Noninvasive Continuous Cardiac Output by the Nexfin Before and After Preload-Modifying Maneuvers

A Comparison with Intermittent Thermodilution Cardiac Output

Abstract

BACKGROUND: 

The Nexfin uses an uncalibrated pulse contour method for the continuous measurement of cardiac output (CO) in a totally noninvasive manner. Since the accuracy of pulse contour methods and their ability to track changes in CO have been repeatedly questioned, we have compared the CO measured by the Nexfin (NAPCO) with the CO measured by the pulmonary artery catheter (PACCO) in cardiosurgical patients before and after preload-modifying maneuvers.

METHODS: 

Twenty-eight patients who underwent on-pump cardiac surgery, of whom 18 were receiving vasopressor and/or inotropic therapy, were studied during the first postoperative hours. Preload modification, in the form of either a fluid challenge or a passive leg raising maneuver, was done whenever clinically indicated, with PACCO and NAPCO being simultaneously measured before and after each intervention.

RESULTS: 

A fluid challenge was administered to 22 patients, and the passive leg raising maneuver was performed in 6 patients. These interventions were repeated in 19 patients producing a total of 47 pairs of measurements. At baseline, mean (±SD) CO was 4.9 ± 1.1 and 5.0 ± 1.4 L·min⁻¹, for the PACCO and NAPCO, respectively, bias 0.1 ± 1.0, 95% prediction interval −2.5 to 2.4 L·min⁻¹, and 39% of error. After preload modification, the mean CO was 5.6 ± 1.3 and 5.6± 1.5 L·min⁻¹ for the PACCO and NAPCO, respectively, bias −0.0 ± 1.1, 95% prediction interval −2.6 to 2.7 L·min⁻¹, and 38% of error. The correlation coefficients (r) between the PACCO and NAPCO before and after preload modification were 0.71 (95% confidence interval [95% CI], 0.53–0.82) and 0.70 (95% CI, 0.52–0.82), respectively. Preload modification induced similar absolute changes in PACCO and NAPCO (r = 0.9, P < 0.0001). A 4-quadrant scatter plot showed a concordance rate of 100% (95% CI, 80.5%–100%) between the changes in NAPCO and PACCO. Polar plot analysis demonstrated a small polar angle and radial limits of agreement well below the 30° benchmark. The area under a receiver operating characteristic curve, testing the ability of Nexfin to detect an increase of ≥15% in PACCO, was 0.974 (95% CI, 0.93–0.99).

CONCLUSIONS: 

Although the Nexfin has limited accuracy when compared with the pulmonary artery catheter, it can reliably track preload-induced changes in CO in stable patients after cardiac surgery in the presence of moderate vasopressor and inotropic therapy. This ability, combined with its total noninvasiveness, fast installation, and ease of use, make the Nexfin a suitable monitor for the perioperative continuous measurement of CO. The reliability of this monitor in tracking the CO when significant changes in peripheral resistance take place still needs to be established.

In Brief

Published ahead of print June 11, 2013.

The measurement of cardiac output (CO) has been traditionally limited to severely ill patients in the intensive care unit (ICU). However, newer technologies have made this main determinant of oxygen delivery to be more available at the bedside in a continuous and less-invasive manner.^1,2 Many reports have also shown that goal-directed therapy aimed at maximizing the CO may significantly improve outcome in patients undergoing high-risk surgery.^3,4 Many of the CO monitors that have been introduced into clinical practice use a pulse contour method for the determination of continuous CO (CCO).⁵ The value of the CCO which is measured by some of these devices is independently calculated by proprietary algorithms and is not dependent on calibration against a better established technique (e.g., thermodilution). The clinical use of these devices has been under debate since their accuracy has been repeatedly questioned.^6–8 Others, however, have claimed that the fundamental question is whether the new device can replace thermodilution as a guide to clinical decisions,^9,10 and that CO obtained by a less invasive technique, even if slightly less accurate, may be preferable if it can be obtained more rapidly and easily, and track the short-term effects of diagnostic and therapeutic interventions.^2,10

One of the recently introduced CCO monitors is the Nexfin (BMEYE, Amsterdam, Netherlands) which uses an inflatable finger cuff as the only interface to the patient.¹¹ Since the noninvasiveness of this device makes it especially suitable for use in the perioperative period, we have compared it with the pulmonary artery catheter (PAC) in patients who were subjected to preload-modifying maneuvers (PMMs) during the immediate postoperative period after cardiac surgery.

