Noninvasive versus invasive assessment of cardiac output after cardiac surgery: clinical validation

Noninvasive

Versus Invasive Assessment of Cardiac Output After Cardiac Surgery: Clinical Validation

Donat R. Spahn, MD, Edith R. Schmid,

MD, Mica Tornic, MD, Rolf Jenni, MD, Ludwig von Segesser,

Marko Turina, MD, and Andreas The accuracy of noninvasive cardiac output (CO) measurement techniques, such as electrical bioimpedance (BIO), suprasternal continuous-wave Doppler (CWD), pulsed-wave Doppler (PWD). and transesophageal continuous-wave Doppler (TED) ultrasound has been variably judged in recent years. In addition, clinical comparisons are hampered by the fact that there is no generally accepted gold standard in CO measurement. After coronary artery bypass surgery in 25 patients, CO was simultaneously determined by invasive standard techniques (thermodilution [TD] and Fick methods) plus BIO, CWD, PWD, and TED. There was an excellent agreement found between TD and the Fick method (CO, = r = 0.96; n = 99). Thermodilu0.13 + 1 .Ol * co,,: tion was thus chosen to be the reference method. Bioimpedance underestimated CO,, (CO,,, = 0.47 + 0.60 . CO,,; r = 0.76; n = 11 I). Allowing physiological ejection times only led to an improved agreement between BIO and TD (CO,,, = 0.05 + 0.69 co,,: r = 0.82; n = 79). but BIO still significantly underestimated

CO,, (P < O.ooO5). Using physiologic

H

MONITORING and EMODYNAMIC correction of abnormal hemodynamic parameters are prerequisites in the perioperative management of high-risk patients.’ Determination of cardiac output (CO) is of particular importance, since it cannot be estimated reliably based on clinical assessment,’ nor can it be derived from other hemodynamic parameters.3v4 Besides invasive methods to measure CO, such as thermodilution (TD), dye dilution, and the Fick method, alternative techniques exist to noninva-

From the Division of Cardiovascular Anesthesia. Institute of Anesthesiology, the Division of Cardiology, Department of Internal Medicine. Medical Policlinic, and the Clinic for Cardiovascular Surgery, University Hospital, Zurich, Switzerland. Portions of this work have been presented at the Annual Meeting of the European Association of Cardiothoracic Anaesthesiologsts. June S-10, 1988. Lyon, France; and the Annual Meeting of the American Society of Anesthesiologists, October 8-l 2. 1988. San Francisco, CA. Address reprint requests to Edith R. Schmid, IUD, Division of Cardiovascular Anesthesia, Institute of Anesthesiology. University Hospital, Raemistrasse 100. CH-8091 Zurich, Switzerland. 0 1990 by W.B. Saunders Company. 0888~296/90/0401-0009$03.00/0 46

Journal of

Baetscher,

MD,

MD

ejection times during CO,, determination reduced the scatter of data as compared with TD; however, CWD still considerably overestimated CO,,. when COcbv, computation was based on the echocardiographicaortic diameter (ECHO) (CO,, rc,,,, = 0.79 + 1.46 * co,,; r = 0.64: n = 52). With the surgical aortic diameter (SURG), the agreement improved = 0.75 + 1 .I6 - CO,,; r = 0.69: n = 44). (CO,,. S”RO but overestimation of CO,, remained significant (P < 0.05). Irrespective of the aortic diameter, CO,,, values showed a considerable scatter of data compared with CO,, (CO,, ECH,,= 1.26 + 0.60 . CO,,; r = 0.62; n = 64 and CO PWDSURO= 1.42 + 0.41 co,,; r = 0.47: n = 61). Correlation of absolute values to thermodilution depended on the CO,,, method used for calibration. All investigated noninvasive CO measurement techniques unreliably measured relative CO changes. Despite its invasiveness, TD remains the method of choice for accurate CO determination in adult patients following cardiac surgery. 0 1990 by W.B. Saunders Company.

sively determine CO, such as different Doppler methods and electrical bioimpedance (BIO). Although there is no generally accepted gold standard for CO assessment,5 the values measured by different invasive techniques show a satisfactory agreement. 64o In contrast, varying correlations

INDEX

OF ABBREVIATIONS

(Used Also as Subscripts)

EST TD F BIO CWD PWD TED TED,, irrET 200-350 ms 90%-110%

Estimated (clinical assessment) Thermodilution Fick method Electrical bioimpedance Suprasternal continuous-wave Doppler ultrasound Suprasternal pulsed-wave Doppler ultrasound Transesophageal continuous-wave Doppler ultrasound TED, calibrated by TD Irrespective of ejection time Ejection time 200 to 350 ms irrespective of the actual heart rate Ejection time 90% to 110% of the physiologic ejection time at the particular heart rate

Cardiofhoracic Anesthesia, Vol4,

No 1 (February), 1990: pp 46-59

NONINVASIVE

v INVASIVE CO ASSESSMENT

have been reported between noninvasively determined and invasively measured CO values in humans.“-** There have been few comparative investigations in patients after cardiac sur17.29-31.53 gery. The aim of the present study was to assess the clinical applicability, reproducibility, and accuracy of four different, simultaneously applied noninvasive CO measurement techniques compared with two invasive methods in patients after coronary artery bypass surgery. MATERIALS AND METHODS Following institutional approval, the study was performed after elective coronary artery bypass surgery. Twentyfive patients (4 women, 21 men) gave their informed consent. The mean age was 59 f 8 years (SD) (range, 43 to 72 years), and the mean preoperative ejection fraction was 62% + 12% (36% to 72%). In addition to the routine preoperative evaluation, an echocardiographic (ECHO) examination to measure the aortic diameter and to exclude aortic valve dysfunction or subvalvular aortic stenosis was performed. Exclusion criteria were (1) aortic valve dysfunction, artificial aortic valve, or subvalvular aortic stenosis; (2) intracardiac shunts; (3) dilatation of the ascending aorta; and (4) significant preoperative dysrhythmias, such as atria1 fibrillation. Total intravenous anesthesia (fentanyl, flunitrazepam, pancuronium) combined with controlled ventilation with an oxygen-air mixture was used for the operative procedures. A total of 3.6 + 0.9 (2 to 5) coronary artery bypasses were done. After the extracorporal circulation and in the early postoperative period, nitroglycerin, 0 to 6.8 pg/kg/min, was used in 23 patients, and dopamine, 0 to 6.3 pg/kg/min, was used in 4 patients. Postoperatively, the patients were sedated with nicomorphin, combined with flunitrazepam, and artificially ventilated. Simultaneous CO determinations by the Fick method, thermodilution (TD), suprasternal continuous-wave Doppler (CWD), pulsed-wave Doppler (PWD), transesophageal continuous-wave Doppler (TED) ultrasound, and electrical bioimpedance (BIO) were performed in the surgical intensive care unit within the first 6 to 8 hours postoperatively. Based on clinical assessment, one of the most experienced physicians estimated CO (COssr) immediately prior to the first set of CO measurements. The physician was aware of systemic and pulmonary arterial pressure as well as right- and leftsided filling pressures. Between two measurement periods, it was attempted to change CO by raising the filling pressures with volume replacement, varying the heart rate by atria1 pacing, or changing the infusion rate of vasodilators or catecholamines. When a new steady state was reached after 20 to 30 minutes, a further set of CO measurements was performed. During a single measurement period, a steady state with regard to oxygen consumption, hemodynamics, and sedation was required. At the beginning and end of each measurement period, mean arterial, pulmonary arterial, central venous, and pulmonary capillary wedge pressures were measured, and arterial and mixed venous blood samples

47

were analyzed for oxygen saturation, oxygen and carbon dioxide tensions, pH, and hemoglobin concentration. Stability of the on-line measurement of oxygen consumption, the adequacy of the state of sedation, and the absence of a significant difference of hemodynamic data over one measurement period were the basis of the assessment of steady state.3*-34 A pulmonary artery catheter (7SF; VIP Edwards, Santa Ana, CA) was inserted preoperatively through the internal jugular vein. Ten milliliters of cold 0.9% saline were injected manually with a closed injectate system (CO-set; Edwards) in 1 to 3 seconds. Cardiac output was determined by a CO computer (9520A; Edwards); the TD curve was displayed on a recorder for direct inspection. A single CO measurement was considered technically correct if the shape of the TD curve fulfilled the criteria of Levett and Replogle” and if the injectate temperature was below 10°C. Injection was not timed with the respiratory cycle, but was randomly distributed.8,9,36The CO,, value was expressed as the calculated mean of five single CO,, determinations during one measurement period. Fick CO (CO,) equals oxygen consumption divided by the arteriovenous oxygen content difference, which was determined by standard formulae.” Oxygen consumption was measured by means of indirect calorimetry (MMC Horizon; Sensor Medics, Anaheim, CA). Before each individual measurement series, the tightness of the respiratory circuit, consisting of the ventilator (Siemens Elema 900 B; Solna, Sweden), as well as the inspiratory and expiratory tubings, was tested in vitro. A rubber bag of 1.5 L volume was connected to the respiratory circuit. F,O, and F,O, were measured with the ventilator set at a minute volume of 6 L and a respiratory frequency of 20/minute, leading to peak pressures of 20 cm H,O. A maximal 0, difference, F,O, F,Or, of 0.001 was tolerated. Once connected to the patient, the endotracheal tube was cuffed tightly and all thoracic drainage tubes were examined for air leaks. Hemoglobin (Hgb) concentration as well as Hgb oxygen saturation were measured photometrically (Hemoxymeter OSM2; Radiometer, Copenhagen, Denmark). Arterial as well as mixed venous blood gas analyses were performed with an automatic gas check analyzer (AVL 940; AVL, SchaBhausen, Switzerland). The arterio-venous oxygen content difference was the mean of the values, determined at the beginning and end of each measurement period. Oxygen consumption was calculated as the mean value over one measurement period. Electrical BIO CO measurements were obtained using the NCCOM-3, Var 6 (BoMed, Hilliard, OH) continuous CO monitor. After careful skin preparation, eight ECG spot electrodes were situated, as described by Bernstein.” The electrode placement was considered correct if (1) the NCCOM-3 sensed every heart beat, (2) no noise signals were detected, and (3) adequate dZ/dt curves were recorded by a strip chart recorder. The electrically participating thoracic segment, defined by its length (L), was determined with the aid of the nomogram of Bernstein,” taking into account the patient’s deviation from ideal body weight. Cardiac output measured by BIO (CO,,,) was calculated as the mean of five single COslo values, determined simultaneously with CO,, during one measurement period. For a single COslo determi-

SPAHN ET AL

48

nation, the NCCOM-3 automatically averaged all accepted signals over a period of 10 seconds. Blood flow velocity in the ascending aorta was measured by a suprasternal approach using CWD (Accucom, Datascope; Paramus, NJ) as well as PWD ultrasound (Cardioflo; Cardionics, Houston, TX). The transmitted frequency of the CWD probe was 2.5 MHz. The backscattered frequency spectrum was transformed to mean Doppler frequency shift by a fast Fourier transformation. Blood flow velocity was averaged over 12 consecutive heart beats. The CWD displayed heart rate, signal level, and ejection time. A high-pitched Doppler sound without low-frequency noise, at the highest possible signal level, a heart rate equal to the heart rate of the ECG monitor, and an ejection time of 200 to 350 ms (as recommended by the manufacturer) were searched for by proper positioning of the suprasternal probe. As aortic diameter, the built-in nomogram, which estimated the inner diameter of the ascending aorta just superior to the sinus of Valsalva in relation to the patient’s height, weight, age, and sex, was initially used. In addition, the correlation of CO,,, to CO,, was compared when the nomogram diameter was changed to the diameter of the ascending aorta, which was measured by ECHO or surgically. Cardiac output was measured three times during one measurement period. The definition of CO,-wu was the mean of the two highest COcwo values obtained. The transmitted frequency of the PWD probe was 3.5 MHz. The mean frequency shift was determined by a zero-crossing counter analysis. The aiming protocol was the same as with the CWD. In addition, the velocity-time curve, written by a strip chart recorder, was analyzed. A rapid upstroke during systole, a distinct negative flow vector at the time of the aortic valve closure, and a positive flow during early diastole were sought. Blood flow velocity was averaged during an effective measuring period of 10 seconds. Cardiac output computation was based on the aortic diameter determined by ECHO. The esophageal Doppler probe (Accucom, Datascope) measured blood flow velocity in the descending aorta. The transmitted frequency of the transesophageal CWD probe was 2.5 MHz. The optimal probe position was searched for with the same aiming protocol as described for the suprasternal approach. When the optimal position was found, the transesophageal Doppler was calibrated by suprasternal CWD ultrasound. The highest valueof three COcwnmeasurements was entered as the calibration value. The displayed CO, measured by transesophageal Doppler (CO,,,), was averaged over 12 consecutive heart beats. The CO,,, value was expressed as the calculated mean value of five single CO,,, determinations during one measurement period. To test whether TED displayed relative CO changes correctly, TED was calibrated in a second group of 18 additional patients by TD before the individual measurement series. The internal aortic diameter was determined echocardiographically (ZD-Echo) at the height of the right pulmonary artery by the leading edge technique. In addition, the outer diameter of the ascending aorta was measured intraop eratively by the cardiac surgeons with a specially designed slide caliper. Furthermore, the circumference at the same location was measured by encircling the ascending aorta with a suture string. Because of the required mobilization of the

