After introduction of a variety of new less-invasive methods for the measurement of cardiac output as alternatives to right heart catheterization, the widespread use of pulmonary artery catheters (PAC) in critically ill patients has been debated repeatedly in recent times [1,2]. Advantages and limitations of the various methods for cardiac output measurement have been the subject of immense discussion. Several very heterogeneously designed evaluation studies have been published [3-6], but due to different designs they are not comparable. There is no study that compares these new methods of cardiac output assessment simultaneously and equally weighed against a real reference method. The evaluation of PAC and the alternative methods against a reference seems to be the best way to achieve comparable results.
Besides accuracy an important issue of the clinical practicability of a cardiac output measurement technique is the detection of acute haemodynamic changes and critical circulatory states.
This study was designed to compare cardiac output measurement by the invasive pulmonary-artery-catheter-based thermodilution technique, a less-invasive transpulmonary thermodilution technique (PiCCO™; Pulsion Medical Systems, Munich, Germany), both based on indicator dilution technique, using the Henriques-Stewart-Hamilton calculation. Furthermore the non-invasive trans-oesophageal echo-Doppler method (HemoSonic™; Arrow International, Reading, USA); and the partial carbon dioxide (CO2) rebreathing method based on Fick equation (NICO™; Novametrix Medical Systems, Wallingford, USA) were compared to a direct measurement of the aortic blood flow by an ultrasonic peri-vascular transit-time flow-probe (A-Series, Flow Meter T 208; Transonic® Systems Inc., Ithaca, USA), placed around the ascending aorta in a porcine model. It should be elaborated whether PiCCO™, NICO™ and Hemosonic™ show comparable accuracy and precision to the reference measure with PAC thermodilution. Possible differences under high- or low-flow conditions were to be recognized. The correlations to the beat-to-beat reference method as a measure for the reflection of acute haemodynamic changes should be compared. Cardiac output measurements were performed over a wide cardiac output range, obtained by numerous haemodynamic interventions. The comparative measurements were carried out in a clinically realistic model very similar to an operating room situation.
After obtaining approval from the Local Ethics Committee on Animal Research, nine domestic pigs with an average weight of 30 kg were instrumented under general anaesthesia. After intramuscular premedication using 0.2 mg kg−1 bodyweight flunitrazepam (Rohypnol®; Hoffmann-La Roche, Grenzach-Wyhlen, Germany) and 15 mg kg−1 ketamine (Ketanest®; Parke-Davis, Freiburg, Germany), total intravenous anaesthesia was induced and maintained using ketamine 1.6-3.3 mg kg−1, 7-10 mg kg−1 h−1, flunitrazepam 0.07-0.1 mg kg−1 h−1, fentanyl (Fentanyl-Janssen®; Janssen-Cilag, Neuss, Germany) 3 μg kg−1, 2 μg kg−1 h−1 and rocuronium (Esmeron; Organon-Teknika, Eppelheim, Germany) 1 mg kg−1, 2 mg kg−1 h−1. The trachea was intubated and the lungs were ventilated (volume controlled, tidal volume 10 mL kg−1, f: 16) by a Servo ventilator (Siemens®, Germany). Respiratory minute volume was adjusted to receive normoxia and normocapnia. For maintenance of normovolemia all animals received full electrolyte solution 12-15 mL kg−1 h−1 (Jonosteril®; Fresenius, Bad Homburg, Germany) to maintain mean arterial pressure (MAP), central venous pressure (CVP), haemoglobin and haematocrit values at a stable state before and during instrumentation.