METHODS

This prospective clinical study was conducted in an ICU of a tertiary care teaching hospital after obtaining institutional ethics committee approval and written patient consent. The study included patients who had undergone elective on-pump cardiac surgery and who had the following inclusion criteria: written informed consent; age >18 years; the presence of a PAC that was inserted intraoperatively due to medical indications unrelated to the study; and clinical indication for intravascular fluid administration or for clarifying fluid responsiveness by passive leg raising (PLR) as determined by the ICU medical staff.

Exclusion criteria included emergency surgery, intracardiac shunt, severe residual valvular disease, mechanical circulatory support, and renal failure requiring hemo- or peritoneal dialysis. Since the performance of the Nexfin may be compromised by the presence of severe vasoconstriction and significant finger edema, we have also excluded patients with body temperature <36°C, ejection fraction <40%, significant hemodynamic instability defined as systolic arterial blood pressure <90 mm Hg, high-dose norepinephrine (defined as >0.1 µg·kg·min⁻¹), clear evidence of excessive intravascular volume, and significant edema of the fingers.

At the end of the surgical procedure and after the transfer to the ICU, all patients were treated according to the local routine clinical protocol. For the first 6 postoperative hours, all patients’ lungs were mechanically ventilated with a tidal volume of 8 to 10 mL·kg⁻¹ body weight, peak pressure <25 cm H₂O, and a positive end-expiratory pressure of 5 cm H₂O. Neuromuscular blockade, analgesia, and sedation were achieved with IV pancuronium, morphine, and propofol, when needed. A PAC (Edwards Lifesciences, Irvine, CA) was introduced in the operating room after induction of anesthesia and connected to a monitor (Spacelabs Ultraview SL, Issaquah, WA) on arrival to the ICU. Hemodynamic status was stabilized with fluids, inotropes, and/or vasoactive drugs at the discretion of the ICU staff.

All eligible patients were connected to a Nexfin monitor by an inflatable finger cuff that was placed on the mid-phalanx of the middle or index finger of the hand with no radial artery catheter. The Nexfin continuously measures the finger arterial pressure using the volume clamp method and the Physiocal calibration.^12,13 After the reconstruction of a brachial pressure waveform from the finger pressure waveform using a transfer function, the pulsatile systolic area (the time integral of the pressure curve between the beginning of the systolic upstroke and the dicrotic notch) is calculated for each beat. The arterial input impedance is simultaneously determined from a 3-element Windkessel model, considering the nonlinear pressure dependency of the model elements and patient-specific data (gender, age, height, weight) which are entered into the monitor. The division of the pulsatile systolic area by the impedance produces the stroke volume on a beat-to-beat basis, while for each beat, mean arterial blood pressure is used to correct the elements of the model that are pressure-dependent in a nonlinear way.¹⁴

During the first 6 postoperative hours in the ICU, whenever the physician in charge considered a patient to be in need of a fluid challenge¹⁵ or a PLR maneuver¹⁶ to assess fluid responsiveness status, a baseline set of hemodynamic measurements was done before the required intervention. Hemodynamic measurements included the measurement of CO by both the Nexfin (NAPCO) and by the PAC (PACCO). PACCO was measured by a triplicate injection of a bolus of cold (<10°C) saline through the proximal port of the PAC, done randomly throughout the respiratory cycle and completed within 3 seconds. The morphology of the thermodilution curve was inspected to identify any artifacts. When 1 of the first 3 measurements differed by >10%, 2 more measurements were done. The average of the 3 closest measurements was used as the PACCO value. To minimize variation between investigators, the thermodilution measurements were performed by 1 of 2 well-experienced investigators (SIB-Tor IM). Each PACCO measurement was immediately followed by the recording of the corresponding NAPCO. The NAPCO value was obtained by averaging the Nexfin CCO for 15 seconds, using the averaging option that is included in the monitor.