ascending aorta, its circumference was not measured in patients undergoing reoperation. The surgical measurements took place just before cannulating the patient for extracorporeal circulation. Aortic wall thickness was assumed to be 2 mm for all patients.” Cardiac output values were compared by linear regression analysis. For every comparison of CO values, determined by different techniques, there were two samples of CO values chosen: one with all available values and the second one with an equal number of measuring periods per patient. Both samples were tested with regard to significant difference by means of analysis of variance, 32The regression coefficient (b) and the y-axis intercept (a) of the regression line equation (v = n + bx) were tested with regard to significant deviation from the ideal values of 1 and 0, respectively, by the one-sample t test.33 Cardiac output changes from one measurement period to the next were tested with regard to a significant increase or decrease as follows: the SEM for each CO measurement technique was determined as the mean of all SEM values of the particular technique. According to Stetz et al,” two CO values, measured by the same technique, were considered to be significantly different (P < 0.05) if the difference between these two values was higher than 3 SEM. Furthermore, whether a particular CO measurement technique would display the same direction of CO changes as the reference method was examined. When a particular method displayed a significant CO change in the opposite direction to a significant CO,, change, the frequency of events was tested with regard to significant difference to zero events by means of the x2 test. RESULTS

Measurement periods lasted 11 + 3 minutes (4 to 20 minutes). Steady-state conditions were reached in 116 of 122 measurement periods. The deviations of hemodynamic and metabolic parameters at the beginning compared with the end of a steady-state period, expressed as coefficient of variation, were as follows: heart rate, 1.09% + 1.43% (0% to 8.57%); mean arterial pressure, 2.86% + 2.43% (0% to 11.71%); arteriovenous oxygen content difference, 3.36% f 2.83% (0% to 11.12%); and oxygen consumption, 1.5 1% + 1.42% (0% to 10.69%). The coefficient of variation was 12.37% to 16.69% in the remaining six measurement periods (unsteady state), and they were therefore not analyzed. Cardiac output values of five additional periods had to be rejected because of technical problems with TD. Thus, 111 measurement periods remained for further data analysis. In all comparisons between CO values, determined by different techniques, the analysis of variance showed that there were no differences among samples of CO values, whether all avail-

NONINVASIVE

v INVASIVE CO ASSESSMENT

49

Table 1. Linear Regression Equations of Absolute Cardiac Output Values as Compared With Thermodikttion y-Axis

Linear

InterceptRegression Method

n

(Llmin)

0.13

1.01

0.96

99

0.43

0.474

0.600

0.78

111

0.67

BlO,c,-, 10%

0.05

0.695

0.82

79

0.67

CWD,ws,,

1.52t

0.84

0.66

81

1.38

NOMO 0.90* ECHO 0.79

0.93

0.77

52

1.14

1.40t

0.84

52

1.32

1.16.

0.89

44

0.93

NOMO

10%SURG 0.754 1.79$ NOMO

CWDsw,

0.83

0.65

79

1.37

TEDsw, 10%NOMO

0.99*

0.96

0.76

49

1.21

P’A’D,c,o

1.26t

O.SO§

0.62

64

1.15

P’J%,R~

1.42t

0.41§

0.47

61

1.17

TED,,,,,

SEE=0.43

SEE

r

BfOirrt

CW’&-,I,,

n=99

Coefficient

Fick

CWUw+-,,,~

r=0.96

tL/min)

Abbreviations: NOMO, aortic diameter derived from nomo-

0

5

COTD

10

gram; ECHO, aortic diameter

determined by ECHO; SURG,

surgically measured aortic diameter.

I/min

Significantly greater or smaller than the ideal value of yaxis intercept (0) and linear regression coefficient (1). respectively;

-,

Fig 1. Fick cardiac output (CO,) versus TD (CO,,). Linear regression line:,----, line of identity.

able CO data or an equal number per patient were used (P > 0.05). Accordingly, the linear regression analyses were based on all available sets of CO values. In 22 of 25 patients (88%) the Fick method could be performed: Two patients had a bronchopleural leak; in one patient, it was impossible to reach the required tightness of the respiratory circuit. In 99 measurement periods in the remaining 22 patients, CO, and CO,, showed an excellent agreement (Fig 1 and Table 1). Cardiac output changes were also reflected equally well by the two methods (Fig 2 and Table 2). The Fick method did not show a significant change in the opposite direction from a significant CO,, change (Table 3). An additional in vitro study showed a close agreement between values, obtained simultaneously by TD and an electromagnetic flow measurement (CO&, calibrated by volumetric tank: Corn = 0.036 + COELM 0.998 (r = 0.998; n = 8; standard error of the estimate = 0.166; range, 1 to 8 L/min). Based on these data, TD was chosen as the reference method. In all 25 patients, it was possible to measure CO by BIO. When all CO,,, values were included in the linear regression analysis, irrespective of ejection time, BIOirrET underestimated CO,, in the whole range of 1.97 to 8.07 L/min (Fig 3 and Table 1). Allowing ejection times of

+P < 0.05; tP < 0.005;

$P < 0.001;

and §P < 0.0005.

90% to 110% of the physiologic value at the particular heart rate only, but including ejection times shorter than 90% in low-output states (cardiac index < 2.5 L/min/m2),39 led to an improved agreement between BI090ao-L108 and TD (Fig 3 and Table l), but BIO,,,_,,,, still distinctly underestimated CO,,. The slope of the regression line (b = 0.69) was still far smaller than the ideal value of 1.O (P < 0.0005). Relative CO changes were underestimated by BIOirrETas well as by BIO,%_,,, compared with TD (Fig 4 +50%

+25% S

Y h

E

0%

a

-50% / -50%

-25%

0%

+25%

+50%

A co,,% Fig 2. Relative CO changes (as percentage to previous value): Fick method (ACO,) versus TD (ACO,,). -_ Linear rearession line: - - --. line of identitv. ,

50

SPAHN ET AL

lo-

Table 2. Linear Regression Equation of Cardiac Output Changes Compared With Thermodilution Y-Axis Intercept (%I

Method Fick

0.56 -0.04

B LO,,,,,

Linear

Regression Coefficient

r

SEE f%)

n

0.95

0.87

77

9.00

0.754

0.57

86

17.64

BLD,, IOX

0.35

0.734

0.66

56

17.41

=WD,w,,, CWD 90110% PWD

0.19

0.71*

0.58

60

15.46

1.79

0.83

0.67

34

16.21

3.00

0.86

0.50

49

25.03

TED,,

2.16

0.67T

0.55

96

11.80

.E F E .-t

7

-

5-

0 z 8

?? P < 0.05. tP < 0.005.

