation, the absolute differences and absolute values of the relative error between PCCO and BCO were significantly higher than between CCO and BCO measurements at T10 and T12 in group 1 and at T12 in group 2. As sedation was not given in the postoperative period, short-term episodes such as increases in arterial pressure occurred during early recovery of anaesthesia, where patients underwent continuous mechanical ventilation. Our data suggest that rapid haemodynamic changes during the first 2 h in the ICU, such as increases in arterial pressure caused by increases in systemic vascular resistance, may account for the observed differences in the degree of agreement between PCCO and CCO, compared with BCO. The peripheral pulse wave reflects the interplay between left ventricular output and capacitance of the vascular tree. 27 The ability to trend changes in CO, despite changes in peripheral resistance, is ascribed to factors in the pulse contour algorithms that compensate for changes in arterial characteristics. We found that although changes in vascular tone of approximately 20% had no effect on the pulse contour method, greater changes in arterial pressure of 529

6 Rödig et al. 50% may affect PCCO measurement. When marked changes in systemic vascular resistance are expected after administration of vasoactive drugs or during recovery from anaesthesia, then our data indicate that the device should be recalibrated to compensate for changes in vascular resistance. In summary, assessment of continuous CO by pulse wave form analysis, and by a thermodilution technique, provided comparable measurements during coronary bypass surgery as long as significant changes in systemic vascular resistance did not occur. Recalibration of the PCCO device is necessary after marked changes in systemic vascular resistance. References 1 Böttiger BW, Soder M, Rauch H, et al. Semi-continuous injectate cardiac output measurement in intensive care patients after cardiac surgery. Intensive Care Med 1996; 22: Boldt J, Menges T, Wollbruck M, Hammermann H, Hempelmann G. Is continuous cardiac output measurement using thermodilution reliable in the critically ill patient? Crit Care Med 1994; 22: Burchell SA, Yu M, Takiguchi SA, Ohta RM, Myers SA. Evaluation of a continuous cardiac output and mixed venous oxygen saturation catheter in critically ill surgical patients. Crit Care Med 1997; 25: Jakobsen CJ, Melsen NC, Andresen EB. Continuous cardiac output measurements in the perioperative period. Acta Anaesthesiol Scand 1995; 39: Jacquet L, Hanique G, Glorieux D, Matte P, Goenen M. Analysis of the accuracy of continuous thermodilution cardiac output measurement. Intensive Care Med 1996; 22: Lefrant JY, Bruelle P, Ripart J, et al. Cardiac ouput measurement in critically ill patients: comparison of continuous and conventional thermodilution techniques. Can J Anaesth 1995; 42: Yelderman ML, Ramsya MA, Quinn MD, Paulsen AW, McKnown RC, Gillman PH. Continuous thermodilution cardiac output measurement in intensive care unit patients. J Cardiothorac Vasc Anesth 1992; 6: Rödig G, Keyl C, Liebold A, Hobbhahn J. Intra-operative evaluation of a continuous versus intermittent bolus thermodilution technique of cardiac output measurement in cardiac surgical patients. Eur J Anaesthesiol 1998; 15: Nelson LD. The new pulmonary arterial catheters. Crit Care Clin 1996; 12: Wesseling KH, de Wit B, Weber JAP, Ty Smith N. A simple device for the continuous measurement of cardiac output. Adv Cardiovasc Physiol 1983; 5: Wesseling KH, Jansen JRC, Settels JJ, Schreuder JJ. Computation of aortic flow from pressure in humans using a nonlinear, threeelement model. J Appl Physiol 1993; 74: Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: LeTulzo Y, Belgith M, Seguin P, et al. Reproducibility of thermodilution cardiac output in critically ill patients: comparison between bolus and continuous method. J Clin Monit 1996; 12: Taylor SH, Silke B. Is the measurement of cardiac output useful in clinical practice? Br J Anaesth 1988; 60: 90 8S 15 Stetz CW, Miller RG, Kelly GE, Raffin TA. Reliability of the thermodilution method in the determination of cardiac output in clinical practice. Am Rev Respir Dis 1982; 126: Latson TW, Whitten CW, O Flaherty D. Ventilation, thermal noise, and errors in cardiac output measurements after cardiopulmonary bypass. Anesthesiology 1993; 79: Nishikawa T, Dohi S. Errors in the measurement of cardiac output by thermodilution. Can J Anaesth 1993; 40: Böttiger BW, Rauch H, Böhrer H, et al. Continuous versus intermittent cardiac output measurement in cardiac surgical patients undergoing hypothermic cardiopulmonary bypass. J Cardiothorac Vasc Anesth 1995; 9: Böttiger BW, Sinner B, Motsch J, Bach A, Bauer H, Martin E. Continuous versus intermittent thermodilution cardiac output measurement during orthotopic liver transplantation. Anaesthesia 1997; 52: Gödje O, Höke K, Fischlein T, Vetter H, Reichart B. Less invasive, continuous cardiac-output measurement through pulse contour analysis versus conventional thermal dilution. Intensive Care Med 1996; 22: 58S 21 Apenburg C, Specht M. Continuous cardiac-output measurement by pulse contour analysis. Intensive Care Med 1996; 22 (Suppl. 3): Weissman C, Ornstein EJ, Young WL. Arterial pulse contour analysis trending of cardiac output. Hemodynamic manipulations during arteriovenous malformation resection. J Clin Monit 1993; 9: Irlbeck M, Forst H, Briegel J, Haller M, Peter K. Continuous measurement of cardiac output using pulse contour analysis. Anaesthesist 1995; 44: Spiegel T, Wietasch G, Bürsch J, Hoeft A. Cardiac output measurement by transpulmonary indicator dilution technique. Anaesthesist 1996; 45: Faltlhauser A, Erhardt W, Gödje O, Reichart B, Pfeiffer U. Accuracy of a less invasive device for rapid beat-to beat continuous cardiac output monitoring. Intensive Care Med 1996; 22: (Suppl. 3): Rauch H, Böttiger BW, Motsch J, Müller M, Fleischer F, Martin E. Pulse contour cardiac output measurement corresponds better with standard dilution technique than continuous thermodilution technique after hypothermic bypass. Br J Anaesth 1997; 78 (Suppl. 2): O Rourke MF, Yaginuma T. Wave reflections and the arterial pulse. Arch Intern Med 1984; 144: