Multiple Cardiac Output Measurements in Man

Multiple Cardiac Output Measurements in Man* Evaluation of a New Closed-System Thermodilution Method Jesse J. Stawicki, M.S.B.E.; Fred D. Holford, M.D.; Eric L. Michelson, M.D.;"" and Mark E. Josephson,

M.D.f

Cardiac output estimates by a new closed-system automatic injection thermodilution method ( C O I - T D ) were compared serially with the direct Fick technique ( C O F I C K ) and the standard open-system manual injec-

tion thermodilution method ( C O T D ) . Comparison with cardiac outputs determined simultaneously by the direct Fick technique in 1 0 0 measurements involving ten patients showed close agreement with the new closed system method using both 25°C and 3 ° C injectates. The cardiac output range was between 1.9 to 11.6 L/min. The open-system manual injection thermodilution method under identical conditions produced a wide scatter of measurements when compared to the direct

A ggressive hemodynamic monitoring for the optimal treatment of critically ill patients with ischemic or valvular cardiac disease requires accurate and frequent assessment of myocardial performance. The present study demonstrates the ease and accuracy with which cardiac function can be quantitated at the bedside in intensive care units using a closed system automatic injection thermodilution cardiac ouput method ( C O I - T D ) . Previous clinical investigations of cardiac output determination by thermodilution in man have used an open-system approach where syringes were loaded with injectate manually and then injected directly by hand. The open-system manual injection thermodilution ( C O T D ) method was first demonstrated in 1954 by Fegler and has been studied 1

Fick

technique.

Reproducibility

of

simultaneous

C O I - T D and C O T D measurements was examined in quad-

ruplicate.

The reproducibility

of measurements was

within 1.9 percent with C O I - T D and 5.9 percent with the C O T D method using both 2 5 ° C and 3 ° C injectates. The C O I - T D method eliminates the technical problems

of recirculation, unstable indicator baseline changes, thermal continuity, and reproducibility encountered with the current C O T D method. Analysis of cardiac output by

the closed-system automatic injection thermodilution method provides a simple, rapid, reproducible, and highly accurate method for multiple cardiac output measurements suitable for use at the bedside.

extensively in laboratory animals by numerous investigators. Studies spanning the past 14 years have documented the relative reliability of the C O T D method in man. ' " Correlation of C O T D with other techniques for measurement of cardiac output, for example, dye dilution and the Fick technique, also has been done previously. Correlation of these methods over the clinical range of cardiac outputs has in general been good. Although the Fick technique is often the reference by which other cardiac output determinations are evaluated, the technical difficulties in obtaining reliable Fick outputs have precluded extensive controlled comparative observations in man. On the other hand, comparison between dye dilution and C O T D has demonstrated outputs by the two techniques may differ by about 7 percent. Dye dilution methods have not gained wide acceptance because of their complexity and inaccuracy at high and low flow states. The accuracy of thermodilution methods has similarly been questioned at high and low flow states. 212

11

13

16

1113

14

" F r o m the Cardiac Catheterization Laboratory and Medical Intensive Care Unit, Hospital of the University of Pennsylvania, Cardiovascular Section, Department of Medicine, University of Pennsylvania School of Medicine, Philadelphia. Supported in part b y grants from the American Heart Association, Southeastern Pennsylvania Chapter, and by the Commonwealth of Pennsylvania, Pennsylvania Department of Health, Division of Chronic Diseases Grant No. 493737. ""Recipient, Research Fellowship Award, Southeastern Pennsylvania Chapter, American Heart Association and National Research Service Award of the National Heart, Lung and Blood Institutes. fRecipient, Research Career Development Award National Heart, Lung and Blood Institute Grant No. 1 K 0 4 HL00361-01. Manuscript received November 1 3 ; revision accepted February 19. Reprint requests: Dr. Micliekon, Hospital of the University of Pennsylvania, Philadelphia 19104

CHEST, 76: 2, AUGUST, 1979

Despite its inherent simplicity, the C O T D method has not gained wide clinical acceptance because of several practical problems. First, the temperature of the injectate must be known, and for outputs to be reproducible, the temperature of each bolus of injectate should be the same. An error of 1°C introduced for either temperature of blood or injectate will alter the cardiac output determination by ap17