and Table 2). In 4 of 86 cases (4.65%), BIOirraT showed a significant change in the opposite direction to a significant COro change (Table 3), and BI%-, 10%in 3 of 56 situations (5.63%). The inner aortic diameter (Fig 5) was determined correspondingly by the surgical outer diameter measurement (2.70 + 0.39 cm) as well as by the measurement of the aortic circumference (2.83 f 0.43 cm) (P > 0.05). For CWD and PWD, the surgical aortic diameter was defined as the inner diameter of the ascending aorta, determined by the surgical circumference measurement (assuming an aortic wall thickness of 2 mm). 31 The aortic diameter measured via ECHO overestimated the surgically determined values, particularly at aortic diameters below 3 cm (Fig 5). In the whole range of echocardiographically and surgically determined aortic diameters of 2.5 to 4.0 cm and 2.2 to 3.7 cm, the nomogram indicated a variation of only 2.41 to 2.97 cm and showed hardly any correlation with the measured aortic diameter values (Fig 5).

0

5

COTD

10 Vmin

Fig 3. Bioimpedance CO (CO,,) versus TD (CO,,) (A) irrespective of ejection time (BIO,,,,) and (B) 90% to 110% of physiologic ejection time (BIO,,,,). -, Linear regression line: - - --, line of identity.

Table 3. Significant Opposite Cardiac Output Changes Compared With Thermodilution CardiacOutputChanges Method

n

Opposite Significantf%)

Fick

77

0

B’D,,,ET

86

4.65

BID,,

56

5.63*

CWD,w,s,

60

8.33t

CWD,,, PWD

34

5.66.

49

10.20t

96

2.08

TED,,

10% ,096

NOTE. Opposite Significant (%) indicates percentage of situstions in which the investigated method displayed e significant CO change in the direction opposite a significant thermodilution CO change. Significantly different from 0%; +P < 0.05; tP = t0.01.

In 21 of 25 patients (84%) and 81 of 111 measurement periods (73%), respectively, CO,, measurements in the ascending aorta were possible with ejection times of 200 to 350 ms, as recommended by the manufacturer. Based on the nomogram diameter, CO,,, (ejection time, 200 to 350 ms) showed a distinct scatter and a slight CO overestimation compared with TD (Fig 6 and Table 1). Allowing physiologic ejection times only improved the correlation between CWD and TD (Fig 6 and Table 1). With this restriction, however, 29 measurement periods dropped out, and CO measurement by CWD was thus possible only in 52 of 111 measurement periods (47%).

NONINVASIVE v INVASIVE CO ASSESSMENT

51

With ejection times of 200 to 350 ms, irrespective of the actual heart rate, CWD unreliably displayed CO changes compared with TD (Fig 7 and Table 2). In 5 of 60 situations (8.33%), CWD*w-,sornSshowed a significant CO change in the opposite direction from a significant CO,, change (Table 3). Using physiologic ejection times improved the correlation between CWD and TD (Fig 7 and Table 2), but, in 2 of 34 cases still displayed a significant (5.88%), CWD,,,,,, change in the direction opposite a significant Corn change (Table 3). When CO,,, determination, restricted to physiologic ejection times, was based on the ECHO instead of the nomo-

ml%

4-c

0"

/' / / / / / /

2

A"-oMSURG

’

/’

-50% -50%

.

-25%

lbcl.00 pdJ.05)

‘1 0% A&o

~25%

+50%

-75%

3

4

CIRCcm

Fig 5. Determination of the inner diameter of the ascending aorta (AO-DM). (A) Surgical outer diameter measurement (AO-DM SURsJ versus surgical circurnferante (AO-DM SURsCIRC).(6) Echocardiographic (AD-DM,,,,) versus surgical circumference measurement (AODM svRs CIRC).(Cj Nomogram (AO-DM,,,,) versus surgical circumference measurement (AO-DM,,,, C,RCJ. -, Linear regression line; ----, line of identity.

%

Fig 4. Relative CO changes (as percentage to previous value): BIO (ACO,,,) versus TD (ACO,,) (A) irrespective of ejection time (ABIO,,,,,j and (6) 90% to 110% of physiologic ejection time (ABIOs0,,09L).-, Linear regression line: ----, line of identity.

gram aortic diameter, a better correlation to Corn resulted, but CO,, was consistently overestimated (Fig 8 and Table 1). The regression coefficient (b = 1.40) was significantly higher than the ideal value of 1.O (P < 0.005). With the

SPAHN ET AL

52

surgical aortic diameter, a quite high correlation coefficient (r = 0.89) between CWD and TD was reached. The slope of the regression line (b = 1.16), however, was still higher than 1.O (P < 0.05). In contrast, the correlation of relative CO changes, determined by CWD and TD, was not influenced by the aortic diameter on which the COcwD computation was based.* An acceptable suprasternal pulsed-wave Doppler (PWD) signal could be obtained in only 15 of 25 patients (60%) and in 64 of 111 measurement periods (58%). The CO,, value showed a distinct scatter of data compared with irrespective of the aortic diameter, on coTD, which the COrwn computation was based (Fig 9 and Table 1). Cardiac output changes were also unreliably displayed (Fig 10 and Table 2). In 5 of 49 instances (10.2%), PWD showed significant CO changes in the direction opposite significant Corn changes (Table 3). Continuous-wave transesophageal CO determination (CO,) was not possible in two measurement periods because of system failure. When calibrated before each measurement period by a COcwD measurement with an ejection time of 200 to 350 ms, irrespective of the corresponding heart rate, TED distinctly overestimated CO,, (Fig 11 and Table 1). With ejection times of the calibration measurement within 90% to 110% of the physiologic value, an improved correlation of COrnn to Corn resulted (Fig 11 and Table 1). In 15 of 18 patients (83%) from the second study group, in which TED was calibrated by TD before the individual measurement series, stable Doppler signals could be obtained by the esophageal probe. One hundred eleven comparative CO determinations were performed, and 96 relative CO changes were analyzed. Changes in CO,, were underestimated by TED (Fig 12 and

*Relative COcwo change = (COCWO, - cOcwo*)/COcwo, = (V,

. CSA - iQ. CSA)/(I,

. CSA)

= (V, - V,)/V, where CO,,,,, CO,,,, = CO,,, at measuring period 1 and 2, respectively; 8,, V2 = mean blood flow velocity at measuring period 1 and 2, respectively; and CSA = cross-sectional area of ascending aorta on which the COcwD computation was based.