Immediately after the completion of the baseline measurements, 1 of the 2 PMM procedures was performed. These included either the infusion of 500 mL of a gelatin solution over 15 minutes or the performance of a PLR maneuver. The PLR maneuver was performed using automatic bed tilt, moving the patient from a 45° head-up position to a supine position with 45° legs elevation, as previously described.¹⁶ The measurements of PACCO and NAPCO were started 45 seconds after achieving the PLR position and were completed within the next 3 minutes. A second identical set of hemodynamic measurements was done either immediately after the completion of the fluid administration or immediately after the PLR. It is of note that the Nexfin includes a Heart Reference System which is supposed to compensate for hydrostatic pressure changes which may occur during the PLR maneuver.

Statistics

Mean values, 95% confidence intervals (95% CIs), and 95% prediction intervals (95% PIs) were calculated using Student t test distribution for the measurements presented. Student t test was used to test if the intercept was different than zero. Accuracy and precision were derived from the values of the mean ± SD of the NAPCO–PACCO differences during baseline (BL), postmaneuver (PM), and for all results (ALL). Mean CO, bias (the mean difference between the PACCO and NAPCO), precision (±1.96 SD of the mean difference), 95% CI and 95% PI (2.5% lower confidence limit for the lower limit and 97.5% upper confidence limit for the upper limit), the percentage of error (ratio of precision to mean CO), and the upper confidence limit of the percentage error were computed for PACCO and NAPCO at BL, PM, and ALL. A linear regression and Bland–Altman analysis were used to assess agreement and precision. However, it has recently been claimed that the classic Bland–Altman analysis is not enough to define the agreement between 2 different methods of measuring CO since the precision of the reference technique, i.e., intermittent thermodilution by the PAC, has to be taken into consideration as well.¹⁷ We have therefore calculated the coefficients of variation (CV) (SD/mean, %) and the coefficients of error (CE = CV/√3) for each triplicate measurement of PACCO and NAPCO. The mean precisions of the PACCO and the NAPCO were calculated as precision = 2 × CE. We also calculated the least significant change (LSC) in CO (1.96 × CE × √2) for each technique (the minimal change that can detect a significant change in CO with a probability of 95%).

To assess the ability of the Nexfin to track the changes induced by the PMM, we used a 4-quadrant scatter plot of the respective changes in PACCO and NAPCO and performed a concordance analysis as recently suggested.⁸ The concordance rate is the percentage of data points in which the direction of change in CO measured by the 2 devices is in agreement. To eliminate the less-predictive data points that lie near the center of the plot, we applied an exclusion zone of 15% and considered 90% to 95% of concordance rate to indicate reliable trending ability.⁸ In addition, we have constructed a polar plot and calculated the radial limits of agreement and the polar angle, as recently proposed.^18,19 The benchmark values that are considered to signify good tracking ability of the tested CO device compared with the PAC are <30° angles for the radial limits of agreement (which are based on the 95% confidence limits) and <5° for the polar angle, indicating a good agreement between the calibration and reference methods of the 2 devices being compared.^18,19 Additionally, we used a receiver operating characteristic curve to test the ability of the Nexfin to detect an increase in PACCO of ≥15%. The statistical analysis was performed using Excel (Microsoft, Redmond, WA) and SPSS version 15.0 statistical package (SPSS, Chicago, IL).

RESULTS

We enrolled 28 patients in the study (20 males and 8 females; age 62 ± 11 years; weight 80 ± 17 kg; height 169 ± 10 cm). These patients underwent the following surgical procedures: coronary artery bypass grafts (14), combined coronary and valve operation (8), aortic (4), and mitral (2) valve replacement. At the time of the study, 18 patients were treated by a low-to-moderate dose of dobutamine (3–6µg·kg·min⁻¹) and noradrenaline (0.03–0.1 µg·kg·min⁻¹), and 10 patients required the addition of levosimendan (0.1–0.2 µg·kg·min⁻¹) as well. Twenty-six patients were in sinus rhythm and 2 in atrial fibrillation. Fluid challenges were done in 22 patients and the PLR maneuver in 6. These interventions were repeated in 19 patients so that a total of 47 pairs of measurements were analyzed.

