Cardiac Output Determination From the Arterial Pressure Wave: Clinical Testing of a Novel Algorithm That Does Not Require Calibration

ORIGINAL ARTICLES

Cardiac Output Determination From the Arterial Pressure Wave: Clinical Testing of a Novel Algorithm That Does Not Require Calibration Gerard R. Manecke Jr, MD,* and William R. Auger, MD† Objective: The purpose of this study was to evaluate the accuracy and precision of a novel algorithm that evaluates cardiac output by using arterial pressure waveform characteristics. Design: Prospective, observational study comparing the cardiac output values of intermittent thermodilution, continuous thermodilution, and continuous arterial pressure wave assessment. Setting: The intensive care unit in a tertiary care university hospital. Participants: Fifty postoperative cardiac surgical patients, within the first 12 hours after surgery. Interventions: All patients received a pulmonary artery catheter (PAC) and at least 1 systemic arterial pressure catheter. The data from the arterial catheter were processed by using a new arterial pressure cardiac output (APCO) algorithm. The data from the PAC (continuous and intermittent assessments) were collected for comparison with the APCO. Measurements: Two hundred ninety-five cardiac output measurements using intermittent thermodilution (ICO), continuous thermodilution (CCO), and arterial pressure-based

output (APCO) were obtained during various times during the first 12 postoperative hours. The measurements of each method at each time point were compared by using BlandAltman analysis. Results: The mean cardiac output ranged from 2.77 to 9.60 L/min. APCO, compared with ICO, revealed a bias of 0.55 L/min and precision of 0.98 L/min. APCO, compared with CCO, revealed a bias of 0.06 L/min and precision of 1.06 L/min. The APCO agreement between femoral and radial arterial catheters was close; the bias was ⴚ0.15 L/min, and the precision was 0.56 L/min. Conclusions: This novel arterial pressure cardiac output algorithm provides cardiac output assessments that agree satisfactorily for clinical purposes with intermittent and continuous thermodilution techniques in postoperative cardiac surgical patients. Further study is required for other patient populations and clinical situations. © 2007 Elsevier Inc. All rights reserved.

A

refused, if pulmonary artery catheterization was contraindicated (eg, pulmonary artery tumor and superior vena cava obstruction), or if they presented with clinically significant intracardiac shunts. Patients were also excluded if, after surgery, moderate or severe tricuspid regurgitation, aortic regurgitation, or aortic stenosis were evident on transesophageal echocardiography. Of the 69 patients enrolled, the first 12 were studied for final adjustments of the algorithm. Of the remaining 57, 7 were excluded because of technical factors. These included inability to place a pulmonary artery catheter (2), unreliable arterial pressure waveform (2, overdamped), and equipment failure in data collection and/or processing (3). In the remaining 50, the APCO algorithm was tested against ICO and CCO by using a continuous cardiac output pulmonary artery catheter (Swan-Ganz CCOmbo V, Edwards Lifesciences, LLC). Thirty of these patients underwent pulmonary thromboendarterectomy (PTE),

SSESSMENT OF CARDIAC OUTPUT in critically ill patients may be useful, and pulmonary artery catheterization with thermodilution has generally been accepted as the clinical gold standard. This method, either intermittent or continuous, is robust but has the disadvantage of requiring the placement of a pulmonary artery catheter. Other methods of determining cardiac output have been introduced, including continuous monitors using the arterial pressure waveform analysis. Edwards Lifesciences, LLC (Irvine, CA) has recently introduced a system for monitoring cardiac output continuously (Vigileo/FloTrac) that does not require thermodilution or dye dilution. Instead, it calculates the cardiac output by using arterial waveform characteristics in conjunction with patient demographic data. It is unique among commercially available arterial waveform cardiac output systems in that it does not require calibration with another method. The goal of this study was to test the accuracy and precision of the algorithm used by the Vigileo/FloTrac system in patients presenting to the intensive care unit after cardiac surgery. This was done by comparing the arterial pressure-based cardiac output (APCO) with intermittent pulmonary artery catheter thermodilution (ICO) and continuous thermodilution (CCO). METHODS

