Accuracy of a noninvasive estimated continuous cardiac output measurement under different respiratory conditions: a prospective observational study

The estimated continuous cardiac output (esCCO) system was recently developed as a noninvasive hemodynamic monitoring alternative to the thermodilution cardiac output (TDCO). However, the accuracy of continuous cardiac output measurements by the esCCO system compared to TDCO under different respiratory conditions remains unclear. This prospective study aimed to assess the clinical accuracy of the esCCO system by continuously measuring the esCCO and TDCO. Forty patients who had undergone cardiac surgery with a pulmonary artery catheter were enrolled. We compared the esCCO with TDCO from mechanical ventilation to spontaneous respiration through extubation. Patients undergoing cardiac pacing during esCCO measurement, those receiving treatment with an intra-aortic balloon pump, and those with measurement errors or missing data were excluded. In total, 23 patients were included. Agreement between the esCCO and TDCO measurements was evaluated using Bland–Altman analysis with a 20 min moving average of the esCCO. The paired esCCO and TDCO measurements (939 points before extubation and 1112 points after extubation) were compared. The respective bias and standard deviation (SD) values were 0.13 L/min and 0.60 L/min before extubation, and − 0.48 L/min and 0.78 L/min after extubation. There was a significant difference in bias before and after extubation (P < 0.001); the SD before and after extubation was not significant (P = 0.315). The percentage errors were 25.1% before extubation and 29.6% after extubation, which is the criterion for acceptance of a new technique. The accuracy of the esCCO system is clinically acceptable to that of TDCO under mechanical ventilation and spontaneous respiration.


Introduction
Cardiac output (CO) is an important hemodynamic variable for assessing cardiac function and for therapy guidance. Thermodilution CO (TDCO) measurement using a pulmonary artery catheter (PAC) is the gold standard CO measurement [1][2][3]. Hemodynamic monitoring using a PAC, including pulmonary artery pressure and mixed venous oxygen saturation measurement, has been used in patients with unstable hemodynamics in the intensive care unit (ICU) or during cardiac surgery [4,5]. Recently, the clinical benefits of a PAC have been questioned [6,7]. Moreover, PAC placement is an invasive procedure with a high risk of complications, such as arrhythmia, pulmonary artery perforation, pulmonary infarction, pulmonary embolism, and infection [8][9][10]. To avoid these complications, some studies recommend that a PAC should not be routinely used, even in critically ill patients or in cardiac surgery [11,12]. As an alternative to a PAC, noninvasive or minimally invasive hemodynamic monitoring techniques have been developed for continuous CO (CCO) monitoring [13]. The estimated continuous cardiac output (esCCO) system (Nihon Kohden Corporation, Tokyo, Japan) was recently developed based on the correlation between the pulse wave transit time (PWTT) and stroke volume [14]. The PWTT is defined as the time from the peak point of the R wave on the electrocardiogram (ECG) to the rise time of the pulse oximeter wave. The esCCO provides estimated CCO measurements using ECG findings, the pulse oximeter wave, and blood pressure (noninvasive or invasive). Therefore, the esCCO is a noninvasive system.
Previous studies have confirmed the accuracy of CO measurements between the esCCO and intermittent bolus thermodilution [14,15]. Ball et al. compared the CO of cardiac patients under general anesthesia in the operating room [14]. Yamada et al. compared their CO measurements in the ICU or operating room; however, they did not specify the patient status, especially the respiratory conditions [15]. Therefore, the accuracy of CCO measurements using the esCCO system compared to the TDCO under different respiratory conditions of mechanical ventilation and spontaneous respiration remains unclear. In particular, whether the esCCO system is affected by spontaneous respiration is unknown. Therefore, this prospective study aimed to assess the clinical accuracy of the esCCO system by continuously measuring the esCCO and TDCO from mechanical ventilation to spontaneous respiration through extubation after cardiac surgery with a PAC in the ICU. We hypothesized that the esCCO system could accurately measure the CCO under different respiratory conditions.

Ethics statements
This study was approved by the Nagoya University Hospital Ethics Committee (identifier number: 2021-0198) and registered with the UMIN Clinical Trials Registry (identifier: UMIN000044712). All patients provided written informed consent to participate in this study.