10

5

0

CO,,

Vmin

Fig 6. Suprasternal CWD CO, based on the CWD nomogram aortic diameter (CO CWD-0 j versus TD (CO,,) with (A) ejection time of 200 to 350 ms and (B) 90% to 110% of the physiologic value. -, Linear regression line; ----, line of identity.

Table 2). The correlation of CO,, changes to Corn changes showed a considerable scatter of data, and individual correlation coefficients ranged from -0.13 to 0.96. In 2 of 96 situations (2.08%), CO changes in direction opposite TD were displayed by TED (Table 3). Although CO was estimated by two experienced cardiovascular anesthesiologists, there was only a faint correlation of CO, to CO,, (Fig 13). Low CO values were systematically overestimated, and CO values above 6 L/min were increasingly underestimated.

NONINVASIVE

v INVASIVE CO ASSESSMENT

All investigated CO measurement techniques showed a comparable reproducibility. The average SEM of the different techniques was SEM,, = 2.55%; SEM, = 2.73%; SEM,to = 2.94%; SEMcwD = 2.33%; SEM,,n = 3.03%; and SEMrsn = 2.07% and 1.44% in the second group. DISCUSSION

In the present study, four noninvasive CO measurement methods, BIO and various Doppler techniques, such as the suprasternal CWD, PWD, and TED were investigated simultaneously with two invasive techniques, TD and the Fick method.

/ .’ /

/

/

/

/

/

/

/

0

5

10

C0l-j~ I/min Fig 8. Supraeternal CWD (CO,,j. restricted to physiologic ejection times versus TD (CO,,) with special

-50%

0% A&O

+500/a

+lOO%

96

Fig 7. Relative CO changes (as percentage to versus TD previous value): suprasternal CWD (ACO,,,) (ACO,,) with (A) ejection time of 200 to 350 ms, irrespective of heart rate (ACWD,, ,,_I, and (6) 90% to 110% of physiologic ejection time (ACWD,,,,,). -, Linear regression line; ----,

~55 based. IA) Echocardiographically determined aortic diameter. (B) Surgically determined aortio diameter. -, Linear regression line; ----, line of identity.

tion

,/

line of identity.

Based on an excellent agreement with the Fick method and a perfect correlation with an electromagnetic flow measurement in an in vitro study, thermodilution was chosen as the reference method for the comparison of reliability and accuracy of the noninvasive techniques.

SPAHN ET AL

54

mean of the 22 CO values below 3.50 L/min was 2.67 f 0.43 L/min with TD and 2.76 + 0.42 L/min with the Fick method. Cardiac output changes were also reflected reliably by both invasive methods (Fig 2 and Table 3).4’*47 During mechanical ventilation with an inspiratory oxygen fraction higher than 0.21, the tightness of the ventilator-patient circuit was of major importance, since any leak caused an exchange of room air with the gas mixture within the circuit. Thus, a strict test of tightness of the respiratory circuit is essential for the clinical use of the Fick principle for CO measurement.40 Consequently, in patients with bronchopleural leakage, the Fick principle is not applicable.

y=l.26+0.6O.x

Bioimpedance

. s’

/

0

/

/

/

I_‘1

0

/

,‘s .

?? .

1.r

1

.

?? / ?? * , 0s . . 43

1

.

I

The NCCOM-3 device is based on the methodology originally described by Kubicek et a142and modified by Bernstein.38 Bioimpedance was easy to handle, and COBI measurements were possible in all patients during all measurement periods. Correlation coefficients (CO,10 versus CO-& of 0.78 (BIOirrar) and 0.82 (BIO,,,,,,,) (Fig 3) were comparable to prior studies with r values of 0.63 to 0.85,11~‘3*42~49~50 although correlation coefficients higher than 0.85’4*‘s*26 and lower than 0.6328 have been reported. A general CO underestimation by BIO, as was found in this study, has not been reported previously, but a reduced agreement between

??

??

. . ,

I

. I

5

COTO

I

I

,

10 I/min

Fig 9. Suprasternal PWD (CDrwo) versus TD (CD,,) with special reference to the aortic diameter on which the computation was based. (A) Echocardiographically CO,, determined aortic diameter. (B) Surgically determined aortic diameter. -, Linear regression line: ----, line of identity.

Fick Versus Thermodilution

Versus Thermodilution

s

+50%

s Q

I

Methods

The excellent agreement of CO, and CO,, in the whole range of 1.97 to 8.09 L/min (Fig 1) was in accordance with the findings of Hillis et al in 252 patients during cardiac catheterization.6 There was no systematic CO overestimation by TD at flow rates below 3.50 L/min, as was found by van Grondelle et al7 and Davis et a1.40The

-50%

0% ACOTD

+50%

+lOO%

96

Fig 10. Relative CO changes (as percentage to previous value): suprasternal PWD IACO,,,) versus TD (ACO,,). -, Linear regression line; ----, line of identity.

NONINVASIVE

v INVASIVE CO ASSESSMENT

55

+SO% -

+40%

y=1.79+0.83

/

.*

-x

r=0.65,n=79 SEE=1.37 a>0

p
-40%

-20%

0%

+20%

+40%

+60%

ACOT,% Fig 12. Relative CO changes (as percentage to previous value): TED (ACO,,,) versus TD (ACO,,). Transesophageal Doppler was calibrated by TD before the individual measurement series. -. Linear regression line; ----, line of identity.

r= 0.76, n=49 SEE=1.21 a>0 ~'0.05

Z0 = base impedance. After coronary artery bypass surgery, the thoracic fluid content is likely to be increased. The correspondingly lowered base impedance, Z, = 23.5 f 2.6 ohms (normal, 24 to 45 ohms), cannot explain the CO,, underestimation by BIO. The considerable CO, underestimation can be explained by a distinct underestimation of the maximum rate of impedance change (dZ/dt) or by a slight underestimation of the distance between the voltage-sensing elec-

COT0 I/min Fig 11. Transesophageal Doppler CO (CO,,,) versus TD (CO,,) with special reference to the ejection time of the COcwo calibration (aortic diameter, CWD nomogram): (A) ejection time of suprasternal CWD calibration is 200 to 3W ms and (gj ejection time of suprasternal CWD celibrstion is 90% to 110% of the physiologic value. -, Lineer regression line; - - --, line of identity.

CO,,o and CO, was described during artificial ventilation.” The stroke volume determined by BIO is as follows:

r=0.53 n=25 SEE=0.61 oy, 0

SVruo = (dZ/dtMAx . LVET . L3)/(4.25 . Z,) where dZ/dtMAx = maximum rate of impedance change, LVET = left ventricular ejection time, L = length of the truncated cone (distance from the base of the neck to the xiphoid process),38 and

I

,

be1.00 p.=O.O005 1 ,I1 , , , , , 5 10

COTD I/min -,

Fig 13. Estimated (CO,,,) versus TD CO (CO,,). Linear regression line; ----, line of identity.