The intrinsic precision of our reference technique (PACCO) was 2.8% ± 1.9 % at BL and 2.5% ± 1.5% at PM, and the LSC values were 3.8% at BL and 3.4% at PM. The intrinsic precision of NAPCO was 4.3% ± 3.3% at BL and 3.9% ± 3.5% at PM, and the LSC values were 5.9% at BL and 6.3% at PM. Because the LSC values were <10% for both techniques at any time of our study, we considered the intrinsic precisions of both techniques to be appropriate. The result of intercept BL was −1.03 (P = 0.10) and of intercept PM was −1.14 (P = 0.11), respectively. These results consistently showed that no systematic errors have been detected or no significant differences have been found between the 2 methods. The mean values, accuracy, 95% PI, precision, and percentage of error of the CO measured by both techniques are shown in Table 1.

Table 1: The Mean Values and Related 95% Confidence Intervals, Accuracy, 95% Prediction Intervals, Precision, and Percentage of Error (PE) of the CO Measured by the Pulmonary Artery Catheter (PACCO) and the Nexfin (NAPCO) at Baseline (BL), at After Preload-Modifying Maneuver (PM), and at Both Times Combined (ALL)

The correlation and Bland–Altman analysis of corresponding PACCO and NAPCO values are shown in Figure 1. The upper confidence limits of the 95% CI were 53%, 52%, and 48% for the calculated percentages of error: 39% at BL, 38% at PM, and 38% for ALL, respectively. The correlation coefficients (r) between the PACCO and NAPCO at BL, post-PMM, and for all measurements combined were 0.71 (95% CI, 0.53–0.82), 0.70 (95% CI, 0.52–0.82), and 0.72 (95% CI, 0.60–0.80), respectively (Fig. 1). The PMM induced similar increases in the PACCO and the NAPCO (0.66 ± 0.55 L·min⁻¹ with a 95% CI of 0.51–0.83 and 0.54 ± 0.59 L·min⁻¹ with a 95% CI of 0.38–0.72, respectively; P = 0.9). The bias between the absolute changes in NAPCO and PACCO after the PMM was −0.12 ± 0.25 L·min⁻¹. The correlation coefficients (r) of the absolute values and the percentage of change between the PACCO and the NAPCO were both 0.9 (P < 0.0001). The concordance rate between the changes (in percent) in PACCO and NAPCO was 100 % (95% CI was 80.5%–100%) as shown in the 4-quadrant scatter plot (Fig. 2).

Figure 1: Bland–Altman analysis (left) and correlation plots (right) showing correlation and agreement between the absolute values of the cardiac output (CO) measured by pulmonary intermittent thermodilution (PACCO) and by the Nexfin (NAPCO) at baseline, after a preload-modifying maneuver and for both periods combined. CO is expressed as L·min⁻¹.

Figure 2: A 4-quadrant scatter plot of the changes in cardiac output (CO) that were measured by pulmonary intermittent thermodilution (PACCO) and by the Nexfin (NAPCO) (n = 47). The concordance rate between changes in NAPCO and PACCO is 100% (95% confidence interval, 80.5%–100%) considering an exclusion zone of 15%.

The changes in the CO are also presented in a polar plot (Fig. 3), which demonstrates a polar angle of −1.2° indicating a small nonsignificant offset in calibration between the Nexfin and the PAC. The radial limits of agreement, which contain 95% of the data pairs of the changes in CO, are 22.4° to 24.9°, well below the 30° benchmark.^18,19

Figure 3: Polar plot showing radial limits of agreement (the dashed lines which contain 95% of the data pairs of the changes in cardiac output [CO]) of 22.4° and 24.9°, well below the 30° benchmark. The polar angle is −1.2° indicating a small nonsignificant offset in calibration between the Nexfin and the pulmonary artery catheter.

The area under the receiver operating characteristic curve that was created for testing the ability of the Nexfin device to detect an increase of ≥15% in the PACCO was 0.974 (95% CI, 0.93–0.99), with sensitivity and specificity of 94% and 90%, respectively (Fig. 4).

Figure 4: A receiver operating characteristic (ROC) curve testing the ability of the cardiac output (CO) measured by the Nexfin device to detect an increase of ≥15% in CO as measured by intermittent pulmonary thermodilution after a preload-modifying maneuver.