Study design and patients
This single-center prospective observational study included 40 patients who underwent cardiac surgery with a PAC and were admitted to our institutional ICU between December 2021 and May 2022. At our institution, a PAC is usually inserted during cardiac surgery with a cardiopulmonary bypass. Therefore, a PAC was not used for the purpose of this study, and we did not include criteria for the procedure in the methodology. We excluded patients undergoing cardiac pacing during esCCO measurement in the ICU (n = 12) and those who received treatment with an intra-aortic balloon pump (IABP) (n = 1) because pacemaker and circulatory support devices affect the PWTT, which might not be measured precisely [15]. We also excluded five patients with measurement errors or missing data, as shown in Fig. 1.

Test methods and measurement
Arterial catheter insertion into the radial artery and PAC insertion via the internal jugular vein were performed in the operating room. The arterial blood pressure and TDCO were continuously measured in the operating room and ICU. At the beginning of monitoring in the ICU, hemodynamic monitoring (HemoSphere Advanced Monitoring Platform; Edwards Lifesciences Corporation, Irvine, CA, USA) was calibrated to measure the TDCO based on the age, height, weight, and sex obtained from patients' electronic medical records. A bedside patient monitor (BSM-6501; Nihon Kohden, Tokyo, Japan) was also calibrated to measure the esCCO based on the TDCO and PWTT calculated using ECG and pulse oximeter waves. ECG monitoring was performed using lead II, and a disposable oxygen saturation sensor (TL-281T-IB; Nihon Kohden) was placed on the patient's fingertip. The esCCO and TDCO measurements were performed in all cases after calibration. Continuous TDCO and esCCO datasets were obtained every 5 min during time periods of 4 h before and after extubation, except for 1 h before and after extubation (Fig. 2).

Management of respiratory, analgesia, and sedation status
All patients received assist control ventilation with pressure control upon admission to the ICU. The ventilator mode was changed from assist control to pressure support ventilation for the spontaneous breathing trial. Decisions regarding the mechanical ventilator setting (peak inspiratory pressure, pressure support level, positive end-expiratory pressure, and Fig. 1 Flowchart of the selection of the study population. The inclusion criteria were patients who had undergone cardiac surgery with pulmonary arterial catheter insertion and admitted to our institutional intensive care unit between December 2021 and May 2022. IABP intra-aortic balloon pump respiratory rate) and when to wean patients from mechanical ventilation were at the discretion of individual anesthesiologists. Therefore, the ventilator settings during esCCO and TDCO measurements were not specified. A spontaneous awakening trial was performed in all patients to target the Richmond Agitation-Sedation Scale score of − 1 to 0 adjusted doses of dexmedetomidine, propofol, and fentanyl. Propofol was discontinued in all cases until extubation. Then, a spontaneous breathing trial was performed for at least 30 min in all patients. The patients were extubated following the successful spontaneous awakening and spontaneous breathing trials. After extubation, all patients received venturi mask oxygen therapy. For the analgesia and sedation status after extubation, individual anesthesiologists adjusted the dexmedetomidine and fentanyl with a target Richmond Agitation-Sedation Scale score of 0 to − 1.

Outcome measurements
For outcome measurements, we assessed the clinical accuracy of the esCCO in various clinical settings. Therefore, we compared the esCCO with TDCO from mechanical ventilation to spontaneous respiration through extubation.
The esCCO is calculated from 64 heartbeat-averaged PWTTs based on data retrieved within every 1 s interval [16]. Therefore, the esCCO is considered as real-time hemodynamic monitoring, including a 32-heartbeat delay as almost half of 64 heartbeats [17]. In contrast, a characteristic of a continuous TDCO is the presence of a delayed time response derived from the averaging process [18]. Continuous TDCO measurement requires 7-8 min for deliberation as the CO measurements are calculated, and it can be delayed by up to 15 min [19]. Therefore, the esCCO and TDCO measurements were simultaneously compared with and without time adjustment. We adjusted for time by averaging the esCCO data over the last 20 min (20 min moving average esCCO measurement [Ave20]).

Statistical analysis
A sample size of 930 points, paired esCCO and TDCO measurements, was calculated to ensure a bias of 0.13 L/min and a standard deviation (SD) of differences of 1.15 L/min with 80% power and a significance level of 0.05, based on methods in previous studies [15]. Assuming that 48 points of CO measurements could be extracted before and after extubation in one patient, 20 patients were required for this study. We expected a 50% attrition rate due to postoperative pacing or arrhythmia; therefore, we included 40 participants in this study.
The baseline characteristics of the patients are expressed as a numerical value (proportion), mean ± SD, or median [interquartile range]. Agreement between the esCCO and TDCO was evaluated using Bland-Altman analysis [20,21]. This method provides bias, a percentage error ([2 SD of the bias]/[mean of the reference method]), and 95% limits of agreement. Critchley and Critchley reported that an acceptable level of precision requires a percentage error of less than 30% and that the acceptance of a new technique should rely on limits of agreement of up to approximately 30% [22]. The correlation between the esCCO and TDCO was analyzed using the Spearman rank correlation coefficient. The bias and SD were compared using the Student's t test. Statistical significance was set at P < 0.05. All statistical analyses were performed using R software, version 4.2.0 (The R Foundation for Statistical Computing, Vienna, Austria).

