Abstract
Background
The ultrasonic cardiac output monitor (USCOM; USCOM Pty. Ltd., Sydney, NSW, Australia) has been accepted as a noninvasive device for measuring cardiac function in various clinical conditions. The present study aimed at comparing the accuracy of this device with that of the thermodilution technique in recipients in the early postoperative period after liver transplantation.
Methods
Fifteen mechanically ventilated patients were studied on the first postoperative day after liver transplantation. We compared the left-sided and right-sided cardiac output (CO) determined by USCOM with that obtained from the thermodilution technique with a pulmonary artery catheter every 8 hours in the intensive care unit. Each patient received a total of four paired measurements. Bland-Altman analysis was used for bias and precision testing. The CO measured by USCOM and the thermodilution method were considered interchangeable if the limits of agreement lay within ± 1 L per minute or 20% of the mean CO.
Results
Forty-eight paired left-sided CO measurements were obtained from 12 patients. Three patients were excluded due to unacceptable signals. Comparison of these two techniques revealed a bias of 0.13 L per minute and limits of agreement at −0.65 L and 0.92 L per minute. Fifty-six paired right-sided CO measurements were obtained from 14 patients with one patient excluded due to an unobtainable optimal signal. A bias of 0.11 L per minute with limits of agreement at −0.51 L and 0.72 L per minute were found for these two techniques.
Conclusion
This is the first study to evaluate the accuracy of USCOM in the post-liver transplant setting. This device is accurate in measuring CO in liver transplant recipients postoperatively. Possible risks of arrhythmia, infection and pulmonary artery rupture can be avoided because of its noninvasive nature. USCOM should be considered as an alternative in hemodynamic monitoring after liver transplantation.
Keywords
cardiac output; hemodynamics; intensive care units; liver transplantation; monitoring, physiologic ultrasonics;
1. Introduction
Cardiac output (CO) monitoring plays a crucial role in the postoperative care of liver transplant patients.1 Since the invention of the balloon-directed thermistor-tipped pulmonary artery catheter (PAC) by Swan and Ganz in 1971,2 PAC has been the gold standard of CO monitoring in clinical practice. However, arrhythmia, infection and possible pulmonary artery disruption have always been concerns related to the use of a PAC and led to a growing interest in the development of noninvasive hemodynamic monitoring devices.3 Invasive procedures in liver transplant recipients may increase the risk of infection after the administration of immunosuppressive medication.4 A noninvasive reliable hemodynamic monitoring device could be a good substitute in contemporary clinical practice.
The ultrasonic CO monitor (USCOM; USCOM Pty. Ltd., Sydney, NSW, Australia) is a noninvasive device for measuring CO, utilizing the latest continuous wave Doppler technique. The flow profile is obtained by placing an ultrasonic transducer (2.2 MHz) over the chest. Right-sided CO is determined by placing the probe over the left parasternal position to calculate the transpulmonary blood flow, whereas left-sided CO is detected by placing the probe over the suprasternal position to calculate the transaortic blood flow. Optimization of the direction and the angle of the ultrasonic probe will obtain a good flow profile over the time-velocity spectral display. The velocity-time integral is calculated from the peak velocity of each cardiac contraction representing the distance a blood column travels with each stroke. In turn, the CO value is calculated from the two equations:
CO = heart rate × stroke volume
Stroke volume = velocity time integral × cross-sectional area
where heart rate can be derived from the cardiac cycle and the cross-sectional area is the crosssectional area of the chosen valve (i.e. pulmonary valve in the right-sided CO measurement and aortic valve in the left-sided CO measurement). The valve area is given by the height-based algorithm built into the machine.5
Recently, a substantial body of research has documented the accuracy of CO measurement by USCOM in patients after cardiac surgery.6,7 However, the accuracy of CO estimation by USCOM at high CO values remains ambiguous.6 Liver cirrhosis patients have a unique cardiovascular change characterized by hyperdynamic, hyporeactive circulation, low systemic vascular resistance and cirrhotic cardiomyopathy.8 This hyperdynamic circulatory state usually remains for weeks after transplantation.
