Assessment of cardiac function after paediatric cardiac surgery relies on echocardiography, on the combination of indirect signs of systemic perfusion (e.g. diuresis, lactates, acid-base balance, and superior caval vein oxygen saturation), and on standard (e.g. blood pressure) and advanced (e.g. stroke volume) haemodynamic parameters to provide an accurate picture of the haemodynamic state of the patient. Reference Skowno1 However, this evaluation can be extremely challenging since, first, echocardiography requires an expert operator, is not continuous, and in the first post-operative hours is not conducted as a routine exam; furthermore, indirect signs of systemic perfusion can be altered by many conditions (e.g. diuresis can be reduced due to acute kidney injury, regardless of cardiac function), and continuous advanced haemodynamic monitoring in children is not commonly applied due to the absence of a gold standard method.
Recently, the pressure recording analytical method, a non-calibrated pulse contour method, has been applied to children after cardiac surgery. Reference Favia, Rizza and Garisto2–Reference Favia, Saxena and Steely5 Pressure recording analytical method is powered with the MostCareTM monitor (Vygon, Vytech, Padua, Italy). Pulse contour method application seems rational in children and neonates undergoing cardiac surgery since invasive blood pressure monitoring is part of the standard of care in all post-operative patients, and pressure recording analytical method estimates stroke volume from the analysis at a high sampling frequency (1000 Hz) of the arterial waveform. Reference Romagnoli, Franchi, Ricci, Scolletta and Payen6 Moreover, pressure recording analytical method does not require dedicated arterial catheters or calibration, and these characteristics make the application of this monitor particularly suitable in the paediatric setting. On the other hand, among all the signs of systemic perfusion routinely measured at bedside (i.e. blood pressure, heart rate, arterial blood gas analysis, lactates, superior caval vein oxygen saturation, and arterial pH), venous to arterial carbon dioxide difference is gaining attention as a promising marker of cardiac output adequacy. Reference Lamia, Monnet and Teboul7 Venous to arterial carbon dioxide difference represents the capability of venous blood flow (which equals cardiac output) to drag venous CO2 from the peripheral tissues to the right atrium. Reference Gavelli, Teboul and Monnet8 Essentially, the relationship between venous to arterial carbon dioxide difference and cardiac output reveals that when cardiac output is low, CO2 stagnates at the venous side, venous CO2 increases with respect to arterial CO2, and, ultimately, the CO2 gap (venous to arterial carbon dioxide difference) increases. Theoretically, for a given level of CO2 production, a decrease in cardiac output results in an increased venous to arterial carbon dioxide difference and vice versa. Reference Lamia, Monnet and Teboul7,Reference Gavelli, Teboul and Monnet8 Differently, respiratory failure, hypoxia, and microcirculatory injury with impairment of oxygen extraction may affect parameters such as superior caval vein oxygen saturation and arterial-venous oxygen difference, because oxygen delivery may be compromised even with normal cardiac output. Reference Lamia, Monnet and Teboul7,Reference Gavelli, Teboul and Monnet8
This study aimed to evaluate the correlation between venous to arterial carbon dioxide difference and cardiac index estimated by MostCareTM in neonates and children after cardiac surgery, upon admission to the paediatric cardiac intensive care unit and in the following days, as well as to assess venous to arterial carbon dioxide difference in relation to other systemic perfusion parameters. Finally, we evaluated whether venous to arterial carbon dioxide difference and other direct and indirect perfusion signs at paediatric cardiac intensive care unit admission were predictive of complex clinical course in our patients.