MULTIPLE CARDIAC OUTPUT MEASUREMENTS IN MAN 193

proximately 3 percent. Second, distinct fluctuations in pulmonary arterial temperature, related to respiratory cycling, result in an unstable thermal baseline. Third, although there is no distinct recirculation of the thermal tracer, there is a slow return of the thermistor temperature to baseline. The closed system automatic injection thermodilution method provides a solution for each of the three problems associated with the C O T D method. The COi-m method increases the accuracy, reliability, and reproducibility of the basic thermodilution technique. MATERIALS AND METHODS Serial measurements of cardiac output by the closed-system automatic injection thermodilution method, open-system manual injection thermodilution, and Fick technique were performed on ten patients in our cardiac catheterization laboratory and medical intensive care unit. The patient population consisted of 6 men and 4 women ranging from 3 6 to 6 4 years of age. All patients were undergoing evaluation for either coronary artery or valvular disease and were studied with informed consent. Catheterization of the right and left sides of the heart were performed with lidocaine ( X y l o c a i n e ) local anesthesia and the patients were sedated breathing room air. The first five patients were injected with 3 ° C 5 percent dextrose in water solution ( D 5 W ) and the remaining five with 2 5 ° C D 5 W injectate. A cutdown was performed in the right antecubital fossa in seven patients. For this group, a size 7 F B-D Electrodyne thermodilution balloon-tipped triple lumen catheter was passed via the brachial vein to the pulmonary artery. Also, a size 7 pigtail catheter was passed retrograde from the brachial artery to the ascending aorta and left ventricle. The remaining three patients had the thermodilution catheter placed percutaneously into the right femoral vein and passed to the pulmonary artery. For O , saturation measurements, a size 7 pigtail catheter was then introduced into the right femoral artery after systemic heparinization and then advanced into the left ventricular cavity. Measurement of cardiac output by thermodilution involves the placement of a triple lumen temperature-sensing catheter into the right atrium. This catheter is then flow-directed through the right ventricle to the first lobar branch of the pulmonary artery. A measured amount of injectate of known temperature is injected through the proximal lumen of the catheter and released into the right atrium. A thermistor located at the distal tip of the catheter in the pulmonary artery measures baseline temperature, detects the change in blood temperature, and transmits this signal to the computer. The decay in voltage is integrated with respect to time, from the temperature-time curve, and the cardiac output is electronically computed and displayed in liters per minute. Cardiac output according to the thermodilution principle was calculated automatically using the B-D Electrodyne C O 100 portable cardiac output computer. The computer was externally connected to a slow speed ( 1 m m / s e c } strip chart recorder which produced a thermal curve with each injection. To evaluate optimal injectate temperature, reproducibility, and catheter position the following procedure was carried out: An automatic thermodilution O M P Laboratories injector model 3 7 0 0 powered by a disposable OMP No. 3 8 9 0 C O cartridge was used with a three-way stopcock which connected to both the proximal lumen of the thermodilution a

194 STAWTCKI ET AL

catheter and the D 5 W injectate reservoir. An integral nonrelieving regulator within the injector reduces 9 0 0 psi primary pressure to 5 0 psi during injection, which conforms with hand injection techniques as suggested by N o r m a n n . In this way, no human contact was made with the injectate, thus reducing possible thermal error in our closed system injection. A single disposable carbon dioxide cartridge pressurized to 9 0 0 psi yields 20 cycles of continuous injection and aspiration. The 10 ml syringe was filled automatically in 8 sec by the injector by means of an open port to the reservoir. At this point, the stopcock was adjusted to close the intravenous line flow and open the syringe to the proximal lumen of the thermodilution catheter ready for injection. The volume of injectate delivered into the right atrium and the difference between body and injectate temperature (both measured via the thermistor) were manually dialed in on scaled potentiometers. The computer was then put into the compute mode and within three seconds, the injector holding the 10 ml of D 5 W was triggered. The injection time was exactly two seconds and provided a steady rate of flow. The cardiac output in liters per minute was displayed within 15 seconds following administration of the injectate. After the first readout, the injector was reloaded automatically and the computer was autobalanced within ten seconds. This allowed two cardiac output determinations per minute. 18