The initial instrumentations performed included a PAC (7 F, Arrow™ International) via introducer system (right jugular vein), a femoral arterial catheter (4 F, PiCCO™) and a suprapubic urinary catheter placement in all the animals. Typical pressure tracings documented the correct positions of the intravascular catheters. This was followed by placement of an ultrasonic peri-vascular transit-time flow-probe of appropriate size (Flow Meter T 208; Transonic® Systems Inc., Ithaca, USA) around the ascending aorta facilitated by a right thoracotomy in each animal. Additionally, a Fogarty occlusion catheter (Edwards Lifescience®, Irvine, USA) was positioned in the inferior vena cava via a small incision in the right atrium. The ultrasound probe with combined M-mode and Doppler transducers (HemoSonic™) was then placed in the oesophagus and positioned at the optimal proximity to the aortic walls. The partial CO2-rebreathing monitor (NICO™) was then attached to the ventilatory circuit between endotracheal tube and Y-piece. A Datex® AS/3 (Datex-Ohmeda®, Helsinki, Finland) monitor was used to calculate pulmonary arterial (PA) cardiac output and to apply basic anaesthetic monitoring. The trans-oesophageal Doppler flow-meter and the aortic flow-meter provided continuous beat-to-beat data and the CO2-rebreathing monitor provided cardiac output values once every 3 min. Thermodilution cardiac output was measured using 10 mL bolus injections of normal saline at a temperature of approximately 5°C and coinciding with every cardiac output measurement of the CO2-rebreathing technique (every 3 min). Average values were calculated out of three pulmonary arterial thermodilution cardiac output measurements. Besides, all the haemodynamic (MAP, CVP, PAP, HR) and ventilatory data of the anaesthetic monitor, and the cardiac output data were recorded both manually and using a computerized multi-interface data recording and processing system that is usually used for research, concerning closed loop control of different parameters of general anaesthesia .
Arterial blood samples were analysed using an ABL 3 autoanalyzer (Radiometer®; Copenhagen, Denmark) at the beginning of the experimental protocol and approximately every 30 min to maintain constant serum electrolytes and normocapnia. Inspiratory oxygen concentration (FiO2), arterial partial pressures (PaO2, PaCO2) and haemoglobin concentration were used to calibrate the partial CO2-rebreathing monitor (NICO™) at regular intervals. The core temperature was kept over 35°C by active warming of the animals and infused fluids.
At the end of surgical preparation, at least 60 min were allowed for stabilization before at least 10 baseline readings were obtained. Different haemodynamic situations were induced in every animal by several analogously performed interventions. Modulation of cardiac output was then elicited by the infusion of dopamine (10/20/40 μg kg−1 min−1) and epinephrine (0.3 μg kg−1 min−1). Low cardiac output levels were first accomplished by blocking the inferior caval vein, using the intravascular Fogarty catheter, until the CVP decreased to half of the baseline level. Subsequently, the infusion of 40 mL kg−1 hydroxyethylstarch 130/6% (HAES-steril®; Fresenius, Bad Homburg, Germany) and additional application of epinephrine (0.3 μg kg−1 min−1) led to maximum cardiac output levels. Finally, a hypovolaemic shock was induced by controlled bleeding. Every haemodynamic intervention was followed by a stabilization period to equilibrate the haemodynamic situation, before a consecutive intervention. All the paired cardiac output data were recorded simultaneously in accordance with the NICO™ monitor as automated slowest-reacting measurement technique. Measurements were only carried out when the haemodynamic condition remained at a stable state, judged by intravascular pressures and transit-time aortic blood flow. An exemplary trend plot of the competing cardiac output measurements of animal 3 during the course of interventions is given in Figure 1.
The aortic blood flow measured in the ascending aorta was taken as reference method, assuming that it represents the best estimation of a true value of the cardiac output.
A plot of the differences between the reference method and every tested technique was done according to the method described by Bland and Altman . The plot allows visual assessment of agreement or disagreement between two clinical methods. The mean of the methods is treated as the best estimation of the true values. Bias, the mean of the differences of the two methods, 95% confidence intervals (CI) and the limits of agreement, i.e. the double standard deviation (2SD) of the differences are given for every comparison. Bias and limits of agreement were calculated for the entire dataset and separately for a reference cardiac output below and above 4 L min−1 to reveal differences between the haemodynamic states. Correlation between techniques was determined using linear regression analyses. Beside the fitted regression line an identity line, representing x = y is given in the regression plots.