After approval of the institutional Human Subjects Protection Program and written informed consent was obtained, 69 adult patients scheduled to undergo cardiac surgery in a single-university hospital setting were enrolled. Patients were excluded from the study if they

KEY WORDS: cardiac output measurement, arterial pressure-based cardiac output, thermodilution cardiac output

From the Departments of *Anesthesiology and †Internal Medicine, UCSD Medical Center, San Diego, CA. Supported in part by a clinical trial agreement with Edwards Lifesciences, LLC, Irvine, CA. Institutional support was provided by UCSD Departments of Medicine and Anesthesiology. Presented in part at the 34th Annual Society of Critical Care Medicine Conference, Phoenix, AZ, January 15-19, 2005. Address reprint requests to Gerard R. Manecke Jr, MD, Department of Anesthesiology, UCSD Medical Center, 200 West Arbor Drive, San Diego, CA 92103. E-mail: [email protected] © 2007 Elsevier Inc. All rights reserved. 1053-0770/07/2101-0002$32.00/0 doi:10.1053/j.jvca.2006.08.004

Journal of Cardiothoracic and Vascular Anesthesia, Vol 21, No 1 (February), 2007: pp 3-7

3

4

MANECKE AND AUGER

Table 1. Patient Demographic Characteristics Sex, male (%) Age (y) Height (mean ⫹ SD) (cm) Weight (mean ⫹ SD) (kg) BSA (mean ⫹ SD) (m2) CABG surgery (%) PTE (%)

28 (56) 61.2 ⫾ 14 171.1 ⫾ 11.7 83.2 ⫾ 17.7 2.0 ⫾ 0.2 21 (42) 30 (60)

20 coronary artery bypass graft (CABG) surgery, and 1 a combined CABG/PTE. All PTEs were performed by using cardiopulmonary bypass (CPB), hypothermia, and intermittent circulatory arrest; whereas the CABG surgeries were performed at normothermia without CPB (“beating heart”). In the operating room, in addition to continuous electrocardiograms, pulse oximetry, end-tidal carbon dioxide, and noninvasive blood pressure monitoring, all patients received a 20-G 1.88” radial arterial catheter (Becton Dickinson, Sandy, UT) on the side with the more prominent pulse. The radial arterial catheters were placed before anesthetic induction under local anesthesia. Thirty-one of the patients (all PTE patients) also had an 18-G 6” femoral arterial catheter (Argon Medical, Athens, TX) placed, and this was done after induction. It is standard practice in this institution to place a femoral arterial catheter in patients undergoing PTE because, after the required hypothermia and circulatory arrest, the radial arterial pressure may not be indicative of central arterial pressure.1 All patients received a right internal jugular introducer (AVA HF Triple Lumen, Edwards Lifesciences, LLC) and pulmonary artery catheter capable of both intermittent and continuous thermodilution (see earlier). The pulmonary artery catheter was connected to its companion processing unit (Vigilance, Edwards Lifesciences, LLC). Anesthetic induction consisted of the intravenous administration of midazolam, 0.15 mg/kg, fentanyl, 10 to 15 ␮g/kg, and vecuronium, 0.1 mg/kg. Anesthesia was maintained with isoflurane, 0.5% to 2% in oxygen, and propofol, 25 to 75 ␮g/kg/min. Separation from CPB was facilitated with dopamine, 3 to 7 ␮g/kg/min, and epinephrine, 0.05 to 0.1 ␮g/kg/min, if necessary, in the PTE patients; and dopamine, 2 to 5 ␮g/kg/min, was used for hemodynamic support in the CABG patients, as indicated. Nitroprusside was used as the primary vasodilator in the operating room and intensive care unit, titrated to maintain mean arterial pressure in the 60- to 80-mmHg range. Inotropes were continued in the postoperative period as necessary. In 25 patients, the radial arterial pressure wave was used for APCO calculation; in the other 25, the femoral arterial pressure wave was used. Of the 31 patients with a femoral catheter, 21 of them were randomly selected for comparison of the APCO calculations, femoral versus radial. The arterial catheter signals were digitized at 100 Hz by using an analog-to-digital converter (DAQPad 6020E; National Instruments, Austin, TX) and routed to a personal laptop computer for off-line analysis. APCO was calculated by applying the algorithm to the digitized waves. Likewise, the data output from the Vigilance module were routed to the same computer for storage of ICO and CCO. The time period for analysis was the 12-hour period after surgery, and cardiac output comparison times were chosen from the ICO times dictated by protocols in the intensive care unit (every 2 hours unless clinical conditions dictated otherwise). At each selected ICO collection time, approximately 4 bolus injections were performed. The injections were performed at random times within the respiratory cycle, within an approximate 5-minute time interval. The injections consisted of 10 mL of room-temperature 5% dextrose in water, according to the protocols of the intensive care unit. A batch of ICO injections was considered valid if it contained at least 3 replicates. The average of the valid batch of ICO replicates was