Results
We screened 40 patients for this study. Based on the exclusion criteria, such as cardiac pacing and IABP, we included 23 patients in the analysis. Postoperative atrial fibrillation was not observed in our study's cases during esCCO measurements before and after extubation. The patient characteristics are shown in Table 1.

Comparison of esCCO and TDCO measurements without time adjustment
In 23 cases, we compared esCCO and TDCO measurements between 956 points before extubation and 1072 points after extubation. The correlation coefficients between the esCCO and TDCO measurements were r = 0.851 before extubation and r = 0.803 after extubation (Fig. 3a, b). Using Bland-Altman analysis, we found that the bias, SD, and 95% limits of agreement values were 0.13 L/min, 0.62 L/min, and − 1.11 to 1.38 L/min before extubation (Fig. 3c), and − 0.46 L/min, 0.82 L/min, and − 2.10 to 1.17 L/min after extubation (Fig. 3d), respectively. There was a significant difference in bias before and after extubation (P < 0.001) but not in the SD (P = 0.639). The percentage errors were 26.2% and 31.1% before and after extubation, respectively. The mean TDCO values were 4.8 ± 1.0 L/min before extubation and 5.3 ± 1.0 L/ min after extubation, and the mean esCCO values were 4.9 ± 1.2 L/min before extubation and 4.8 ± 1.4 L/min after extubation (Table 2a).