DISCUSSION

Our study demonstrates that in patients after cardiac surgery, the majority of which were receiving moderate doses of inotropic therapy and low-dose vasopressor therapy, there was a significant correlation between NAPCO and PACCO. Although the percentage of error between the 2 methods was close to 40%, higher than the traditionally recommended 30% limit,²⁰ the changes in the CO that were induced by a PMM were tracked reliably by the Nexfin monitor as evidenced by polar plot analysis and by a 100% concordance rate (95% CI was 80.5%–100%) with the simultaneous values measured by the PAC.

The measurement of CO is becoming increasingly prevalent in clinical practice.^1,2,10,20 This seems to be a result of several factors, one of which is the growing awareness that clinical assessment and conventional hemodynamic monitoring may not be enough to accurately assess the hemodynamic status and titrate fluids and inotropic therapy. Another major factor is the development of less-invasive technologies for the measurement of CO that have obviated the need for the use of the PAC for this purpose. Last but not least, the many studies that have shown that maximizing the CO in the perioperative period improves patient outcome have supplied the much needed evidence that the measurement of CO may indeed make a difference. Such evidence, as a rule, has always been very scarce in the field of hemodynamic monitoring.²¹ It is no wonder therefore that recent literature frequently recommends that minimally invasive CO monitoring for preload optimization should be considered in all major surgery.^3,4,22

The Nexfin is a recently introduced monitor that measures noninvasively and continuously both the arterial blood pressure and the CO by an inflatable finger cuff that serves as the only interface with the patient. The continuous beat-to-beat arterial blood pressure measured intraoperatively by the Nexfin has been recently shown to be closely correlated to invasive blood pressure monitoring over a wide range of pressure changes²³ and to have the potential to decrease the time of hypotension and hypertension during surgery compared with conventional intermittent blood pressure monitoring.²⁴ The CCO that is measured by the Nexfin has been validated against CO measured by the PAC,^11,25,26 PiCCO,²⁷ esophageal Doppler,²⁸ and transthoracic Doppler echocardiography.²⁹ All these studies reported significant correlations to the reference techniques, and percentage of error values of 29%,²⁵ 23%,²⁷ 37%,²⁸ and 39%,²⁹ comparable with the 38% of error found in our study.

Of note is 1 study²⁶ in which the concordance between NAPCO and PACCO was unaffected by the presence of atrial fibrillation. This may have been due to the fact that the Nexfin’s CO values were averaged over a period of 10 seconds.²⁶ In our study, we used 15-second averages for the Nexfin’s CO and have therefore included the 2 patients with atrial fibrillation in the final analysis.

In 2 of the aforementioned studies, hemodynamic interventions were performed to examine the ability of the Nexfin to track changes in CO.^27,28 Broch et al.,²⁷ who used a PLR maneuver to modify the preload status, found a good correlation between the changes in NAPCO and by the PiCCO, and concluded that the Nexfin is a reliable method of measuring CO during and after cardiac surgery. Chen et al.²⁸ administered a bolus of phenylephrine which induced similar changes in NAPCO and by the esophageal Doppler monitor with high concordance rates. Their results are of particular interest since this strong concordance was observed despite the intense vasoconstriction that was induced by phenylephrine. Other pulse contour methods were less able to accurately track changes in CO induced by vasopressors in both surgical^30–33 and septic patients.³⁴ We have not tested the performance of the Nexfin in the presence of acute changes in vascular tone. Our results, however, show that it can be reliably used to track preload-induced changes in CO in patients who receive a combination of vasopressor and inotropic therapy, similar to what was previously shown in a similar patient population.²⁵