Besides a trend score was computed for the paired measurements during hemodynamic interventions. The trend score is derived from the changes in consecutive cardiac output values. If both methods simultaneously indicate a positive trend, the changes compare positively and a positive score is counted. If both show a negative trend, they again compare positively. When the changes in cardiac output are in opposite directions they compare negatively and a negative score is counted. Ideally, only positive scores are present.
A total of 366 paired measurements were carried out in nine animals in accordance with the different interventions described above. The median weight of the animals was 30 kg (26-32). A summary of the haemodynamic parameters is given in Table 1. Besides the confluent data, the haemodynamic data under hypo- and hyper-dynamic conditions are given separately. Mean cardiac output values, SD and ranges are given analogously.
The correlation coefficients of the four different methods against the reference technique are presented in Table 2. Bias, limits of agreement and 95% CI are given for each tested method for the confluent data and hypo- and hyper-dynamic state as well.
The Bland-Altman plots (Fig. 2) show for PA thermodilution and transpulmonary thermodilution that as the average cardiac output increases, the difference between the methods increases in relation to the reference measure as well. The Bland-Altman plots and the regression analysis demonstrate that the values of thermodilution measurements were higher than the reference cardiac output values.
The Bland-Altman plots for HemoSonic™ and NICO™ reveal that the mean difference to the reference method is close to zero for both, in contrast to the thermodilution techniques. The regression plots demonstrate that at a range of transit-time-flow of 7 L min−1 the corresponding maximum values for both thermodilution techniques are 10.1 and 10.6 L min−1, respectively, whereas trans-oesophageal Doppler and partial CO2-rebreathing technique did not show such an overestimation of the reference cardiac output.
The trend score was calculated for 290 paired cardiac output measurements in relation the continuous reference measurement. 76 readings during the stabilizing periods between the interventions were not considered for trend analysis. Of 290 trends in cardiac output, 244 (84%) and 251 (86%) were scored in the same direction by pulmonary and trans-pulmonary thermodilution. For trans-oesophageal Doppler flow-meter and partial CO2-rebreathing 191 (65%) and 169 (58%) were scored in the same trend.
Since the 1970s, the pulmonary arterial thermodilution technique has been considered the clinical standard of advanced haemodynamic monitoring in critically ill patients . However, the PAC technique is an indirect procedure that has to be judged critically regarding accuracy and clinical impact. With the introduction of alternative, less-invasive methods of cardiac output monitoring, people increasingly started using the technical innovations in daily practice. Differences in application and reliability of the tested methods seem to be obvious, considering the absolutely different technical approaches to cardiac output assessment. The design of the current study allows, specifying differences, in performance and practicability of the contributing methods. In contrast to other studies, our investigation allowed competitive measurements by four different methods over a wide cardiac output range and the employment of a real reference method.
The most important findings of this study are: cardiac output measurement by pulmonary and trans-pulmonary thermodilution may be used interchangeably. Both methods overestimated cardiac output in general. PiCCO™ was able to measure cardiac output with comparable bias and precision weighed against PAC. Acute haemodynamic changes were reflected instantly and nearly identically. HemoSonic™ can assess aortic blood flow continuously on a non-invasive basis and showed an overestimation of aortic flow in this study. Acute changes are reflected reliably and absolute measurements should not be weighed over trend monitoring. Partial CO2-rebreathing is able to assess cardiac output non-invasively. Due to the overestimation of low cardiac output measures and underestimation of high cardiac measures, the application of the partial CO2-rebreathing technique seems to be limited during acute haemodynamic changes. The variety of methods available enables the clinician to apply patient- and situation-adapted cardiac output monitoring.