Comparison of esCCO and TDCO measurements with time adjustment
We compared esCCO and TDCO measurements using Ave20 as the esCCO measurement. The esCCO and TDCO measurements from 939 points before extubation and 1112 points after extubation in 23 patients were compared. The correlation coefficients between the esCCO and TDCO were r = 0.859 before extubation and r = 0.818 after extubation (Fig. 4a, b). Using Bland-Altman analysis, we found that the bias, SD, and 95% limits of agreement values were 0.13 L/min, 0.60 L/ min, and − 1.06 to 1.33 L/min before extubation (Fig. 4c), and − 0.48 L/min, 0.78 L/min, and − 2.04 to 1.07 L/min after extubation (Fig. 4d), respectively. There was a significant difference in bias before and after extubation (P < 0.001); however, in the SD, the difference between before and after extubation was not significant (P = 0.315). The percentage errors were 25.1% and 29.6% before and after extubation, respectively. The mean TDCO values were 4.8 ± 1.0 L/min before extubation and 5.3 ± 1.0 L/min after extubation, and the mean esCCO values were 4.9 ± 1.2 L/min before extubation and 4.8 ± 1.3 L/ min after extubation (Table 2a). Herein, the esCCO measurement was acceptable as an alternative to the TDCO measurement under mechanical ventilation, resulting in a percentage error of 25.1% with time adjustment by Ave20. Additionally, the esCCO measurement was acceptable under spontaneous respiration, resulting in a percentage error of 29.6% with time adjustment by Ave20. Based on a previous report [22], our percentage error findings are clinically acceptable; however, the percentage error for the esCCO measurements was smaller under mechanical ventilation than under spontaneous respiration. The difference in the SD between the esCCO and TDCO measurements was acceptable both before and after extubation. However, this study showed that the bias after extubation was significantly greater than that before extubation. The reason for the significant difference in bias was related to the change in mean CO before and after extubation. The mean TDCO gradually increased before and after extubation, whereas the esCCO did not change under different respiratory conditions, either with or without time adjustment, as shown in Table 2b. Kirkeby-Garstad et al. reported that the clinical precision of TDCO measurements is significantly reduced during spontaneous respiration compared with mechanical ventilation [23]. They showed that differences in intrathoracic pressure changes between mechanical ventilation and spontaneous respiration and irregular respiratory rate patterns in spontaneous respiration might reduce the accuracy of the TDCO measurement. They also indicated that spontaneous respiration induces more fluctuations in pulmonary artery blood temperature than mechanical ventilation, affecting the TDCO measurement's accuracy. In this study, the accuracy of TDCO measurements under spontaneous respiration may have influenced the bias between the esCCO and TDCO. In contrast, the reduction in the intrathoracic pressure resulting from extubation reduces the duration of the pre-ejection period (PEP), which is Table 2 Comparison of cardiac output before and after extubation (a) The esCCO and TDCO measurements were compared with and without time adjustment using the 20 min moving average esCCO measurement (Ave20). The mean cardiac output was calculated as the average of every 5 min measurement point for 4 h before and after extubation. (b) Comparison of hourly cardiac output measurements between the esCCO and TDCO from 4 h before extubation to 4 h after extubation using the 20 min moving average esCCO measurement (Ave20 the period just before the blood is pumped into the aorta [24]. Since the PWTT includes the PEP, we speculate that a decrease in the PEP leads to a decrease in the PWTT. However, the physiological changes in the PWTT in response to intrathoracic pressure change and other factors caused by extubation remain unclear. Several factors might influence the PWTT under spontaneous respiration, such as intrathoracic pressure changes, irregular respiratory rate patterns, and body movement, which could cause disturbances in the ECG findings. Therefore, further studies are needed to investigate the accuracy of esCCO and TDCO measurements under spontaneous respiration in combination with other techniques, such as echocardiography. Our study showed that the percentage errors between the esCCO and TDCO from mechanical ventilation to spontaneous respiration were 25.1-29.6% with time adjustment using Ave20. This result was lower than the 44-60% reported in previous studies [14,15]. Additionally, in the Bland-Altman analysis in our study, the bias and SD between the esCCO and TDCO were − 0.48 to 0.13 L/min and 0.60-0.78 L/min, respectively, even for continuous measurements under different respiratory conditions, which were lower than those in a previous study: 0.66-0.95 L/min and 1.16-1.59 L/min [14]. Our study used data from the ICU, where hemodynamics were stable despite changes in the respiratory condition due to extubation, whereas the previous report used intraoperative data such as hemodynamics that are prone to fluctuations [14]. Sugo et al. reported that the PWTT through the peripheral arteries was incorporated into the calculations in the present method to mitigate the effect of systemic vascular resistance [25]. However, the accuracy of the esCCO may be lower than that of the TDCO in clinical situations with large CO variations. In addition, the possible weaknesses of the PWTT that may cause inaccurate esCCO measurements to include an unstable pulse wave owing to lower perfusion and too many arrhythmias, such as atrial fibrillation. Postoperative atrial fibrillation was not observed in our study's cases. We consider this finding to be due to the timing of the esCCO measurements in the early postoperative period, within 24 h of surgery. Although the TDCO is the gold standard for CO measurement in any clinical setting, the percentage error under spontaneous respiration after extubation in this study was acceptable for a clinical index. Therefore, this study suggests that the esCCO could be used as a noninvasive hemodynamic monitoring technique for CCO monitoring in patients under spontaneous respiration conditions in general hospital beds and ICUs. Currently, there is no equipment for continuously monitoring CO in general hospital beds. The esCCO device was safe and simple to use. Thus, noninvasive hemodynamic monitors such as the esCCO system are more likely to be used than invasive hemodynamic monitors such as a PAC for patients in general hospital beds. This study has several limitations. First, this study only included patients who underwent cardiac surgery with a PAC and were admitted to our institutional ICU, and only ICU periods were used to collect the data. Further studies under various clinical settings are required to improve the generalizability of the findings of this study. Second, this study did not specify the ventilation pressure setting (peak inspiratory and positive end-expiratory pressures) during esCCO measurement, which may have affected the accuracy of the measurements. However, the percentage errors of esCCO were clinically acceptable compared to the TDCO, even at various ventilator settings, reflecting real-world data. Third, we excluded patients undergoing cardiac pacing during esCCO measurement in the ICU because the PWTT might not be precisely measured [15]. Therefore, it is necessary to consider other appropriate hemodynamic monitors for these patients as an alternative to the TDCO with a PAC. Fourth, this study did not measure the PEP, a component of PWTT, to clarify the physiological changes in PWTT resulting from extubation. Further studies should measure changes to various respiratory and circulatory parameters caused by intrathoracic pressure and other factors related to spontaneous respiration.
In conclusion, this study suggests that the accuracy of the noninvasive esCCO measurement is clinically acceptable as an alternative to TDCO measurement under different respiratory conditions of mechanical ventilation and spontaneous respiration.