56

trodes (L) by the Bernstein nomogram,38 because the L value enters the COaro computation in the third power. In an additional series of 20 patients, the distance from the base of the neck to the xiphoid process was measured, and a perfect agreement with the nomogram was found. Thus, BIO seems to considerably underestimate the maximum rate of impedance change. Besides an intraoperative cardiac study by Siegel et al, **BIO has been reported to correctly display relative CO changes.14~26~42*49~50 This judgment was either based on high correlation coefficients in linear regression analyses of CO changes (r = 0.77 to 0.87) or on corresponding mean CO changes of whole patient groups between difHowever, in 3 of 10 and ferent study periods. 42V49 in 4 of 10 situations, COBI changes were in the opposite direction from the values obtained by dye dilution49; even grouped CO changes from one particular study period to the next were in the direction opposite from the invasive reference method. I4 In the original report of Kubicek et aL4* 5 of 10 subjects showed individual ratios of CO values, determined by BIO and dye dilution, which varied between different study periods from distinctly smaller than 1 to values considerably higher than 1 (range, 0.63 to 1.51). It is therefore unlikely that individual CO changes were always displayed correspondingly by BIO and the reference method. Such misleading information occurred in 4.65% (BIOirrET) and 5.63% (B1090-1,096),respectively, in the present study (Table 3). Doppler Techniques Versus Thermodilution

The volume flow through the ascending aorta at time t corresponds to mean flow velocity times cross-sectional area. The Doppler-derived stroke volume equals the time integral of volume flow over the ejection period.43 Hence, there are three essential parameters to be precisely determined in accurate Doppler CO assessment: the cross-sectional area of the ascending aorta, blood flow velocity, and ejection time (ET). When Doppler CO was determined for data with ET ~90% of physiologically accepted values,39 CO was underestimated; while for ET >l 10% of physiologic values, it was distinctly overestimated. Selecting data with ET in the physiologic range greatly increased the time required for positioning of the Doppler probe. Although periods as long as 90 minutes were used for supraster-

SPAHN ET AL

nal probe positioning, physiologic ETs were achieved in only 52 of 111 COcwD measurements (47%). This compared with 81 of 111 (73%) ET values in the range of 200 to 350 ms, ie, the range recommended by the manufacturer of the TED device as acceptable for “usual” heart rates. The former restriction, however, decreased data scatter and improved the correlation between COcwD and CO,, (Fig 6). The aortic diameter is usually determined by ECHO. The accuracy of this measurement technique, however, has rarely been tested. Few comparative surgical aortic diameter determinations have been performed, which in part showed an acceptable agreement with 7 of 7 and 25 of 31 measurements within a difference of 2 mm.29*43 Mark et al reported a distinct scatter of aortic diameter values, determined surgically and via ECHO (r = 0.31).31 In the present study, the aortic diameter was determined surgically, 3 to 4 cm above the aortic valve, either by outer diameter measurement by means of a slide caliper or by measuring the circumference that showed a close agreement (Fig 5). However, ECHO measurement, at the height of the right pulmonary artery, showed an increasing overestimation of the aortic diameter in the low-diameter range (Fig 5). Although the exact location of the aortic diameter measurement might have varied slightly between these techniques, it appears that the ECHO method overestimated the true aortic diameter because the highest correlation of to CO,, was reached, when the COcwD %WD computation was based on the surgically measured aortic diameter (Fig 8). A further problem, which may impair the reliability of CO determination by Doppler ultrasound, is the fact that the aortic diameter is known to change with varying hemodynamic situations.45 This could only be overcome by assessment of the aortic diameter before each individual CO determination. The third essential parameter is the precise determination of a representative blood flow velocity. The investigated suprasternal Doppler techniques measure the blood flow velocity only at one particular location in the ascending aorta, but the velocity profiles in the ascending aorta are by no means blunt. 46-48The temporal mean velocity across the ascending aorta, o(r), in relation to the true mean flow velocity, o(tm), came to O(r)/V(tm) ratios of 2.35 to 0.1646 and

NONINVASIVE

v INVASIVE CO ASSESSMENT

1.7 1 to 0.60,48 respectively. In addition, Lucas et a147 found the velocity p rofiles to be COdependent. Compared with the small sample volume of the PWD, the CWD technique detects a larger area of different blood flow velocities across and along the aorta. A larger part of the velocity profile may thus be covered by the ultrasound beam with the CWD technique. This leads to a certain averaging of the measured blood flow velocity and to partial inclusion of the highest blood flow velocities across the velocity profile. These highest velocities, however, determine the envelope of the backscattered frequency spectrum; thus, overestimation of the mean blood flow velocity and the corresponding COcwn may result. In the PWD method, the aiming protocol for a proper probe position was also to search for the highest flow velocity in the ascending aorta. With the small sample volume, a distinct CO overestimation compared with TD would be expected. The present study, however, basically showed a considerable scatter of data and no CO overestimation (Fig 9). Thus, the exact determination of the true mean flow velocity in the ascending aorta is one of the main problems in the PWD technology in adult patients. The true angle 0 will remain unknown, even in combination with 2D-Echo, because the vector deviation in the plane perpendicular to the 2Dplane cannot be estimated. However, any deviation of the angle 6 from this ideal value of O” would cause the Doppler technique to underestimate CO. Several investigators have reported an acceptable agreement of the Doppler-derived CO changes to those measured invasively.‘6V2’-23 Thus, Doppler techniques have been proposed to be valid trend monitors. Although this was true for grouped measurements, there were individual situations in which the Doppler-derived CO change was in direction opposite TD. Apart from a considerable scatter of CO changes, determined by CWD or PWD compared with TD (Figs 7 and 10; Table 2), such misleading results occurred in 5.88% to 12.28% (Table 3). The precise location of the velocity measurement across the aortic cross-sectional area cannot be held constant at repeated CO determinations. Because there are local temporal mean velocityto-true mean flow velocity ratios of 0.16 to 2.35,46 deceptive Doppler CO changes or significant CO changes in the opposite direction to real CO