considered as the analysis endpoint at a given ICO collection time. For each ICO collection time, an ICO observation was considered to be an outlier if it was ⬎50% different from the median of the ICOs in the individual series. For each selected ICO collection time, the average of the nearest 3 CCO measurements before the bolus ICO injectates and nearest 3 CCO measurements after the bolus injectates defined the analysis endpoints (CCO measurements during the period of ICO determination were not used). The time interval for observing 6 CCO measurements was approximately 5 minutes. To ensure that there was no overlap of comparison sets, the midpoint of a valid set was required to be at least 30 minutes from the previous valid set. The average APCO value (6 measurements) for each ICO/CCO determination period was used for comparison. Bland-Altman analysis was used to compare the ICO, CCO, and APCO determinations, as well as the femoral and arterial APCO values. RESULTS

Thirty of the patients had received PTE for chronic thromboembolic pulmonary hypertension, and 21 had undergone CABG surgery for coronary artery disease. The demographic characteristics are presented in Table 1. Four of the patients had type 2 diabetes mellitus, and 4 had a history of congestive heart failure. No patients exhibited significant aortic valvular regurgitation or stenosis. No patients had to be excluded on the basis of postsurgery valvular abnormalities. Those with significant tricuspid regurgitation in the preoperative period (PTE patients) no longer exhibited this postoperatively, presumably because of successful pulmonary endarterectomy. The mean cardiac output ranged from 2.77 to 9.60 L/min. Mean arterial pressure was successfully maintained in the 60to 80-mmHg range by using modest doses of nitroprusside, propofol, and inotropes as described in the methods section. No patients required vasoconstrictors (norepinephrine or vasopressin) or high-dose inotropic support. A total of 295 measurement pairs were obtained comparing APCO and ICO. The BlandAltman analysis for the comparison is shown in Figure 1. The

Fig 1. Bland-Altman analysis comparing APCO with that determined by ICO. The bias was 0.55 L/min (solid line), and the precision (standard deviation) was 0.98 L/min. The dashed lines indicate the limits of agreement (2 times the standard deviation). (Color version of figure is available online.)

CARDIAC OUTPUT DETERMINATION

Fig 2. Bland-Altman analysis comparing APCO with that determined by CCO. The bias was 0.06 L/min (solid line), and the precision (standard deviation) was 1.06 L/min. The dashed lines indicate the limits of agreement (2 times the standard deviation). (Color version of figure is available online.)

bias was 0.55 L/min, and the precision (standard deviation) was 0.98 L/min. APCO versus CCO comparison, also on 295 measurements, revealed a bias of 0.06 L/min, with precision of 1.05 L/min (Fig 2). Bland-Altman comparison of APCO derived from femoral and radial arterial catheters (132 measurements) is shown in Figure 3. The bias between the measurements was ⫺0.15 L/min, and the precision was 0.56 L/min. For informa-

Fig 3. Bland-Altman analysis comparing APCO from the femoral arterial catheter with that derived from the radial arterial catheter in 21 patients. The bias was ⴚ0.15 L/min (solid line), and the precision (standard deviation) was 0.56 L/min. The dashed lines indicate the limits of agreement (2 times the standard deviation). (Color version of figure is available online.)