Cardiac morbidity after successful liver transplantation may be as high as 20%,9 which may be detected early with a CO monitoring device. The objective of this study was to evaluate the reliability of CO measured by USCOM in the intensive care unit (ICU) in the early post-liver transplant period. To determine this, we compared CO measured by USCOM with that measured by the thermodilution PAC technique.
2. Methods
2.1. Patient preparation
After obtaining approval from the Institutional Ethics Committee of Chang Gung Memorial Hospital (No. 95-0910B) and written informed consent from all the patients, the present study was performed between January and May 2007. Fifteen patients (11 males, 4 females) with mean ± SD age of 53.3 ± 9.3 years (body weight, 67.1 ± 9.46 kg; height, 165.6 ± 7.36 cm) who underwent liver transplantation because of end-stage liver cirrhosis or acute liver failure were included. Each patient received a routine preoperative examination including an electrocardiogram, blood tests, arterial blood gas analysis and a pulmonary function test. Precordial echocardiography was performed to exclude moderate to severe valvular cardiac pathology, intracardiac shunt or severe intrapulmonary shunting. Other exclusion criteria included severe arrhythmia, wounds over the suprasternal, left parasternal or intercostal regions, an ICU stay less than 24 hours and unwillingness to take part in the clinical trial.
Routine intraoperative monitoring, such as an electrocardiogram, anesthetic gas analysis, blood pressure, and pulse oximetry, was carried out as per the institutional standard. Right radial artery cannulation was performed for arterial blood pressure monitoring. A 7.5 Fr pulmonary artery catheter (Opti-Q; Abbott Laboratories, North Chicago, IL, USA) was inserted through the right internal jugular vein for measuring CO. General anesthesia was induced with intravenous midazolam (0.15 mg/kg), fentanyl (3−5 μg/kg) and rocuronium bromide (0.8 mg/kg). Anesthesia was maintained by 1.5% isoflurane in oxygen, and cisatracurium was given for muscle relaxation. The procedure of liver transplantation was accomplished according to the standard guidelines of the hospital and all operations were performed by the same surgical team. At the end of the surgery, the patient was sent to the ICU. The room temperature was maintained at 20ºC, and the patient’s body temperature was kept within 35−37ºC using blankets, and infusion fluids were warmed to prevent hypothermia and shivering.
Propofol infusion 0.3−0.5 mg/kg per hour and intravenous fentanyl 50−100 μg per hour were given for postoperative sedation and analgesia. Chest X-ray was taken postoperatively to confirm the correct position of the PAC. Patients were mechanically ventilated for at least 1 day depending on their condition.
2.2. CO measurements
CO measurement was performed upon arrival of patients at the ICU, and was repeated every 8 hours on the first postoperative day in a stable hemodynamic condition. Each patient received a total of four sets of paired CO measurements, which were considered independent samples.
The thermodilution technique, the reference method, was performed by injecting a bolus of 10 mL of 0.9% saline at room temperature through the PAC port at the expiratory phase of ventilation and was calculated with a CO computer (Oximetrix; Abbott Laboratories). The average of four CO measurements by this method was recorded as COPAC, all of which were within a 10% range.
The USCOM, which noninvasively measures CO, was operated by aiming its probe at the areas of the aortic valve (COUS-A) and pulmonary valve (COUS-P). The left-sided CO (COUS-A) measurement was obtained by aiming the probe toward the aortic valve area on the suprasternal notch, while the right-sided CO (COUS-P) was measured with the probe placed over the 3rd to 5th parasternal intercostal space aimed at the pulmonary valve to search for optimal Doppler signals. The optimal Doppler signal is characterized by a sharp, welldefined waveform with the clearest audible sound. The optimal flow profile must also fulfill Dey and Sprivulis’ criteria,10 which stipulates: (1) a well defined image base and peak; (2) a clear commencement of flow or heart sound; (3) a well defined cessation of flow or heart sound; (4) an appropriate scale used with minimal acoustic interference.