Materials and methods
We conducted a retrospective study collecting data from all patients younger than 18 years of age with biventricular congenital heart diseases who underwent elective cardiac surgery, with cardiopulmonary bypass, between June 2019 and January 2020. Inclusion criteria were also the presence of a superior caval vein central venous catheter, the availability of at least one pair of simultaneous (within 30 minutes) withdrawals for arterial and venous blood gas analyses after paediatric cardiac intensive care unit admission, sedation, controlled mechanical ventilation, and Most-CareTM monitoring data. Exclusion criteria were gestational age <37 weeks, intracardiac or extracardiac shunts, haemodynamic instability (i.e. bleeding with transfusion of more than 20 ml/kg of packed red blood cells and an increase of adrenaline and/or vasopressin of more than 25% of admission dose in the first 2 hours), post-operative pulmonary damage (i.e. PaO2/FIO2 ratio < 100), and application of extracorporeal membrane oxygenation at any time in the post-operative phase. Standard continuous post-operative monitoring included electrocardiogram, invasive (systolic, diastolic, mean) arterial blood pressure, peripheral oxygen saturation, rectal and peripheral temperature, cerebral near infrared spectroscopy (all recorded in the electronic medical chart every hour and retrieved at the same time point of all blood gas analyses pairs), arterial blood gas analyses, sampled every 2 hours during the first post-operative day and about every 4 hours thereafter, and venous blood gas analyses, routinely sampled twice a day, unless further requests were made. In our unit, MostCareTM is applied at the discretion of attending physician, typically in post-operative patients for the first 24–48 hours, with particular attention to those who are expected to require mechanical ventilation for more than 24 hours. When applied, MostCareTM parameters (i.e. cardiac index estimated by MostCareTM, systemic vascular resistances indexed, and dP/dtMAX) are recorded every hour. dP/dTMAX represents the measurement (not an estimation) of maximal pressure-to-time ratio or maximal slope of systolic upstroke. Previous studies have found dP/dtMAX to be associated with heart contractility Reference Saxena, Durward, Steeley, Murdoch and Tibby9 and ventriculo-arterial coupling. Reference Marinari, Rizza and Iacobelli10 Arterial waveform is always carefully checked and MostCareTM is not applied in the case of an overdamped or underdamped signal that cannot be adjusted or when dicrotic notch is not correctly detected. For the purposes of this analysis, we retrieved the following blood gas analyses data: arterial and venous CO2, arterial and venous oxygen saturation, arterial and venous pH, and arterial lactates. The following data were also collected: age and weight, cardiac diagnoses and surgical procedures, cardiopulmonary bypass duration, vasopressor doses (retrieved at the same time point of all blood gas analyses pairs) in order to calculate vasoactive inotropic score, Reference Gaies, Gurney and Yen11 and length of paediatric cardiac intensive care unit stay.
We defined the complex clinical course as the requirement for a vasoactive inotropic score above 15 at any time during paediatric cardiac intensive care unit admission or a paediatric cardiac intensive care unit length of stay above the median value of 5 days.
The primary objective was to assess the correlation between venous to arterial carbon dioxide difference and cardiac index estimated by MostCareTM recorded at the same time point (with a tolerance window of ±30 minutes). The secondary objectives were to analyse the association of venous to arterial carbon dioxide difference with other perfusion markers: superior caval vein oxygen saturation, arterio-venous oxygen gap, lactates, cerebral near infrared spectroscopy, vasoactive inotropic score, systolic arterial pressure, diastolic arterial pressure, heart rate, peripheral temperature, systemic vascular resistances indexed, dP/dtMAX, arterial pH, and venous pH. The behaviour over time of the main perfusion parameters (venous to arterial carbon dioxide difference, arterio-venous oxygen gap) was analysed. Finally, we evaluated whether direct and indirect perfusion signs recorded at paediatric cardiac intensive care unit admission were predictive of complex clinical course in our patients.
The institutional review board approved the protocol and waived the need for informed consent because of the retrospective design of the study.
Statistical analysis
Spearman’s correlation and logistic regression were used to analyse the univariate associations when variables were continuous or dichotomous, respectively. Mixed effect analysis was utilised to assess the behaviour of repeated measures over time, with a post-test assessment for trend. Multivariable logistic regression was utilised to assess which variable (reporting OR and 95% CI) was associated with the achieved definition of the complex clinical course. An area under the receiver operator characteristic curve of this model was calculated to explore its performance in predicting the complex clinical course. All data are presented as median (range). A p value of less than 0.05 was considered statistically significant. Statistical analysis was performed with the GraphPad Prism 8.0 software package (GraphPad Software, San Diego, CA).
Results
The flow chart of data collection is depicted in Figure 1. Analysed samples and all the collected clinical variables were retrieved between 0 and 5 days postoperatively. Ultimately, 127 venous to arterial carbon dioxide difference/cardiac index pairs were retrieved. Demographic data and baseline characteristics, including diagnoses, median (range) values of analysed parameters, and outcomes are reported in Table 1. All enrolled patients survived to paediatric cardiac intensive care unit discharge. The median venous to arterial carbon dioxide difference value was 9 and ranged from 1 to 25 mmHg. The correlation of venous to arterial carbon dioxide difference with cardiac index estimated by MostCareTM was not significant (r: −0.16, p = 0.08). It was generally weak with other clinical variables (Table 2) and significant, among other variables, with vasoactive inotropic score (r: 0.21, p = 0.02), systolic arterial pressure (r: −0.43, p = 0.0001), dP/dtMAX (r: 0.26, p = 0.004), and arterio-venous oxygen gap (r: 0.63, p = 0.0001). Cardiac index estimated by MostCareTM showed additional significant correlations with lactates (r: −0.29, p = 0.0009) and systemic vascular resistances indexed (r: −0.59, p < 0.0001) (Table 2).