5

Fig 4. Bland-Altman analysis comparing ICO with CCO. The bias was 0.50 L/min (solid line), and the precision (standard deviation) was 0.66 L/min. The dashed lines indicate the limits of agreement (2 times the standard deviation). (Color version of figure is available online.)

tion purposes, the ICO and CCO measurements were also compared, showing close agreement (Fig 4). DISCUSSION

Thermodilution is widely accepted as a clinical “gold standard” for assessment of cardiac output. It has been adopted globally for the measurement of cardiac output in the intensive care unit setting. However, pulmonary artery catheterization has been associated with serious complications, some fatal.2-4 For that reason, alternative methods of determining cardiac output have been sought. One of the most attractive means of determining cardiac output in a minimally invasive or noninvasive way is by analysis of the arterial pulse wave. The possibility of determining the cardiac output by using the arterial pulse wave has intrigued both scientists and clinicians for decades.5-8 Preliminary successes have been achieved by using techniques involving determination of the area under the arterial pressure curve as well as other methods involving pulsatility and various subtleties of the wave.9-11 An issue has been quantifying the relationship between the amount of blood flow and the pressure wave associated with it. This relationship can vary widely from one individual to another and in a single individual as clinical conditions change. Knowing this relationship for an individual patient and circumstance allows for the calculation of a constant (K), which can be used for subsequent cardiac output assessments. Thus, techniques using the arterial wave have heretofore required calibration with another method of cardiac output assessment. These include the PulseCO system (LiDCO Group, London, UK), which requires calibration using lithium dilution12; the PiCCO system (Pulsion SG, Munich, Germany), requiring transpulmonary thermodilution13; and the Finapres Modelflow system (Finapres Medical Systems, Amsterdam, The Netherlands), with which calibration with another means of cardiac output measurement is advisable

6

to achieve acceptable accuracy.14 The algorithm tested in this study does not require such calibration. The algorithm calculates the arterial pressure by using arterial pulsatility, resistance, and compliance, according to the following general equation: stroke volume ⫽ K * pulsatility, where K is a constant quantifying arterial compliance and vascular resistance, and pulsatility is derived from the standard deviation of the arterial pressure wave over a 20-second interval. K is derived from patient characteristics (sex, age, height, and weight) according to the method described by Langewouters et al.15 The Langewouters’ method involves using aortic compliance data generated from cadaver studies as a “starting point” for the K value. Arterial waveform characteristics (eg, skewness and kurtosis of individual waves) are then used to “fine tune” the value. For example, a wave skewed to the left indicates noncompliance of the vascular tree as does a wave with pronounced kurtosis (shortness). A large, rounded arterial wave, shifted to the right, is more indicative of a compliant arterial tree. This calibration constant is recalculated every minute. The results indicate that the algorithm used for the Edwards Vigileo/FloTrac system is accurate and precise when used in the postoperative period after cardiac surgery. The reported values are reliable over a wide range of cardiac outputs (2.779.60 L/min). The postoperative period after cardiac surgery is a period of rapid changes in hemodynamics, with temperature changes, inotropes, and vasopressors resulting in wide swings in cardiac output values and systemic vascular resistance. Although there are many studies comparing arterial pulse– determined cardiac output with thermodilution, a Medline search revealed only 3 studies involving commercially available APCO devices performed in the post– cardiac surgery period. One such study investigated the lithium dilution-calibrated PulseCO method12 and the other 2 the transpulmonary thermodilution PiCCO-calibrated method.16,17 The APCO algorithm compares favorably with the PiCCO method in the postoperative period in which the PiCCO method showed a bias of ⫺0.40 L/min and precision of 1.30 L/min when compared with intermittent thermodilution measurements. In that same study, PiCCO showed a bias of ⫺0.12 L/min and precision of 1.33 L/min when compared with continuous thermodilution.16 Zollner et al17 obtained similar results in a postoperative study of the PiCCO method (bias PiCCO-thermodilution 0.31 L/min and precision 1.25 L/min). Hamilton et al12 performed a more limited study (20 relatively healthy patients after CABG surgery) of the PulseCO system. Although the actual values for bias and precision were not reported, the Bland-Altman plot they presented revealed very good bias and precision. Also, an experimental method of using standard monitors to determine CO requiring calibration with another method is being developed. This method was recently tested in post– cardiac surgical patients, with good results.18 To the authors’ knowledge, this is the first commercially available arterial pressure– based cardiac output algorithm that does not require calibration to show such a high level of accuracy. The system simply consists of a transducer/preprocessor (FloTrac) connected to a processing/display unit (Vigileo). Preparation for monitoring cardiac output consists only of