The CO measurements by USCOM were undertaken by the same investigator (BC Su). A nurse anesthetist helped to perform the CO measurements with the thermodilution technique. The measurements of CO by the noninvasive ultrasonic technique and the thermodilution technique were carried out almost simultaneously when the patient was at the end-expiratory phase of the ventilation. The measurements of COUS-A and COUS-P were performed separately at an interval of at least 10 minutes. Each CO measurement was coupled with a reference, namely COPAC-A for COUS-A, and COPAC-P for COUS-P. Each operator was ignorant of the results obtained by their counterpart.
2.3. Statistical analysis
Statistical analysis was performed using MedCalc version 9.0.1.0 (MedCalc Inc., Mariakerke, Belgium). Scattergrams of the CO measurements were plotted. Pearson’s correlation coefficient was used to test the intrapatient variability due to repeated measurements over the same site. Regression with correlation coefficients was analyzed between COPAC-A and COUS-A as well as COPAC-P and COUS-P. The CO measurement derived from the reference method (COPAC-A, COPAC-P) was presented on the X axis, while the CO derived from the comparative method (COUS-A, COUS-P) was on the Y axis. A p value less than 0.05 was considered statistically significant.
The Bland-Altman analysis11 was used to estimate the bias and limits of agreement between measurements by the two methods on both left-sided and right-sided CO. The X axis was the average of two measurements by both methods, while the Y axis was the difference between these two methods. Bias was the mean difference between measurements by two methods. Limits of agreement were defined as mean bias ± 1.96 SD. The two methods of measuring CO were judged interchangeable if the limits of agreement lay within a threshold, set a priori at ± 1.0 L per minute or 20% of the mean CO.12,13
3. Results
The classification of these 15 patients was a median Child-Pugh score of C. Two patients were in acute liver failure, and 13 had end-stage liver diseases, of whom eight suffered from malignant hepatic tumors. Thirteen patients received living donor liver transplantation, and two patients received cadaveric livers.
Good Doppler signals could not be obtained over the aorta and pulmonary artery in three patients and one patient, respectively, due mainly to anatomic variability. The detection ability of USCOM was 80% on the left side and 93.3% on the right side. Patients in whom an optimal aortic flow could not be determined could usually obtain a good pulmonary flow, and vice versa. Pearson’s correlation coefficient for testing intraobject variability showed good consistency during USCOM measurement.
For left-sided CO, 48 pairs of measurements were obtained. The measurements of CO by the two different methods (COPAC-A and COUS-A) are shown in Figure 1, and are presented in the Bland-Altman plot (Figure 2). The correlation coefficient was 0.988 (p< 0.0001) according to the regression equation of y = 0.2989 + 0.9462x, where COPAC-A was represented on the X axis and COUS-A on the Y axis. The mean bias was 0.13 L per minute (95% CI, 0.015−0.248) with limits of agreement (mean bias ± 1.96 SD) of −0.65 and 0.92 L per minute. The overall percentage error (1.96 SD/mean CO) was ± 8.9%.
For right-sided CO, 56 pairs of measurements were obtained. The measurements of CO by the two different methods (COPAC-P and COUS-P) are shown in Figure 3 and are presented in the Bland-Altman plot (Figure 4). The correlation coefficient was 0.9950 (p< 0.0001) according to the regression equation of y = 0.2989 + 0.9462x, where COPAC-P was represented on the X axis and COUS-P on the Y axis. The mean bias was 0.11 L per minute (95% CI, 0.02−0.189) with limits of agreement (mean bias ± 1.96 SD) of −0.51 and 0.72 L per minute. The overall percentage error (1.96 SD/mean CO) was ± 7.2%.