F o r the closed-system thermodilution, the D 5 W reservoir was kept refrigerated at 1°C .for one-half hour. Before injection, the D 5 W reservoir bottle was mounted in a specially designed receptacle filled with ice. The intravenous line on the injector syringe was then connected from the three-way stopcock to the D 5 W reservoir inside the receptacle under continuous temperature monitoring. Injection followed in less than 2 0 seconds and quadruple determinations were complete within 2 . 5 minutes. The B-D Electrodyne C O - 1 0 0 computer evaluates the thermal curve area by the following process. During injection, the voltage corresponding to the cooling of the blood rises to a peak voltage. This peak is established by the computer as the lowest temperature sensed. From this point the curve decays exponentially with time until an automatic timing circuit is activated, at 7 5 percent of the peak signal, and is again activated at 30 percent of the peak signal to measure the area under the curve down to 12 percent of peak. Once die timing circuit has been set, no fluctuations in the thermal curve can affect the integrated cutoff point to give an erroneous display. The computer compensates for the area under the curve below the 12 percent cutoff due to artifact by adding a weighting factor to the preceding portion of the curve. The result of the automatic timing circuit computation is improved accuracy and repeatability of the cardiac output display. Measurements of cardiac output by the open-system manual injection thermodilution were done with both iced and room temperature injectates. The syringes were handled carefully and rapidly during manual injection in an effort to eliminate rate of flow and thermal continuity errors. Manual injections were carried out under the same time considerations as in the closed-system injection thermodilution. Fick

Technique

Cardiac output was computed by the classic Fick equation using the Beckman Instruments metabolic measurement system. In order to assure steady state conditions, especially with regard to mean pulmonary volume, the supine and premedicated patients breathing room air were thoroughly familiarized with our measurement apparatus and corrections were made for the respiratory quotient, water vapor in the

CHEST, 76: 2, AUGUST, 1979

air, and air temperature. Cardiac output by the Fick method was measured either simultaneously with the two thermodilution methods or just before they were done. During gas sample collection, the mixed venous blood was withdrawn from the pulmonary artery at the same rate and time as arterial blood was removed from the left ventricle. Oxygen content of the pulmonary and systemic arterial blood samples was determined on the Instrumentation Laboratory I L - 1 8 2 CO-oximeter. T h e expired air collected for at least nine minutes was analyzed for oxygen by a Beckman O M - 1 1 oxygen analyzer with an accuracy of ± 0 . 2 percent 0 „ a n d carbon dioxide was analyzed by the Beckman L B - 2 medical gas analyzer with an accuracy of ± 1 percent CO.,. T h e mixed expired gas concentrations were sampled from an expired gas mixing chamber. Alveolar gas concentrations were sampled from a mouthpiece with the peak (maximum CC\ and minimum 0 . , concentration) held electronically and printed out every 3 0 seconds. T h e gas volume measurements were obtained by internal mechanical transfer to a Collins 120 liter gasometer. RESULTS

0

I

1

0

1

, 2

, 3

, 5

4

, 6

, 7

, B

i 1

1

1

i 2

11

The comparison between C O I - T D and C O T D against C O F I C K is reported ( F i g l a and l b ) respectively for the total combined 2 5 ° C and 3 ° C injectates in 1 0 0 simultaneous measurements. Outputs ranging from 1 . 9 to 1 1 . 6 liters/min were analyzed to produce the linear regression equations. The linear regression equations for the combined injectate temperatures were: COI-TD = COTD

=

.183, r =

-978 C O F I C K + .852 COFICK

+

.531, r

=

.982,

(P

<

.001)

.638,

(P

<

.001)

Linear regression equations for C O I - T D and C O T D against C O F I C K for 2 5 ° C and 3 ° C injectates individually are shown in Table 1 . The mean difference between C O I - T D and C O F I C K was + . 0 6 ± . 5 1 ( S D ) liters/min. No systematic difference was demonstrated between C O I - T D and C O F I C K ( P > . 1 0 ) . F o r C O T D and C O F I C K the mean difference was — . 3 2 ± .67(SD)litcrs/min(P>10). The reproducibility of C O I - T D and C O T D methods was evaluated by examining the percentage of difference between simultaneous quadruple measurements. The percentage of difference from the mean of each set of quadruple determinations was analyzed by subtracting the value of each individual output from the mean of the four outputs. The sum of these differences in percentage from the mean was then divided by the number of determinations for each subset of outputs to derive the mean difference in percentage. The mean percentage difference of 5 0 measurements of C O I - T D using both 2 5 ° C and 3 ° C injectates was 1 . 9 percent ± 1.5 percent ( S D ) , whereas 5 0 measurements of C O T D had a mean difference of 5 . 9 percent ± 5 . 3 percent ( S D ) . Further analysis of reproducibility data using only 2 5 ° C injectates showed that the mean difference of 2 5 measurements of C O I - T D was 2 . 1 percent ± 1 . 6 percent ( S D ) and C O T D was 8 . 2 percent ± 7 . 5 perCHEST, 76: 2, AUGUST, 1979

cardiac

Output

fick

( liters/min

F I G U R E l a ( u p p e r ) and l b (lower).