MANECKE AND AUGER

entering the age, height, weight, and gender into the unit and zeroing the transducer. The ease of use and setup makes this technology particularly attractive in critical situations. Because hemodynamic changes can occur very rapidly in the postoperative period after cardiothoracic surgery, the fact that the system is continuous (values are updated every 20 seconds) is potentially advantageous. The results indicate that, under the study conditions, the radial artery and the femoral artery are both reliable in assessing cardiac output. In 21 of the patients in whom both cannulae were present, the femoral and radial APCO values showed close agreement. This attribute of the system will be useful, in particular when femoral cannulation is contraindicated. The fact that the radial artery wave is sufficient represents an advantage over the PiCCO system, which uses transpulmonary thermodilution for calibration,19 and over the PulseCO system, which requires a lithium dilution calibration.12 The system using this algorithm reports cardiac output, cardiac index, systemic vascular resistance (if a central venous pressure monitor is present), stroke volume variation, and venous oxygen saturation (if a venous oximeter catheter is used). Its capabilities and characteristics have recently been described in detail.20 However, it cannot report the advanced volumetrics that other more invasive systems can provide. The PiCCO system, for example, assesses intrathoracic blood volume, and the Edwards continuous cardiac output pulmonary artery catheter system (Vigilance) can provide right ventricular ejection fraction and end-diastolic volumes. Also, pulmonary artery catheters can determine pulmonary artery pressures, including pulmonary artery occlusion pressure, as well as mixed venous oxygen saturation. Limitations of the device also include those factors that limit the utility of arterial catheters in general, including arterial spasm, dissection, catheter kinking, kinked tubing, underdamping, and overdamping. The accuracy of the cardiac output data reported by the device depends on an arterial waveform of good fidelity. There are some significant limitations to this study. None of the patients suffered aortic regurgitation or stenosis, and none had atrial fibrillation. Both of these conditions have caused inaccuracy in arterial pressure– derived cardiac output. Also, the patients in this study were reasonably stable, not requiring high-dose vasopressor or inotropic support. Future studies should include unstable patients, such as those in septic or cardiogenic shock. The study only tested the algorithm in a specific subgroup of patients and did not assess the performance of the algorithm in the operating room. Further research on the system is in order, on a wider variety of patients both in the intensive care unit and the operating room. In particular, the FloTrac should be tested during cardiac surgery, when potentially confounding factors such as rapid temperature changes, cardiac manipulation, and rapid changes in patient positioning are present. Also, future studies should include physiologic manipulations such as fluid loading, inotropes, vasodilators, and vasopressors so as to further test the device’s ability to track changes in individual patients. The algorithm studied allows for accurate cardiac output assessment in postoperative cardiac surgical patients. This technology is promising and requires further investigation.