Download full-size image
Download full-size image
Download full-size image
Download full-size image
4. Discussion
There are few studies which mention the evaluation of USCOM for measurement of CO in post-liver transplantation patients in the ICU. From the results of our study, it was found that measurements of CO, both left-sided and right-sided, by USCOM are interchangeable with those by the traditional thermodilution method. Each patient can have at least one optimal Doppler profile either over the aorta or the pulmonary artery.
The Bland-Altman bias and precision statistics is a well-established statistical method appropriate for comparison of the CO measurements between new and established devices.11−13 A meta-analysis of previous research of CO measurement showed that the limits of agreement up to ± 30%, or less than ± 1 L per minute are acceptable.13 In our study, the limits of agreement were set a priori at ± 1.0 L per minute or 20% of the mean CO. Our results showed limits of agreement at −0.65 to 0.92 L per minute for the left-sided CO and −0.51 to 0.72 L per minute for the right-sided CO between USCOM and the thermodilution method. In addition, the bias was very small (0.11−0.13 L per minute). We conclude that CO can be accurately measured by USCOM if the thermodilution method is considered the clinical standard reference method.
There has been no article evaluating the reliability of the use of USCOM in the post-liver transplantation period. However, there are two articles comparing the measurement of CO using USCOM and PAC in the post-cardiac surgery period.6,7
Tan et al demonstrated good agreement between the CO measurements by USCOM and by the thermodilution method in patients following cardiac surgery.6 However, USCOM tends to underestimate the real CO value when it is relatively high (> 6 L per minute). From their figure, Tan et al’s CO range mainly stayed between 3 and 7 L per minute, with most of the data clustering around 5 L per minute. On the contrary, such a difference with high CO does not appear in Chand et al’s research.7 This study is the first to focus on the accuracy of CO measured by USCOM at high CO values. Cirrhotic patients were chosen because of their unique hyperdynamic hemodynamic status,8,9 which appears and persists for weeks after liver transplantation. In our study, the CO value ranged from 2.8 to 13.6 L per minute with a median value of 9 L per minute. Even at high CO, USCOM can still reliably measure CO values. Possible explanations for the controversies include: (1) exclusion of cases with a “sick” heart, such as severe arrhythmia and valvular pathology; (2) careful optimization of the Doppler flow to capture the maximum optimal peak flow;10 and (3) more patients with CO > 6 L per minute.
Several factors are considered for accurate measurement of CO by USCOM: (1) the direction of the ultrasound beam and the blood flow must be nearly parallel; (2) the cross-sectional area is set by a nomogram of the pulmonary valve or aortic valve and failure to measure flow across the valve area, or severe valvular pathology can cause an error in CO measurement; (3) if there are no cardiac images during measurement, seeking of optimal flow from the suprasternal or intercostal space may catch signals elsewhere other than from the presumed valve in some anatomically variant patients; (4) the interference of severe arrhythmia with CO measurement still needs to be examined, since the built-in function of USCOM may fail to auto-trace the flow during severe arrhythmia, and manual operation is needed, i.e. pointing at the flow of peak velocity.14
Several techniques are considered as noninvasive methods for CO measurement but each has its drawbacks.15,16 Transesophageal echocardiography provides much diagnostic hemodynamic information.17 However, its routine use in the ICU is cumbersome, operator-dependent and contraindicated in patients with esophageal varices. NICO (Non In vasive Cardiac Output; Novametrix Medical Systems Inc., Wallingford, CT, USA) was developed based on the application of Fick’s principle to carbon dioxide. Inaccuracy in high CO status measurement precludes its use in liver failure, systemic inflammation and other similar situations.15,18 Esophageal Doppler calculates CO under the assumption of a constant ratio between caudal and cephalic blood flow. The CO value measured by esophageal Doppler and thermodilution correlates better with relative changes in CO rather than with absolute CO values.15,16,19 CO estimation utilizing a bioimpedance method measures the overall thoracic bioimpedance with thoracic surface electrodes. A recent meta-analysis