i

Comparison of results

of C O I - T D , C O T D , and C O F I C K for the entire patient popula-

tion using values from both 2 5 ° C and 3 ° C injectates. E a c h one of the ten symbols plotted represents the mean of five individual measurements. C O I - T D = cardiac output by closed system injection thermodilution; C O T D = cardiac output by standard thermodilution; C O F I C K = cardiac output by the Fick technique.

cent ( S D ) . At 3 ° C injectate temperature the mean difference of 2 5 measurements of C O I - T D was Table COTD

1—Linear Regression Equations for C O l - T D and Compared Against C O F I C K Method for Each Given Injectate Temperature Range

Injectate Temperature, Range 25°C + 3 C 0

Linear Regression Equations COI-TD = . 9 7 8 COFIOK + . 1 8 3 , r = 9 8 2 C()

25°C

CO 3°C

= . 8 5 2 CO ICK-|-.531, r = 683

T D

F

CO,-TD = .953 C O M C K - . 0 5 7 , T

D

r - 972

= . 6 1 6 CO ICK + 1.83, r = 354 F

COI-TD = 1.00 C O F I C K + . 0 8 5 , r = 9 9 6 COTD

= . 9 6 5 C O F I C K - - 0 6 2 , r = 974

MULTIPLE CARDIAC OUTPUT MEASUREMENTS IN MAN 195

T a b l e 2—Reproducibility of Cardiac Output Determinations Calculated from 50 Measurements Each Given Injectate Temperature Range

for

Injectate Temperature Range

M e a n P e n •pnt Difference C()l-TD

25°C+3°C 25°C 3°C

COTD (SD)

5.9%+5.3%

(SD)

2 . 1 % ± 1 . 6 % (SD)

8.2%±7.5%

(SD)

1.7%±1.4%

3 . 6 % + 3 . 1 % (SD)

1.9%±1.5%

(SD)

1 . 7 percent ± 1 . 4 percent ( S D ) and C O T D was 3 . 6 percent ± 3 . 1 percent ( S D ) (Table 2 ) . To test whether the site of insertion of the thermodilution catheter in the brachial or femoral vein affected the estimation of C O I - T D and C O T D , 7 7 measurements were performed on seven patients via the brachial vein and 2 3 measurements on three patients by femoral placement. With brachial vein placement the mean difference between C O I - T D and C O F I C K was + . 0 1 ± . 6 4 ( S D ) liters/min, whereas with femoral placement the mean difference was + .17 ± . 7 2 ( S D ) liters/min. Using C O T D in the brachial vein gave a mean difference of — . 1 2 ± . 7 1 ( S D ) L/ min and femoral placement resulted in a — .77 ± . 3 9 ( S D ) liters/min mean difference. None of these differences was statistically significant.

DISCUSSION

The present study has demonstrated the methodologic advantage of the C O I . T D method compared to the standard open-system manual thermodilution method ( C O T D ) . Very accurate control of the volume of injectate and temperature is required for an injection of a perfectly defined and reproducible bolus of injectate. " There are several factors that influence the transfer of heat between the injectate and blood. The heat content of the bolus injectate is modified during injection by an amount which depends upon its intra- and extracorporeal length, rate of injection, and temperature difference between injectate and blood. The C O I - T D method provides a reproducible and smooth rate of injection using an automatic injector with a driving force of 9 kg during the 2 sec injection time. With this type of control of volume and constant flow dynamics of the injectate, we are able to make accurate serial approximations for heat transfer in our closed system. The C O I - T D method involves both the direct temperature measurement of the injectate at the injection site and a correction factor for injectate temperature rise in the catheter. The interna] computer correction factor used for 1 0 18

