Using haemodynamic parameters to monitor patients after coronary artery bypass surgery is an essential element of clinical practice. Such monitoring helps physicians properly manage patients after surgery, when haemodynamic parameters are usually dynamic and complicated. Pulmonary artery catheter-guided thermodilution (PACT) has been the gold standard for assessing the adequacy of cardiac output (CO), and its use after cardiac surgery is associated with a decreased length of stay and reduced cardiopulmonary morbidity.1 However, its invasive nature is associated with more complex procedures and the risk of complications, including infections. With the emergence of new non-invasive haemodynamic monitoring devices, such as continuous non-invasive arterial pressure (CNAP), alternative methods for haemodynamic monitoring are now available. Compared with invasive techniques, CNAP is more user-friendly, carries a lower risk of infection and is more cost-effective due to the reusable CNAP double-finger sensor.
To the best of our knowledge, no studies have been conducted to date in the Asian population evaluating the agreement between CNAP- and PACT-derived CO (CNAP-CO and PACT-CO, respectively). A study was performed in 2015 by Wagner et al. on 38 patients in a German medical intensive care unit (ICU) that compared transpulmonary thermodilution-derived CO (TDCO) obtained using the PiCCO system (Pulsion Medical Systems SE) with CO obtained using the CNAP system (continuous non-invasive CO [CNCO]; CNSystems Medizintechnik AG).2 Two Bland–Altman analyses were performed separately, one for CNCO calibrated to the first TDCO (CNCOcal) and another for CNCO autocalibrated to the biometric data (CNCOauto). The study showed an agreement between CNCOcal and TDCO, with an acceptable percentage error of 25%; however, there was no agreement between CNCOauto and TDCO, with a percentage error of 45%.2
In another study by Wagner et al. performed in 2017, CNCO was compared with an intermittent invasive CO measurement using PACT-CO in 51 post-cardiothoracic surgical patients.3 Using the same method as before, Wagner et al. performed two separate analyses: CNCO calibrated to biometric data (CNCObio) and CNCO calibrated to the first simultaneously measured PACT-CO (CNCOcal). In addition, the authors performed a subgroup analysis of patients after a passive leg raise manoeuvre and determined the trending ability of CNCO. The study concluded that there is good agreement between CNCOcal, but not CNCObio, and PACT-CO.3 However, there was good trending ability with a 100% concordance rate for both the CNCOcal and CNCObio.3
If CNAP is demonstrated to be at par with the gold standard PACT, it can be used as an alternative non-invasive method to measure haemodynamic parameters. Furthermore, it can be used for patients who have physiologic conditions that make PACT less accurate. These conditions include intracardiac shunts or severe tricuspid or pulmonary regurgitations. Continuous monitoring can detect changes in blood pressure and identify haemodynamic instability in real time, providing immense value. Determining and interpreting CO, along with other haemodynamic and metabolic parameters, can change the direction of therapeutic intervention. This will serve as a guide for physicians managing patients, and could equate to better overall prognosis and survival.
The objective of this study was to determine the agreement between CNAP and PACT in monitoring haemodynamic parameters. Furthermore, we explored whether there was a correlation between CNAP and PACT in tracking haemodynamic changes.
Methods
Participants
This prospective study was conducted at the cardiovascular surgical ICU of the Philippine Heart Center from October 2022 to February 2023. The inclusion criteria were adult patients aged ≥19 years scheduled for elective coronary artery bypass surgery with a pulmonary artery catheter and on thermodilution. Patients with mechanical circulatory support (intra-aortic balloon pumps [IABP], left ventricular assist device [LVAD], extracorporeal membrane oxygenation [ECMO]), arrhythmia, significant peripheral artery disease, residual shunts and significant postoperative tricuspid and pulmonary regurgitation were excluded from the study.
Patients who required mechanical circulatory support (IABP, LVAD, ECMP) in the postoperative period and those who had significant bleeding requiring exploration were withdrawn from the study.
Using consecutive sampling, 63 patients scheduled to undergo elective coronary artery bypass graft surgery meeting the study criteria were selected. Informed consent was obtained preoperatively. Immediately after surgery, patients were transferred to the surgical ICU. Information on demographic characteristics, risk factors, laboratory findings, preoperative risk and intra-operative characteristics were obtained through physical charts and electronic medical records.
As part of the standard of care at the surgical ICU, patients were hooked up to both non-invasive and invasive haemodynamic monitoring devices. Non-invasive devices include an ECG, pulse oximeter and non-invasive blood pressure monitoring, whereas invasive haemodynamic monitoring devices include an arterial line and a pulmonary artery catheter with a thermistor that is properly calibrated and transduced.
Pulmonary Artery Catheter Thermodilution Method
For PACT haemodynamic monitoring, a sensor-tipped pulmonary artery catheter or Swan-Ganz catheter (Arrow Thermodilution Catheter) was inserted during surgery via the internal jugular vein access and passed through the venous system into the right atrium, right ventricle and pulmonary artery.