The employment of a highly invasive peri-aortic ultrasound flow-probe is limited to thoracic and experimental surgery. Transit-time ultrasound is the leading technique of continuous cardiac output measurement from both technical and physiological standpoints [10,11]. The technique qualifies to be a standard method by pre-calibration, direct continuous flow measurement, insensitivity to rotation, haematocrit or any electrically charged molecules. The porcine model used by us is well-established for cardio-circulatory investigations. The advantages of an animal experimental approach are obvious and were mentioned above. However, the results should not be transformed to human physiology directly. All tested methods and their algorithms were developed based on human circulatory physiology. Besides, cardiac output measures of 7 L min−1 mark a physiological limit in pigs since their bodyweight of 30 kg, and have to be judged critically. Furthermore, differences in the fractions of regional perfusion, compared to human haemodynamics may cause discrepant absolute cardiac output measures especially by the indicator dilution techniques. Although we cannot exclude a constant bias and an overestimation of cardiac output in our experimental model, there are equivalent conditions for common pulmonary arterial thermodilution and the new methods as well.
The mean global bias measures between PAC and PiCCO™ thermodilution to the reference method in our model, were −0.68 and −1.22 L, respectively. A difference in bias of approximately 0.5 L min−1 seems to be clinically acceptable. The difference between the methods increased for both thermodilution techniques with increasing average cardiac output, especially at cardiac output levels above approximately 4 L min−1. Both methods showed better bias and precision in low cardiac output states. Analogously, bias and precision values distinctly worsened at levels over 4 L min−1. Both thermodilution methods overestimated the cardiac output, but revealed excellent correlation to the reference cardiac output. This may be caused by a number of factors. Coronary perfusion is not included by the transit-time flow-meter, which, however, cannot explain all the differences. Besides, both of the thermodilution computers use algorithms concerning human haemodynamics, which might explain the systematic overestimation, but excellent correlation. Thermodilution measurements are influenced by injection technique, temperature, volume, and localization of injectate and thermo-recording. Overestimation of cardiac output by thermodilution at high cardiac output states have been described previously in clinical and experimental studies . A possible reason for the increasing difference between the methods at hyper-dynamic states could be a relative lack of thermo-indicator, compared to low- and medium-flow states .
Regarding the excellent correlation of PAC thermodilution (r = 0.93) and transpulmonary thermodilution (r = 0.95) with the reference method and with each other (r = 0.98), both methods seem to be interchangeable in many clinical situations, especially when possible complications and disadvantages of the use of PACs are considered. While PA catheterization allows the verification of left ventricular filling and performance by the assessment of left ventricular end-diastolic pressure, the concept of transpulmonary thermodilution suggests a volume-focused assessment of cardiac function. In addition to a reliable assessment of cardiac output, this device is also useful in the measurement of several additional parameters like the preload independent cardiac function index, global end-diastolic volume, intra-thoracic blood volume and extravascular lung water. Furthermore, PiCCO™ offers a thermodilution calibrated beat-to-beat cardiac output monitoring by analysis of the arterial pulse curve , where, in our opinion, trends are more important than absolute numbers.
The trans-oesophageal Doppler method of measuring cardiac output (HemoSonic™) has generated a wide interest due to its non-invasive application and because it offers fast-response continuous assessment of haemodynamic changes. Suggested indications of this monitor are surgical patients with a cardiac history undergoing non-cardiac procedures. Non-invasive Doppler monitoring may also be useful for advanced haemodynamic monitoring in infants where invasive monitoring may be technically difficult . The bias between the descending aortic blood flow of the HemoSonic™ and reference is expected to represent cerebral and upper extremity blood flow. Thus our data reveal an overestimation of aortic blood flow by the HemoSonic™. However, the non-invasive trans-oesophageal Doppler monitor measurements showed a good correlation (r = 0.84) with the reference method in our study. The most sensitive and influencing factor in the accuracy of HemoSonic™ 100 is, of course, the aortic diameter, easily explained by the law of Hagen-Poiseuille. We found that, despite the image of proximal and distal parts of the aorta, even experienced users sometimes encountered difficulties in placing the oesophageal probe instantly in a proper position. However, we were finally able to achieve a proper signal quality in all the animals.