20

ml of injectate at 3 ° C and 2 5 ° C

196

STAWfCKI ET AL

is . 8 3 and . 9 1

respectively. Operationally, the C O T D method, as it is usually executed, involves several poorly controlled variables. There is always the possibility of injectate contamination and considerable heat transfer with an open-system technique. Heat transfer in the syringe and extracorporeal length of the catheter constitute the most serious errors of the open-system manual thermodilution method. To minimize potential rewarming errors in both the accuracy and precision when an iced injectate is employed, the injectate should be used promptly and the syringe not handled. The C O I - T D method avoids any external risks of uncontrolled heat loss, as the temperature of the thermal indicator is measured within the iced bath receptacle holding the injectate reservoir and injection takes place automatically from the reservoir. 21

Using 2 5 ° C injectates can further eliminate extracorporeal heat losses, but the sensitivity of the C O I . T D method decreases by 1 . 5 times necessitating a corresponding increase in the volume of the injectate. Increasing the volume of injectate would increase the catheter's dead-space effect. The dead-space phenomenon occurs at the end of the injection procedure. Thermal exchange takes place between the 3 ° C injectate remaining in the lumen and the blood across the wall of the catheter with the result of increased cooling by conduction to the detected blood. The lag time before equilibrium between blood and catheter after injection will delay the return to baseline of the downslope of the thermal curve, resulting in a non-exponential lower downslope. It has been shown on a model that this delayed return to the baseline is primarily related to the dead-space effect. The dead-space effect is eliminated by the C O I - T D method since injection with rapid and smooth flow dynamics decreases erratic indicator flow patterns along with the time necessary for internal heat transfer. Also the B-D Electrodyne computer takes the skew of the descending portion of the thermal curve into account and obviates the potential problem in question. 22

23

One of the main sources of variability of serial cardiac output determination with the C O T D method derives from the fact that during inspiration cooler blood is drawn into the right heart with resultant spontaneous fluctuations in temperature of pulmonary arterial blood. The thermal curve responds to this temperature change with an unstable nonexponential configuration. This artifact can be lowered partially by increasing the volume of injectate and decreasing its temperature. Hyperpneic and hypovolemic patients can alter the temperature of the pulmonary blood significantly so as to produce a fluctuating thermal baseline approaching the magni24

CHEST, 76: 2, AUGUST, 1979

tude of the temperature change that follows an iced injection. An automatic timing circuit is used by the computer to complete the COI-TD system. With an automatic timer as an internal part of the computer, integration is performed on both baseline variation averaging and analysis of the part of the downslope of the curves from 7 5 to 1 2 percent of the peak signal. Once the timing circuit is set, continuous logarithmic extrapolation occurs, so that no fluctuations in the curve can affect the integration cutoff point of 1 2 percent peak to give an erroneous display. 23

Since the shape of the thermal curve depends upon blood volume and velocity of blood flow, finding a method for the selection of the exponential portion which would prove to be applicable to the entire range of possible thermal curves has been a serious problem for the COTD method. This is mainly because previous thermodilution computers have been limited to evaluation of 30 to 80 percent of the peak thermal signal. The COI-TD method using the automatic timing circuit integrates down to 1 2 percent of the peak signal, making it more accurate in determining low cardiac outputs, because of the increased detection range. Reliability of thermodilution techniques for measurement of cardiac output has been questioned for many years and has limited the routine application of this method for patient care. With the resolution of methodologic requirements correctly fulfilled by the COI-TD method, the overall simplicity, reliability, and accuracy of the method makes it an extremely useful tool for critically ill patients being monitored in special care units where multiple sequential trends of cardiac output are used to aid in decisions regarding therapy. Moreover, the closed-system automatic injection thermodilution method has been demonstrated to be accurate over a wide range of cardiac outputs, including both very low and very high output states. In addition, the high accuracy of this system is maintained even using injectate solution kept at room temperature (eg 2 5 ° C ) , further increasing the ease of use of this method. A C K N O W L E D G M E N T S : T h e authors wish to thank John A. Kastor, M.D., Chief, Cardiovascular Section, Hospital of the University of Pennsylvania, for his continued technical and editorial support, Charlotte Drzewiecki for her secretarial assistance, and John Helgesen, C.V.T., for his technical assistance. REFERENCES 1 Fegler G: Measurement of cardiac output in anesthetized animals by a thermo-dilution method. Quart J Exptl Physiol 3 9 : 1 5 3 - 1 6 4 , 1954 2 Benchimol A, Akre PR, Dimond E G : Clinical experience with the use of computers for calculation of cardiac