Figure 1. Flow chart of data collection. BGA = blood gas analysis; CI-M = cardiac index estimated by MostCareTM.
Table 1. Demographic data, outcome, diagnosis, and perfusion variables. Data are expressed as absolute numbers or as median (range)
AV-O2 = arterio-venous oxygen difference; CI = cardiac index; Cnirs = cerebral near infrared spectroscopy; CPB = cardiopulmonary bypass; HR = heart rate; PCICU LOS = paediatric cardiac intensive care unit length of stay; SaO2/ScvO2 = arterial/venous oxygen saturation; SAP/DAP = systolic/diastolic arterial pressure; SVRI = systemic vascular resistances indexed; VIS = vasoactive inotropic score; VA-CO2 = veno-arterial CO2 difference
Table 2. Correlation between veno-arterial difference of CO2 (VA-CO2)(left side of the table), cardiac index estimated by MostcareTM (CI-M)(right side of the table) and other clinical variables
AV-O2 = arterio-venous oxygen difference; cNIRS = cerebral near infrared spectroscopy; CPB = cardiopulmonary bypass; HCO3 - = plasma bicarbonates; HR = heart rate; Lac = lactates; pH ven = venous pH; PIM3 = paediatric index of mortality 3; pvCO2 = CO2 venous partial pressure; pvO2 = oxygen venous partial pressure; SaO2/ScvO2 = arterial/venous oxygen saturation; SAP/DAP = systolic/diastolic arterial pressure; SVI = stroke volume index; SVRI = systemic vascular resistances indexed; VIS = vasoactive inotropic score
The levels of venous to arterial carbon dioxide difference showed a significant decreasing slope over time (−0.9, p = 0.01). Differently, arterio-venous oxygen gap slope was not significant (−0.2, p = 0.73) (Fig 2). Systolic arterial pressure (OR 0.95, 95% CI 0.90–0.99), dP/dtMAX (OR 0.00, 95% CI 0.00–0.06), lactates (OR 1.87, 95% CI 1.21–3.31), and venous to arterial carbon dioxide difference (OR 1.13, 95% CI 1.01–1.35) at paediatric cardiac intensive care unit admission showed a significant univariate association with the composite complex clinical course. None of these variables remained independently associated when multivariate logistic regression was built, even though this model produced an area under the receiver operator characteristic curve of 0.89 (p < 0.0001) (Fig 3).
Figure 2. Representation of analysed data over time. ( a ) veno-arterial CO2 difference (VA-CO2): significant trend slope; ( b ) arterio-venous oxygen gap (AV-O2): not significant trend slope.
Figure 3. ROC graph of the prediction model: Systolic arterial pressure (OR 0.95, 95% CI 0.90–0.99), dP/dtMAX (OR 0.00, 95% CI 0.00–0.06), lactates (OR 1.87, 95% CI 1.21–3.31) and VA-CO2 (OR 1.13, 95% CI 1.01–1.35) at paediatric cardiac intensive care unit admission were able to predict a prolonged paediatric cardiac intensive care unit stay or a vasoactive inotropic score >15 at any time.