The quality of the echo-Doppler signal is influenced by surgical manipulations that affect the anatomic relation of oesophagus and aorta. Thus, there may be limitations to this equipment during cardiac surgery and procedures involving the upper abdominal region. Nevertheless, HemoSonic™ offers a useful, fast-reacting tool of advanced haemodynamic monitoring with a good correlation to our reference. The acute haemodynamic changes caused by the interventions described above were rapidly and reliably reflected, provided there was a good signal quality.
The NICO™ cardiac output monitor is appealing in view of its absolutely non-invasive character, too. As a method of partial rebreathing, the use of this monitor is limited to patients whose lungs are mechanically ventilated and who are able to tolerate an increase in PaCO2. Minute ventilation has to be kept stable and so controlled-ventilation modes are mandatory. The global mean bias between NICO™ and reference was −190 mL min−1. The bias values for low and high output levels seem to be in clinically acceptable range of 500 mL. However our data revealed an overestimation of low cardiac output measures and an underestimation of high cardiac measures. This bidirectional error, that was described by other authors as well , may cause a measurement compression during acute haemodynamic changes. Differences in pulmonary perfusion and the amount of shunt effects during low flow and hyper-dynamic circulatory states, compared to human hemodynamics may be relevant, especially since only the cardiac output fraction participating in pulmonary gas exchange is considered by the method. On the other hand side these effects are hardly to assess and usually also occur during the clinical use of the device in critically ill patients.
The correlation of the cardiac output measured by this monitor with our reference method was moderate (r = 0.77). Even marked changes in cardiac output were detected only partially and relatively slowly by the technique. We feel that partial CO2-rebreathing may have only a limited application as an interchangeable alternative to invasive cardiac output measurements at the moment.
Measuring cardiac output is expected to be of significant clinical relevance in the monitoring and treatment of anaesthetic and critically ill patients. The assumption of improved patients outcome and decreased perioperative risk by advanced haemodynamic monitoring seems to be likely, but is overshadowed especially by the risks of the invasive measurement techniques themselves. The new less-invasive methods of cardiac output measurement, in spite of their specific limitations and disadvantages, may represent alternatives and complements to the common monitoring tools. Regarding the measurement of an acute changing cardiac output, transpulmonary and pulmonary thermodilution seem to be interchangeable in all haemodynamic states. However, both methods have to be judged critically during hyper-dynamic circulation, where overestimations of cardiac output may occur. Trans-esophageal echo-Doppler is attractive through rapid availability, non-invasive application and fast response to acute haemodynamic changes. Therefore, HemoSonic™ seems to be more predestined to fast cardiac output assessment in acute emergency situations and perioperatively than in sophisticated and long-term intensive care situations. Non-invasive CO2-rebreathing may be recommended as a less-invasive cardiac output and ventilation monitor in addition to a standard monitoring in intensive care and during anaesthesia, where invasive monitoring is not feasible for many reasons. The monitor does not seem to be suitable in acute situations and in patients with severe pulmonary disturbances or after extra-corporal circulation [5,17,18].
The replacement of PAC by less-invasive monitoring techniques is not the most important issue that has to be discussed. The new methods offer the possibility of applying the potential benefit of advanced haemodynamic monitoring to patients that are not in need of it per se . The employment of potentially less-invasive advanced haemodynamic monitoring techniques allows a more large-scaled application in lower-risk patients, with a reduced risk of endangering a patient by the monitoring technique itself.