The results of our study are relevant to the general debate concerning the optimal level of monitoring to be used in high-risk surgery. A recent survey among North American and European anesthesiologists revealed that, despite the mounting evidence about the benefits of CO optimization in high-risk surgery, only few practitioners do it on a routine basis.³⁵ One of the possible reasons that were mentioned for this discrepancy between current practice and evidence is the worry about the inaccuracies of the monitoring techniques.³⁶ Indeed, many of the studies that have examined the accuracy of CO monitors have produced results that do not meet the criteria set by Critchley.¹⁰ A meta-analysis of studies that compared the CO measured by pulse contour techniques, esophageal Doppler, partial carbon dioxide rebreathing, and transthoracic bioimpedance to thermodilution, showed that these methods were associated with a percentage error of approximately 40% to 45%.³⁷ A study of cardiac surgical patients found wide limits of agreement with percentage errors of 65%, 48%, and 54% for the esophageal Doppler, Vigileo/FloTrac, and LiDCOrapid monitors, respectively, compared with thermodilution.³² In another study in cardiac surgery patients,³³ the mean CO values measured by the FloTrac, PiCCO, and LidCO monitors were similar to intermittent thermodilution, although the 3 devices trended differently in response to therapeutic interventions, and showed concordance rates below the proposed 92% cutoff.⁸ In an accompanying editorial, Critchley⁷ claimed there is growing evidence that the pulse contour method is not the solution to providing reliable CO monitoring at the bedside.

However, the relevance in clinical practice of the criteria set by Critchley²⁰ has been repeatedly questioned.¹⁰ Most clinical studies have used the PAC as the reference technique against which new technologies were compared.^8,17 However, the thermodilution method has been shown to have a limits of agreement of about 40% when compared with CO measured by a periaortic transit-time flowprobe as the reference method,³⁸ and to be dependent on the selection of catheter and monitor model.³⁹ Thus, possible high precision errors of thermodilution CO may affect the calculation of the percentage of error and lead to values in excess of the 30% cutoff.¹⁷ This, in part, drove Peyton et al.³⁷ to suggest that a percentage error in agreement with thermodilution of ±45% is a more realistic expectation of achievable precision in clinical practice. Others have recently pointed out that all bedside methods for the assessment of CO are indeed estimates, and that, therefore, a measurement obtained by a less-invasive technique may be preferable if it can be obtained more rapidly and easily, even if it is slightly less accurate.^2,10 Nevertheless, due to the percentage of error of 39% that we found compared with the PAC, the Nexfin may not be a suitable monitor for those instances when the accuracy of the absolute values of CO is absolutely indicated.

The fundamental question therefore is whether a new device can replace thermodilution CO measurement as a guide to clinical decisions.^9,10 The continuity of measurement that is offered by new CCO monitors, provided that they do have a reliable tracking ability, makes these devices potentially superior to intermittent thermodilution since they can better assess the short-term effects of therapeutic and diagnostic interventions which are often performed in the process of perioperative goal-directed therapy. Our study shows that the Nexfin is able to reliably track changes in CO in response to changes in preload in stable cardiac surgery patients, many of whom were receiving low-moderate doses of vasoconstrictor and inotropic therapy. This ability, combined with its totally noninvasive nature, fast installation and ease of use, as attested by other groups as well,^25,27 make the Nexfin a suitable monitor for perioperative flow-based hemodynamic optimization. The ability of the Nexfin to track CO reliably during acute changes in vascular tone still needs to be established.

DISCLOSURES

Name: Serban Ion Bubenek-Turconi, MD.

Contribution: This author proposed and initiated the study design, led the study, oversaw data collection, and has contributed to data analysis and manuscript preparation.

Attestation: Serban Ion Bubenek-Turconi attests to the integrity of the original data and the analysis reported in this manuscript, reviewed the original study data and data analysis, and approved the final manuscript.

Conflicts of Interest: The author has no conflicts of interest to declare.

Name: Mihaela Craciun, PhD.

Contribution: This author participated in data collection, data analysis, and is the archival author.

Attestation: This author is the archival author.

Conflicts of Interest: The author has no conflicts of interest to declare.

Name: Ion Miclea, MD.

Contribution: This author participated in data collection and data analysis.

Attestation: This author reviewed the original study data and data analysis, and Ion Miclea also attests to the integrity of the original data and the analysis reported in this manuscript and approved the final manuscript.

Conflicts of Interest: The author has no conflicts of interest to declare.

Name: Azriel Perel, MD.

Contribution: This author helped design the study and prepare the manuscript.

Attestation: Azriel Perel approved the final manuscript.

Conflicts of Interest: This author serves as a consultant to BMEYE, Amsterdam, The Netherlands, and is a member of the medical advisory board of Pulsion, Munich, Germany.

This manuscript was handled by: Dwayne R. Westenskow, PhD.