CHEST, 76: 2, AUGUST, 1979

output. Am J Cardiol 1 5 : 2 1 3 - 2 1 9 , 1 9 6 5 3 Pavek K, Boska D, Selecky F V : Measurement of cardiac output by thermodilution with constant rate injection of indicator. Circ Res 1 5 : 3 1 1 - 3 1 9 , 1964 4 Evonuk E, Imig C J , Greenfield W , et al: Cardiac output measured by thermal dilution of room temperature injectate. J Appl Physiol 1 6 : 2 7 1 - 2 7 5 , 1 9 6 1 5 Silove E D , Cantez T, Wells B G : Thermodilution measurement of left and right ventricular outputs. Cardiovasc Res 5 : 1 7 4 - 1 7 7 , 1971 6 Fegler G: T h e reliability of the thermodilution method for determination of the cardiac output and the blood flow in central veins. Quart J Exptl Physiol 4 2 : 2 5 4 - 2 6 6 , 1957 7 Fronek A, Ganz V : Measurement of flow in single blood vessels including output by local thermodilution. Circ Res 8 : 1 7 5 - 1 8 2 , 1960 8 Solomon HA, San Marco MA, Ellis RJ, et al: Cardiac output determination: Superiority of thermal dilution. Surg Forum 2 0 : 2 8 - 3 0 , 1 9 6 9 9 Goodyer AVN, Huvos A, Eckhardt W F , et al: Thermal dilution curves in intact animals. Circ Res 7 : 4 3 2 - 4 4 1 , 1959 10 Khalil H H , Richardson TQ, Guyton A C : Measurement of cardiac output by thermal-dilution and direct Fick methods in dogs. J Appl Physiol 2 1 : 1 1 3 1 - 1 1 3 5 , 1966 11 Stenson R, Grouse L, Harrison D: Computer measurement of cardiac output by dye dilution: Comparison of computer, Fick, and Dow techniques. Cardiovasc Res 6:449-456, 1972 12 Wessel H U , James GW, Paul M H : Effects of respiration and circulation on central blood temperature of the dog. Am J Physiol 2 1 1 : 1 4 0 3 - 1 4 1 2 , 1966 13 Branthwaite MA, Bradley R D : Measurement of cardiac output by thermal dilution in man. J Appl Physiol 2 4 : 4 3 4 438, 1 9 6 8 14 Canz W , Donoso R, Marcus H, et al: A new technique for measurement of cardiac output by thermodilution in man. Am J Cardiol 2 7 : 3 9 2 - 3 9 6 , 1 9 7 1 15 Loughman J : Cardiac output measurement by thermal dilution in anesthesia and intensive care. Anesth Intensive Care 1 : 3 9 3 - 3 9 9 , 1 9 7 3 1 6 Weisel R D , Vito L , Dennis R C , et al: Clinical applications of thermodilution cardiac output determinations. Am J Surg 1 2 9 : 4 4 9 - 4 5 4 , 1 9 7 5 17 Weisel RD, Bergcr RL, Hechtman H B : Current concepts measurement of cardiac output by thermodilution. N Engl J Med 2 9 2 : 6 8 2 - 6 8 4 , 1975 18 Normann NA: Thermodilution technique for cardiac output. N Engl J Med 2 9 5 : 4 8 , 1 9 7 6 19 Dizon C T , Gezari W A , Barash PG, et al: Handheld thermodilution cardiac output injection. Crit Care Med 5 : 2 1 0 - 2 1 2 , 1977 2 0 Reiniger E J : Error in thermodilution cardiac output measurement caused by variation in syringe volume. Cathet Cardiovasc Diag 2 : 4 1 5 - 4 1 7 , 1 9 7 6 2 1 Powner D : Thermodilution technique for cardiac output. N Engl J Med 2 9 3 : 1 2 1 0 , 1 9 7 5 22 Saadjian A, Quercy J E , Torresani J : Cardiac output measurement by thermodilution: methodologic problems. Med Prog Technol 3 : 1 6 1 - 1 6 7 , 1 9 7 6 23 Saadjian A, Quercy J F , Lacroix Y, et al: Problems in cardiac output measurement by thermodilution. In Conference on Theory and Practice of Blood Flow Measurement. London, Sector Publishing, Ltd., 2 9 : 1 5 5 - 1 5 8 , 1972 24 Sorensen M B , Bille-Rake N E , Engell H C : Cardiac output by thermodilution. Ann Surg 1 8 3 : 6 7 - 7 1 , 1 9 7 6

MULTIPLE CARDIAC OUTPUT MEASUREMENTS IN MAN

197