Discussion
The key points of this study are 1) venous to arterial carbon dioxide difference did not correlate with cardiac index estimated by MostCareTM. This could be due to the fact that it is a poor marker of organ perfusion. Alternatively, we can speculate that correlating blood gas analyses samples with a continuous monitoring (cardiac index with MostCareTM) is wrong. In fact, it is possible that the changes in cardiac function are detected by these two variables at different time points. However, other perfusion variables appeared significantly associated with venous to arterial carbon dioxide difference, in particular arterial pressure and vasoactive inotropic score. It is possible that modifications of vasoactive drugs according to arterial pressure levels were effective to maintain a steady cardiac index, but some CO2 “stagnation” was still present. Lactate levels, one of the most important indirect perfusion markers in these patients, did not show a direct correlation with venous to arterial carbon dioxide difference, similar to the results of Akamatsu Reference Akamatsu, Inata, Tachibana, Hatachi and Takeuchi12 and Rhodes. Reference Rhodes, Erwin, Borasino, Cleveland and Alten13 2) On this line, based on trend analysis, we speculate that the progression of venous to arterial carbon dioxide difference from higher values to lower ones might express the “normalisation” of perfusion demands at the tissue level in the post cardiopulmonary bypass phase. Differently, arterio-venous oxygen gap did not appear to display a trend to normalisation. Venous oxygen saturation, as a surrogate variable of perfusion, may present significant pathophysiological limitations with respect to venous CO2, as explained by Gavelli and coworkers. Reference Gavelli, Teboul and Monnet8 3) If one might imply that cardiac index estimated by MostCareTM was not accurately estimated with the pressure recording analytical method, still dP/dTMAX could be considered an interesting parameter. Reference Marinari, Rizza and Iacobelli10 dP/dTMAX was associated with venous to arterial carbon dioxide difference and with the complex clinical course. Again, the association between cardiovascular function, vasoactive drugs administration, and systemic perfusion warrants further evaluation. 4) According to the results of this study, it would be difficult to propose cut off values for venous to arterial carbon dioxide difference; however, the median value of venous to arterial carbon dioxide difference of 9 in our patients could be used as a reference value between those with low and high venous to arterial carbon dioxide difference. It is also possible that the trend should be followed in these patients instead of an absolute value, with a reduction being a marker of an improving perfusion. 5) In particular, our data show a complex interaction between different variables. Possibly, the application of multimodal monitoring that included both direct and indirect perfusion parameters might improve prediction and management of low cardiac output in heart surgery children, as shown by the model’s promising area under the receiver operator characteristic curve. Our results, following previous controversial findings, may warrant further research on venous to arterial carbon dioxide difference variations over time and their relationship with other variables.
Limitations
This study enrolled a relatively small sample of patients; it is possible that correlations might have been stronger with a higher number of patients. However, our aim was to include only patients with corrected biventricular anatomies in order to exclude the confounders of intra- or extracardiac shunting and to analyse the most homogeneous population possible. MostCareTM is not a validated monitoring system in children and our primary objective may have been biased by an imprecise estimation of cardiac index. Furthermore, even if we routinely carefully check arterial waveform before applying this monitor, a formal control (i.e. with fast flush Gardner’s test Reference Romagnoli, Ricci and Quattrone14 ) was not performed and this is a major risk of inaccuracy when a pulse contour method is applied. However, there is currently no universally recommended monitor to measure cardiac index in children Reference Singh, Villaescusa and da Cruz15 and to apply as a gold standard. In our centre, MostCareTM has been repeatedly studied Reference Favia, Rizza and Garisto2–Reference Favia, Saxena and Steely5,Reference Marinari, Rizza and Iacobelli10 and we have found cardiac index estimated by MostCareTM to be reliable in paediatric cardiac patients after surgery. Reference Favia, Rizza and Garisto2 Finally, to add consistency, we analysed cardiac index estimated by MostCareTM associations with indirect perfusion parameters and found that it was also associated with lactate levels that are considered among the commonest variable utilised to estimate adequate oxygen delivery and patients outcome. Reference Singh, Villaescusa and da Cruz15,Reference Schumacher, Reichel and Vlasic16 Our study was retrospective, and the enrolled patients could not be controlled for diagnosis and severity of disease or incidence of low cardiac output. It has to be acknowledged that about 10% of recorded cardiac indices estimated by MostCareTM for this study were below 2 L/minute/m, Reference Favia, Rizza and Garisto2 which is the threshold for diagnosis of low cardiac output syndrome in our unit. Such percentage of low cardiac index estimated by MostCareTM should be adequate to assess correlation with venous to arterial carbon dioxide difference in an exploratory retrospective study. Finally, it might be interesting to include exclusively unstable patients, who were excluded in our study.
In conclusion, venous to arterial carbon dioxide difference did not correlate with cardiac index estimated by MostCareTM in our cohort of post-cardiosurgical children. Venous to arterial carbon dioxide difference showed a trend towards an improvement over time, and higher values were associated with the complex clinical course. Further evaluation of the exact clinical value of this parameter and its interaction with direct and indirect perfusion variables, especially over time, should be studied.
Acknowledgements
None.
Financial support
None
Conflict of interests
None
Ethical standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national guidelines on human experimentation (please name) and with the Helsinki Declaration of 1975, as revised in 2008, and has been approved by the institutional committees (Comitato Etico Bambino Gesù).