Undoubtedly, PACs are useful and appropriate for special clinical questions such as pulmonary hypertension. Our animal model study demonstrated that the less-invasive and non-invasive methods of cardiac output measurement can offer sufficient cardiac output measurement, taking into account the differing characteristics in application, additional information and reflection of fast circulatory changes. Thus they possibly could cover a majority of patients who require advanced haemodynamic monitoring in intensive care and anaesthetic settings. If there is a patient- and situation-adapted application of the investigated techniques, there might be clear alternatives to pulmonary-artery-catheter-based cardiac output measurement, especially by transpulmonary thermodilution even under extreme hemodynamic changes. Less-invasive techniques may be applied in numerous clinical situations, adapted to the different patient needs and to suitability.
1. Connors Jr AF, Speroff T, Dawson NV et al
. The effectiveness of right heart catheterization in the initial care of critically ill patients. JAMA
2. Dalen JE, Bone RC. Is it time to pull the pulmonary artery catheter? JAMA
3. Gödje O, Hoeke K, LichtwarcK-Aschoff M et al
. Continuous cardiac output
by femoral arterial thermodilution calibrated pulse contour analysis: comparison with pulmonary arterial thermodilution. Crit Care Med
4. Valtier B, Cholley BP, Belot JP et al
. Noninvasive monitoring of cardiac output
in critically ill patients using transoesophageal Doppler. Am J Respir Crit Care Med
5. Heerden van PV, Baker S, Lim SI et al
. Clinical evaluation of the non-invasive cardiac output
(NICO) monitor in the intensive care unit. Anaesth Intens Care
6. Cariou A, Monchi M, Joly LM. Non invasive cardiac output
monitoring by aortic blood flow determination: evaluation of the sometec Dynemo 3000. Crit Care Med
7. Simanski O, Kaehler R, Pohl B, Hofmockel R. Control in general anaesthesia - a contribution. In: Adlassnig KP. Intelligent Systems in Patient Care
. European Network on Intelligent Technology for Smart Adaptive Systems and the Austrian Society of Artificial Intelligence. Vienna: Austrian Computer Society, 2001: 169-176.
8. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement
9. Ganz W, Swan HJ. Measurement
of blood flow by thermodilution. Am J Cardiol
10. Hartman J, Olszanski DA, Hullinger TG, Brunden MN. In vivo
validation of a transit-time flow meter. J Pharm Tox Meth
11. Lundell A, Bergquist D, Maattson E, Nilsson B. Volume flow measurements with a transit-time flowmeter - an in vivo
and in vitro
variability and validation study. Clin Physiol
12. Aadahl P, Aakhus S, Saether OD, Stomholm T, Myhre HO. Cardiac output measurement
during cross-clamping of the descending thoracic aorta in pigs. Acta Anaesthesiol Scand
13. Renner LE, Morton MJ, Sakuma GY. Indicator amount, temperature, and intrinsic cardiac output
affect thermodilution cardiac output
accuracy and reproducibility. Crit Care Med
14. Rödig G, Prasser C, Keyl et al
. Continuous cardiac output measurement
: pulse contour analysis vs thermodilution technique in cardiac surgical patients. Br J Anaesth
15. Gueugniaud PY, Muchada R, Moussa M et al
. Continuous oesophageal aortic blood flow echo-Doppler measurement
during general anaesthesia in infants. Can J Anaesth
16. Yem YS, Tang Y, Turner MJ, Baker AB. Sources of error in noninvasive pulmonary blood flow measurements by partial rebreathing: a computer model study. Anesthesiology
17. Nilsson LB, Eldrup N, Berthelsen PG. Lack of agreement between thermodilution and carbon dioxide-rebreathing cardiac output
. Acta Anaesthesiol Scand
18. Neuhäuser C, Müller M, Bräu M et al
. Partial CO2
rebreathing technique versus thermodilution: measurement
of cardiac output
before and after operations with extracorporal circulation. Anesthetist
19. Gan TJ, Soppitt A, Maroof M et al
. Goal directed intraoperative fluid administration reduces length of hospital stay after major surgery. Anesthesiology