57

changes are by no means astonishing. On the other hand, there was no significant CO, change in the opposite direction from TD (Table 3). The COTED value, calibrated within each measurement period by suprasternal CWD, showed a close agreement to COcwn. Due to this actual calibration procedure, the ability of the TED to display the CO trend correctly could not be assessed in the first group of patients. In the second group, in which TED was calibrated by thermodilution once before the individual measurement series, a considerable scatter of CO changes determined by TED compared with TD resulted in heavily sedated patients (Fig 12 and Table 2). This confirms a previous study, assessing the ability of TED to track trends in anesthetized cardiovascular patients, in which only changes higher than 4 L/min achieved coTED 95% confidence for a change in CO,,.** Additionally, every movement of the patient may change the esophageal probe position relative to the descending aorta with correspondingly altered blood flow velocity measurement. For complete hemodynamic monitoring, CO, right- and left-sided filling pressures, systemic and pulmonary arterial pressures and mixed venous oxygen saturation of Hgb are of major importance. Together with derived values, such as systemic and pulmonary vascular resistances, left and right ventricular stroke work index, and clinical assessment, a thorough evaluation of the cardiovascular status becomes possible. The ideal CO measurement method should be easy to handle, exhibit a minimal threat to the patient, and display reliable CO values and CO changes. In addition, a continuous CO measurement technique would be desirable. Although the flow-direeted pulmonary artery catheter exhibits a certain morbidity and even mortality, ” this system best fulfills the above-stated criteria of advanced hemodynamic monitoring. The CO,, and CO, values showed an excellent agreement (Fig l), and the hemodynamic management would not have been misled by an aberrant result. With modified TD and Fick methods, continuous CO monitoring has been described.N*52 The prerequisite for reliable CO, determination, however, is the accurate measurement of oxygen consumption, which in turn necessitates a tight ventilator-patient circuit, which may be quite difficult to achieve in the individual patient.

SPAHN ET AL

58

In summary, noninvasively measured CO values and CO changes showed a considerable scatter of results compared with thermodilution (Figs 3- 13). Therapeutic interventions would have been misled by noninvasively determined CO changes in the opposite direction of actual CO changes in a considerable number of situations (4.65% to 12.28%; Table 3). This represents a potential threat for the patients, although there are minimal risks by these techniques themselves. In addition, these methods, with the exception of BIO, were applicable in a reduced percentage of patients only (60% to 84%). Furthermore, none of these techniques indicates right- or left-sided filling pressures, and thus its value for a complete hemodynamic assessment of critically ill patients is limited. The findings of this study are in the strict sense only valid for adult postcardiosurgical patients. However, in children, promising correlations have been reported of noninvasive versus invasive CO determination after correction of congenital heart

lesions.17’s3At present, despite its invasiveness, thermodilution remains the method of choice for comprehensive hemodynamic monitoring of adult patients after cardiac surgery. ACKNOWLEDGMENTS

MS Cardio-Medical (Brunnen, Switzerland) kindly provided the Accucom (Datascope) continuous-wave Doppler; Gambro AG (Htinenberg, Switzerland) provided the Cardioflo (Cardionics) pulsed-wave Doppler, and Hausmann Laboratorien (St Gallen, Switzerland) provided the respiratory gas analyzer (MMC Horizon, Sensor Medics). We would like to thank Dr Koenig, Department of Obstetrics, Research Unit, for assistance with regard to statistics; the nurses of the Cardiosurgical Intensive Care Unit; the graphic artist, Alena Chimburek; and Sylvia Distel and Dragan Popovic of the blood gas laboratory for their outstanding efforts in support of this study.

REFERENCES 1. Rao TLK, Jakobs KH, El-Etr AA: Reinfarction following anesthesia in patients with myocardial infarction. Anesthesiology 59:499-505.1983 2. Connors AF, McCaffree DR, Gray BA: Evaluation of right-heart catheterization in the critically ill patient without acute myocardial infarction. N Engl J Mad 308:263267.1983 3. Kohanna FH, Cunningham JN, Catinella FP, et al: Cardiac output determination after cardiac operation. Lack of correlation between direct measurements and indirect estimates. J Thorac Cardiovasc Surg 82:904-908,198l 4. Magilligan DJ, Teasdale R, Eisinminger R, et al: Mixed venous oxygen saturation as a predictor of cardiac output in the postoperative cardiac surgical patient. Ann Thorac Surg 44:260-262,1987 5. Schuster AH, Nanda NC: Doppler echocardiographic measurement of cardiac output: Comparison with a non-golden standard. Am J Cardiol53:257-259,1984 6. Hillis LD, Firth BG, Winiford MD: Analysis of factors affecting the variability of Fick versus indicator dilution measurement of cardiac output. Am J Cardiol 56~764-768, 1985 7. van Grondelle A, Ditchey RV, Groves BM, et al: Thermodilution method overestimates low cardiac output in humans. Am J Physiol245:H690-H692,1983 8. Jansen JRC, Versprille A: Improvement of cardiac output estimation by the thermodilution method during mechanical ventilation. Intensive Care Med 12:71-79, 1986 9. Jansen JRC, Schreuder JJ, Bogaard JM, et al: Tbermodilution technique for measurement of cardiac output during artificial ventilation. J App. Physiol50:584-591, 1981 10. Stetz CW, Miller RG, Kelly GE, et al: Reliability of the thermodilution method in the determination of cardiac

output in clinical practice. Am Rev Respir Dis 126:10011004,1982 11. Donovan KD, Dopp GJ, Woods WPD, et al: Comparison of transthoracic electrical impedance and thermodilution methods for measuring cardiac output. Crit Care Med 14:1038-1044,1986 12. Bernstein DP: Continuous noninvasive real-time monitoring of stroke volume and cardiac output by thoracic electrical bioimpedance. Crit Care Med l&898-901,1986 13. Appe PL, Kram HB, Mackabee J, et al: Comparison of measurement of cardiac output by bioimpedance and thermodilution in severely ill surgical patients. Crit Care Med 14:933-935, 1986 14. Goldstein DS, Cannon RO, Zimlichman R, et al: Clinical evaluation of impedance cardiography. Clin Physiol 6:235-251,1986 15. Edmunds AT, Godfrey S, Tooley M: Cardiac output measured by transthoracic impedance cardiography at rest, during exercise, and at various lung volumes. Clin Sci 63:107-113,1982 16. Rose JS, Nanna M, Ramintoola SH, et al: Accuracy of determination of changes in cardiac output by transcutaneous continuous-wave Doppler computer. Am J Cardiol54:1099-1101, 1984 17. Rein AJJT, Hsieh KS, Elixson M, et al: Cardiac output estimates in the pediatric intensive care unit using a continuous-wave Doppler computer. Validation and limitations of the technique. Am Heart J 112:97-103,1986 18. Vandenbogaerde JF, Scheldewaert RG, Rijckaert DL, et al: Comparison between ultrasonic and thermodilution cardiac output measurement in intensive care patients. Crit Care Mad 14:294-297.1986 19. Alverson DC, Eldridge M, Dillon T, et al: Nonin-