CARDIAC OUTPUT DETERMINATION

7

REFERENCES 1. Manecke GR Jr, Parimucha M, Stratmann G, et al: Deep hypothermic circulatory arrest and the femoral-to-radial arterial pressure gradient. J Cardiothorac Vasc Anesth 18:175-179, 2004 2. Huang L, Elsharydah A, Nawabi A, et al: Entrapment of pulmonary artery catheter in a suture at the inferior vena cava cannulation site. J Clin Anesth 16:557-559, 2004 3. Abreu AR, Campos MA, Krieger BP: Pulmonary artery rupture induced by a pulmonary artery catheter: A case report and review of the literature. J Intensive Care Med 19:291-296, 2004 4. Bossert T, Schmitt DV, Mohr FW: [Pulmonary artery rupture— Fatal complication of catheter examination (thermodilution catheter)]. Z Kardiol 89:54, 2000 5. Huggins RA, Smith EL, Sinclair MA: Comparison of cardiac output by the direct Fick and pressure pulse contour methods in the open-chest dog. Am J Physiol 159:385-388, 1949 6. Williams RR, Wray RB, Tsagaris TJ, et al: Computer estimation of stroke volume from aortic pulse contour in dogs and humans. Cardiology 59:350-366, 1974 7. English JB, Hodges MR, Sentker C, et al: Comparison of aortic pulse-wave contour analysis and thermodilution methods of measuring cardiac output during anesthesia in the dog. Anesthesiology 52:56-61, 1980 8. Brock-Utne JG, Blake GT, Bosenberg AT, et al: An evaluation of the pulse-contour method of measuring cardiac output. S Afr Med J 66:451-453, 1984 9. Della Rocca G, Costa MG, Coccia C, et al: Cardiac output monitoring: aortic transpulmonary thermodilution and pulse contour analysis agree with standard thermodilution methods in patients undergoing lung transplantation. Can J Anaesth 50:707-711, 2003 10. Jansen JR, Schreuder JJ, Mulier JP, et al: A comparison of cardiac output derived from the arterial pressure wave against thermodilution in cardiac surgery patients. Br J Anaesth 87:212-222, 2001 11. Mukkamala R, Reisner AT, Hojman HM, et al: Continuous cardiac output monitoring by peripheral blood pressure waveform analysis. IEEE Trans Biomed Eng 53:459-467, 2006

12. Hamilton TT, Huber LM, Jessen ME: PulseCO: A less-invasive method to monitor cardiac output from arterial pressure after cardiac surgery. Ann Thorac Surg 74:S1408-S1412, 2002 13. Della Rocca G, Costa MG, Pompei L, et al: Continuous and intermittent cardiac output measurement: Pulmonary artery catheter versus aortic transpulmonary technique. Br J Anaesth 88:350-356, 2002 14. Remmen JJ, Aengevaeren WR, Verheugt FW, et al: Finapres arterial pulse wave analysis with Modelflow is not a reliable noninvasive method for assessment of cardiac output. Clin Sci (Lond) 103:143-149, 2002 15. Langewouters GJ, Wesseling KH, Goedhard WJ: The pressuredependent dynamic elasticity of 35 thoracic and 16 abdominal human aortas in vitro described by a five-component model. J Biomech 18: 613-620, 1985 16. Mielck F, Buhre W, Hanekop G, et al: Comparison of continuous cardiac output measurements in patients after cardiac surgery. J Cardiothorac Vasc Anesth 17:211-216, 2003 17. Zollner C, Haller M, Weis M, et al: Beat-to-beat measurement of cardiac output by intravascular pulse contour analysis: A prospective criterion standard study in patients after cardiac surgery. J Cardiothorac Vasc Anesth 14:125-129, 2000 18. Ishihara H, Okawa H, Tanabe K, et al: A new noninvasive continuous cardiac output trend solely utilizing routine cardiovascular monitors. J Clin Monit Comput 18:313-320, 2004 19. Goedje 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 27:2407-2412, 1999 20. Manecke GR: Edwards FloTrac sensor and Vigileo monitor: Easy, accurate, reliable cardiac output assessment using the arterial pulse wave. Expert Rev Med Devices 2:523-527, 2005