Upon a patient’s arrival at the surgical ICU, standard protocols were followed, including proper calibration and levelling of the transducer to the right atrium and confirming zeroing the system to atmospheric pressure. The nurse in charge injected 10 ml normal saline solution at room temperature through the proximal port of the pulmonary artery catheter during end-expiration. A minimum of three measurements was obtained and any value that deviated >10% from the previous value was removed. Baseline haemodynamic parameters (CO, cardiac index [CI], stroke volume [SV] and systemic vascular resistance [SVR]) were obtained via PACT.
The basic principle of the PACT method is that a known amount of indicator (volume and temperature) is injected as a bolus into the circulation. The injectate then mixes with the blood and cools it. A thermistor located at the tip of the catheter measures blood temperature. The faster the blood flow, the faster the rise and decline of the indicator concentration downstream.4 The monitor can then compute the CO, as well as indirect measurements such as CI, SV and SVR.
Continuous Non-invasive Arterial Pressure
The CNAP devices used in this study were provided by the manufacturer (CNSystems Medizintechnik GmbH) and the local distributor (Philcare Pharma). A double-finger cuff was attached to the patient’s index and middle fingers, while another cuff was attached to the ipsilateral arm. We used the CNAP’s CNCO autocalibrated algorithm, which uses a polynomial function of the patient’s age, sex, height and weight. We did not use an external calibration mode (CO derived from PACT) because we wanted to completely eliminate the influence of PACT on the CNAP’s measurements so that we could demonstrate the latter’s utility. Using a plethysmography sensor, the change in blood volume in the finger was recorded. The cuff on the ipsilateral arm allows the maintenance of appropriate pressure in order to sustain a consistent blood volume in the finger. The pressure measured on the arm corresponds to the arterial blood pressure of every heartbeat, and the monitor shows a real-time, high-resolution, dynamic curve.
The CNAP uses a pulse contour analysis applied to the CNAP blood pressure waveform that allows measurement of the CO, CI, SV and SVR. These values are calculated from the continuously registered pressure curve, and the absolute values are adjusted by oscillometric measurement.5 Conditions that can cause significant changes to the blood pressure waveform, such as mechanical circulatory support (IABP, LVAD, ECMO), arrhythmia and significant peripheral artery disease, can alter the values of the haemodynamic parameters obtained by CNAP. Thus, patients with these conditions were excluded from the study.
Every time PACT was performed, a simultaneous measurement of haemodynamic parameters was obtained using the CNAP method. Three consecutive sets of measurements were recorded for each patient at 3-hour intervals: at the time of arrival in the ICU, then 3 hours, and 6 hours postoperatively. CO readings from the pulmonary artery catheter and CNAP were recorded using the data collection form.
Statistical Analysis
Descriptive statistics were used to summarise the demographic and clinical characteristics of the patients. Frequencies and proportions were used for categorical variables, whereas continuous variables are presented as the median and interquartile range (IQR) if they were not normally distributed and as the mean ( ± SD) if they were normally distributed. Paired-sample t-tests and Bland–Altman analysis were used to determine the difference and agreement, respectively, between CNAP and PACT in terms of CO estimation. Pearson product–moment correlation was used to determine the correlation between CNAP and PACT in terms of CO estimation. The Shapiro-Wilk test was used to test the normality of continuous variables. Missing values were neither replaced nor estimated. Null hypotheses were rejected at α=0.05. Stata 13.1 was used for data analysis.
Results
In all, 63 patients were included in the study; all patients were successfully enrolled with no withdrawals. Table 1 presents a summary of the clinicodemographic profiles of coronary artery bypass patients admitted to the Philippine Heart Center. The mean age of the patients was 62 years, most were male (n=55; 87.3%), 44.44% (n=28) had BMI values in the overweight category and 28.57% (n=18) had a depressed ejection fraction. With regard to surgical risk stratification, the median European System for Cardiac Operative Risk Evaluation (EuroScore) was 0.95% and the median Society of Thoracic Surgery (STS) score was 0.98%. In terms of comorbidities, 87.3% (n=55) of patients were hypertensive, 50.79% (n=32) were diabetic and 11.11% (n=7) had chronic kidney disease Stage III–IV. Most patients (n=56; 88.89%) had three-vessel disease, with 23.81% having left main involvement. In addition, 23.81% (n=7) of patients underwent concomitant valve surgery. The median bypass and ischaemic times (149 and 115 minutes, respectively) were both within acceptable ranges. None of the patients required high doses of vasopressors, defined as a dose of noradrenaline >1 µg/kg/min with or without the need for rescue therapy with vasopressin, adrenaline and dopamine at any time during the duration of the study.
Table 2 presents PACT- and CNAP-derived mean CO, CI, SV and SVR values at 0 (arrival in the ICU), 3 and 6 hours postoperatively. In general, CNAP-derived measurements for CO, CI and SV were slightly higher compared to the PACT measurements, while the SVR was slightly lower for the CNAP across all time points. Moreover, there was a significant difference between the values obtained using both methods except for the SVR values on the 0 and 3 h postoperatively.
Table 3 shows the agreement between PACT and CNAP. All hemodynamic parameters obtained using the PACT and CNAP showed a significant difference in all measured time points (0, 3 and 6 hours). The limits of agreement were broad, suggesting a considerable variability in the measurement between the two devices.