showed that bioimpedance and thermodilution only moderately correlates in CO me asurement.20 Although improvements in the bioim pedance method have been made, it still awaits more clinical validation.21 Pulse contour analysis measures CO based on the theory that the arterial pressure waveform is related to the stroke volume. A moderate to good correlation has been reported.15,16 An arterial line and calibration with lithium22 or cold saline injection23 through a central venous line are needed. FloTrac (Edwards Life Science LLC, Irvine, CA, USA) is a new model where few clinical comparison studies have been presented and its use in CO estimation mandates further investigation.24 Finometer (Finapres Medical Systems, Amsterdam, The Netherlands) is a truly non invasive monitor which measures beat-to-beat blood pressure, CO, and systemic vascular resistance.25 CO derived from a Modelflow method, which carries a bias of 0.3 L per minute and a precision of 20%,26 requires calibration for more precise measurement and thus cannot replace thermodilution as a monitor for absolute levels of CO. 27
An accurate, rapid-to-use, noninvasive CO monitor is of clinical interest. USCOM has the potential to be the ideal noninvasive CO monitoring device for several reasons. It is truly noninvasive and no injection or arterial line is needed. It is easy to use and easy to learn. Two sites for CO measurement are possible, the suprasternal and left parasternal intercostal spaces. A good Doppler profile can be obtained from at least one of the two sites. In fact, other valuable parameters can be obtained by USCOM. Cardiac index, stroke volume, systemic vascular resistance, and systemic vascular resistance index can be calculated by input of body weight, body height and blood pressure of the patient. Stroke volume variance and corrected flow time are potent fluid status indicators in mechanically ventilated patients.28,29 However, there are also some shortcomings of USCOM in comparison with PAC. It does not provide pulmonary artery pressure readings and a means to measure mixed venous oxygenation, both of which play important roles in the clinical management of liver transplant patients postoperatively.
Several limitations exist in this study. First, all of our study patients were screened by electrocardiogram and echocardiography to exclude severe arrhythmia and valvular pathology. The use of USCOM in patients with severe arrhythmia and valvular pathology must be further examined. Besides, USCOM, unlike transesophageal echocardiography, cannot record the full investigative process. This made it impossible to separate the data selector from the USCOM performer which might add additional bias to the study result. Second, only four paired CO measurements were performed in each patient. Repeatability was not tested. It is mandatory to perform a comparison of the two methods over a longer time and wider clinical spectrum. However, repeatability of USCOM had been tested in dogs with good concordance.30 Another study conducted by our group also tested the repeatability of left-sided CO measured by USCOM and PAC intraoperatively, which also showed a good concordance.31 Last, the thermodilution method by PAC was chosen for the reference method because of it being the current clinical gold standard. Unfortunately, the thermodilution method has its own inherent variability.32−34
According to the experience of this study, there are some pointers for the clinical use of USCOM. Although it is an easy-to-use device, determination of optimal Doppler flow by interpretation of the flow profile requires experience. There was a small proportion of patients in whom we failed to measure optimal flow. Also, because of the use of a nomogram in assuming the cross-sectional area of the pulmonary or aortic valve, CO can be miscalculated in patients with severe valvular pathology. All of the above-mentioned issues may lead to misjudgment of the hemodynamic status. However, clinical misinterpretation will lessen as experience grows. Cautious interpretation of CO data obtained by USCOM as well as by other means is suggested.
In conclusion, USCOM is a useful device in CO determination in postoperative liver transplant patients. Clinical use of additional information, such as stroke volume variance and corrected flow time, merits further study. The use of USCOM in the perioperative period of liver transplantation is also promising and warrants further investigation. Other potential clinical uses include hemodynamic monitoring in the ICU, application to early goaldirected therapy, and intraoperative monitoring in noncardiac surgery.