NONINVASIVE

v INVASIVE CO ASSESSMENT

vasive pulsed Doppler determination of cardiac output in neonates and children. J Pediatr 101:46-50, 1982 20. Lewis JF, Kuo LC, Nelson JG, et al: Pulsed Doppler echocardiographic determination of stroke volume and cardiac output: Clinical validation of two new methods using the apical window. Circulation 70~425-431, 1984 21. Huntsman LL, Steward DK, Barnes SR, et al: Noninvasive Doppler determination of cardiac output in man-Clinical validation. Circulation 67:593-602,1983 22. Ihlen H, Amlie JP, Dale J, et al: Determination of cardiac output by Doppler echocardiography. Br Heart J 51:54-60, 1984 23. Darsee JR, Walter PF, Nutter DO: Transcutaneous Doppler method of measuring cardiac output-II. Am J Cardiol46:613-618,198O 24. Loeppky JA, Hoekenga DE, Green ER, et al: Comparison of noninvasive pulsed Doppler and Fick measurements of stroke volume in cardiac patients. Am Heart J 107:339-346,1984 25. Seyde WC, Stephan H, Rieke H: Nicht-invasive Doppler-Ultraschallmessung des Herzzeitvolumens. Anaesthesist 36:504-509, 1987 26. Milsom I, Forssman L, Dottori 0, et al: Measurement of cardiac stroke volume during cesarian section: A comparison between impedance cardiography and the dye dilution technique. Acta Anaesthesiol Stand 27:421-426, 1983 27. Ihlen H, Myhre E, Pamlie JP, et al: Changes in left ventricular stroke volume measured by Doppler echocardiography. Br Heart J 54:378-383,1985 28. Siegel LC, Shafer SL, Martinez GM, et al: Simultaneous measurements of cardiac output by thermodilution, esophageal Doppler, and electrical impedance in anesthetized patients. J Cardiothorac Anesth 2:590-595, 1988 29. Nishimura RA, Callahan MJ, Schaff HV, et al: Noninvasive measurement of continuous-wave Doppler echocardiography: Initial experience and review of the literature. Mayo Clin Proc 59:484-489, 1984 30. Donovan KD, Dopp GJ, Newman MA, et al: Comparison of pulsed Doppler and thermodilution methods for measuring cardiac output in critically ill patients. Crit Care Med 15:853-857,1987 31. Mark JB, Steinbrook RA, Gugino LD, et al: Continuous noninvasive monitoring of cardiac output with esophageal Doppler ultrasound during cardiac surgery. Anesth Analg 65:1013-1020,1986 32. Sachs L: Angewandte Statistik (ed 6). SpringerVerlag, Berlin, West Germany, 1984, pp 205-209 33. Werner J: Medizinische Statistik. Urban and Schwarzenberg, Munich, West Germany, 1984, pp 141-143 34. Fromm RE: The craft of cardiopulmonary profile analysis, in Shnyder JV (ed): Oxygen Transport in the Critically Ill. Year Book Medical, Chicago, IL, 1987, pp 249-269 35. Levett JM, Replogle RL: Thermodilution cardiac output: A critical analysis and review of the literature. J Surg Res 27:392-404, 1979 36. Okamoto K, Komatsu T, Kumar V, et al: Effects of intermittent positive-pressure ventilation on cardiac output measurement by thermodilution. Crit Care Med 14:977-980, 1986

59

37. Nelson LD, Snyder JV: Technical problems in data acquisition, in Shnyder JV (ed): Oxygen Transport in the Critically Ill. Year Book Medical, Chicago, IL, 1987, pp 205-234 38. Bernstein DP: A new stroke volume equation for thoracic electrical bioimpedance: Theory and rationale. Crit Care Med 14:904-909, 1986 39. Meiners S: Messmethoden zur Analyse der Herzund Kreislaufdynamink, in Weber A, Blumberger KJ (eds): Kreislaufmessungen, 1. Freiburger Colloquiums iiber Kreislaufmessungen, Freiburg/Breisgau. Werk-Verlag, Mtlnchen, West Germany, 1958, pp 84-98 40. Davis GG, Jebson PJR, Glasgow BM, et al: Continuous Fick cardiac output compared to thermodilution cardiac output. Crit Care Med 14:881-885, 1986 41. Rubin SA, Siemienczuk D, Nathan MD, et al: Accuracy of cardiac output, oxygen uptake, and arteriovenous oxygen difference at rest, during exercise, and after vasodilator therapy in patients with severe, chronic heart failure. Am J Cardiol50:973-978, 1982 42. Kubicek WG, Karnegis JN, Patterson RP, et al: Development and evaluation of an impedance cardiac output system. Aerospace Med 1208-1212,1966 43. Hottinger CF, Meindl JD: Blood flow measurement using the attenuation-compensated volume flowmeter. Ultrason Imaging l:l-15, 1979 44. Francis GS, Hagan AD, Oury J, et al: Accuracy of echocardiography for assessing aortic root diameter. Br Heart J 37:376-378, 1975 45. Greenfield JC, Pate1 DJ: Relation between pressure and diameter in the ascending aorta of man. Circ Res 10:778-781,1962 46. Segadal L, Matre K: Blood velocity distribution in the human ascending aorta. Circulation 76:90-100, 1987 47. Lucas CL, Keagy BA, Hsiao HS, et al: The velocity profile in the canine ascending aorta and its effects on the accuracy of pulsed Doppler determinations of mean blood velocity. Circ Res 18:282-293,1984 48. Jenni R, Vieli A, Ruffmann K, et al: A comparison between single gate and multigate ultrasonic Doppler measurement for the assessment of the velocity pattern in the human ascending aorta. Eur Heart J 5:948-953, 1984 49. Gabriel S, Atterhog JH, Or6 L, et al: Measurement of cardiac output by impedance cardiography in patients with myocardial infarction-Comparative evaluation of impedance and dye dilution methods. Stand J Clin Lab Invest 36:29-34, 1976 50. Milsom I, Sivertsson R, Biber B, et al: Measurement of stroke volume with impedance cardiography. Clin Physiol2:409-417, 1982 5 1. Shah KB, Rao TLK, Laughlin S, et al: A review of pulmonary artery catheterization in 6245 patients. Anesthesiology 61:271-275, 1984 52. Philip JH, Long MC, Quinn MD, et al: Continuous thermal measurement of cardiac output IEEE Trans Biomed Eng 31:393-400,1984 53. Schmid ER, Jenni R, Tornic M: Nichtinvasive Herzminutenvolumen-Bestimmung in der frilhpostoperativen Phase nach Korrektur Kongenitaler Herzfehler. Anaesthesist 38:573, 1989