The correlation between PACT and CNAP is shown in Table 4. A moderately positive correlation between PACT and CNAP was observed for all haemodynamic parameters after surgery, suggesting that CNAP is capable of tracking haemodynamic changes.
Discussion
With the emergence of new finger-cuff technologies, continuous non-invasive assessment of haemodynamic parameters is now feasible. To date, there are only two commercially available technologies for automated continuous non-invasive assessment of CO, namely Clearsight (Edwards Lifesciences) and the CNAP system (CNSystems Medizintechnik AG). For this study, we used the CNAP system and compared it with the gold standard of PACT.
There was a high mean difference between CNAP- and PACT-derived haemodynamic parameters, suggesting that there was no agreement between the two methods. In addition, the results showed that the mean absolute difference between CNAP-CO and PACT-CO declined after the third hour. This could be due to temperature changes that can have a direct effect on the arterial pulse generated by the CNAP. Patients are initially hypothermic after bypass surgery (0-hour time point). This causes systemic vasoconstriction, resulting in a higher arterial pulse pressure and, subsequently, a higher blood pressure detected by the CNAP. This effect of hypothermia on the arterial pulse wave contour could result in a falsely high SV, CO and CI. However, with time, the mean absolute difference between CNAP-CO and PACT-CO decreased. At 3 and 6 hours postoperatively, patients eventually become normothermic, making the peripheral arterial pulse more congruent with the patient’s central pulse pressure. We can now thus expect the values of CNAP-derived CO, CI and SV to be closer to those obtained using PACT.
The results of our analysis showed that CNAP- and PACT-derived haemodynamic parameters were positively correlated. Thus, we conclude that the CNAP can be used for noting haemodynamic trends. By tracking real-time changes in CO, CNAP can quickly identify haemodynamic instability and monitor the effects of medications and fluids. For example, fluid responsiveness can be assessed by CNAP through continuous monitoring of changes in CO or SV after passive leg raising or administration of a fluid bolus. This will guide physicians in providing appropriate and timely interventions faster compared to a scenario involving frequent rechecking of PACT-CO or SV.
Our study had similar results with those of Wagner et. al. in 2015 and 2017, who showed that haemodynamic parameters derived by CNAP did not agree with those derived from PACT. In their 2017 study, Wagner et al. added a subgroup analysis of patients after a passive leg raise manoeuvre and determined the trending ability of CNAP. The concordance rate was 100% between the two methods, suggesting that giving CNAP is good for trending, as our results have demonstrated. Another paper published in 2022 by Roth et al. involved a small cohort of patients who underwent heart transplant and left ventricular assist device (LVAD) implantation. Through Bland–Altman analysis, they established a poor agreement between PACT and CNAP, concluding that the latter was not suitable for non-invasive evaluation of hemodynamic parameters.6
The findings of our study, as well as those of others, thus suggest that, although there is a high mean difference between the values obtained by both modalities, the CNAP’s utility lies in its positive correlation with PACT. By monitoring trends in its parameters, the CNAP can be used to determine fluid responsiveness and types of shock. There are also clinical scenarios where the CNAP may be beneficial due to its ease of use. Potentially deteriorating patients may be quickly hooked to the CNAP at the wards or emergency room before they are transferred to the ICU where a more advanced hemodynamic monitor may be used. If there are contraindications to putting in an arterial line, the CNAP may also prove a viable alternative. More studies are needed to explore these benefits.
Study Limitations
One limitation of this study is that the subjects were chosen from a pool of patients who were stable preoperatively because they were scheduled for an elective procedure. Furthermore, during the postoperative period, none of the patients required high vasopressor doses and none were withdrawn from the study due to the need for mechanical circulatory support. A larger sample size and a more heterogeneous subset of patients may address these issues in future studies. This may be done by involving other cohorts, such as patients undergoing general surgery or patients with different forms of shock. We also recommend involving other local medical institutions in future studies.
Conclusion
Our study demonstrated a high mean difference between CNAP- and PACT-derived haemodynamic parameters, suggesting that there was no agreement between the two methods. However, there was a positive correlation between the parameters derived by the two modalities suggesting a good trending ability of the CNAP. This means that although CNAP may not precisely approximate PACT measurements, it may still be used to track haemodynamic changes during the cardiac postoperative course. Therefore, CNAP may be utilised in the management of critically ill patients, depending on the setting and clinical scenario.
Clinical Perspective
- CNAP has the potential to be an alternative non-invasive method for monitoring haemodynamic parameters. It may also be useful for patients in whom PACT is less accurate, such as those with intracardiac shunts or severe tricuspid or pulmonary regurgitation.
- CNAP has several advantages, including its non-invasiveness, ease of use, and patient comfort. Additionally, the continuous monitoring allows for realtime detection of blood pressure changes and identification of haemodynamic instability.
- Although we found a high mean difference between CNAP- and PACT-derived haemodynamic parameters, suggesting no agreement between these two methods, we also found a positive correlation between CNAP and PACT. Thus, although CNAP may not precisely approximate PACT measurements, it may be used to track realtime changes in CO to guide appropriate and timely interventions.