Maintenance of adequate global oxygenation to prevent postoperative complications is essential in infants and children undergoing major surgery. Monitoring adequacy of global oxygen delivery in paediatric anaesthesia and intensive care is usually limited to the periodical assessment of acid–base balance, lactate concentration and central venous oxygen saturation (ScvO2). Base excess and lactate concentration, however, are retrospective assessments. ScvO2 requires invasive central venous catheterization and repeated blood sampling with associated risks1,2 and so provides intermittent rather than continuous data.
The PediaSat system (Edwards Lifesciences, Irvine, USA) is a newly developed central venous catheterization system for continuous ScvO2 measurement based on fibreoptic technology (SPediaSatcvO2). So far only limited laboratory and clinical data1,3,4 are available. The aim of this study was to validate the PediaSat system under clinical conditions by testing the accuracy of the fibreoptic measurements compared with values derived from the central venous blood samples routinely taken for laboratory analysis in paediatric patients scheduled for major noncardiac and cardiac surgery.
The study was approved by the Hospital Ethics Committee and written informed consent was obtained from each parent or patient as appropriate. We included children and adolescents undergoing general anaesthesia with central venous catheterization for major orthopaedic, craniofacial or cardiac surgery with and without cardiopulmonary bypass (CPB), with expected blood loss and/or haemodynamic changes.
The PediaSat system (Edwards Lifesciences) used in this study consisted of a central venous catheter with a built-in spectrophotometric probe containing two fibreoptic lines. The tip of this probe protrudes from the distal end of the catheter by 0.1 cm and the proximal part is connected to the optical module of the PediaSat system. For the purpose of monitoring SPediaSatcvO2, light of two wavelengths (660 and 800 nm) is transmitted through one of the two fibreoptic lines of the central venous catheter placed in the superior vena cava. Tissue chromophores, such as haemoglobin (Hb), absorb near-infrared light depending on their oxygenation state. Changes in chromophore concentrations and oxygenation states, revealed by comparing emitted and detected near-infrared light, can be quantified using the modified Lambert–Beer law.5–7 Light reflected from the Hb in the red blood cells is detected by the second fibreoptic line of the probe and transmitted back to the sensor in the optical module of the PediaSat device. This determines light extinction and SPediaSatcvO2 is calculated and displayed on the screen of the Vigileo Monitor (Vigileo Monitor MHM1E; Edwards Lifesciences). When in use, the detection quality of the fibreoptic probe is continuously monitored by the Sensitivity Quality Index (SQI), which is graded in values of 0–5. SQI values less than or equal to 2 indicate adequate sensitivity of the probe.
General anaesthesia was induced by the inhalational or intravenous route depending upon the patient's medical conditions and preference. After establishing neuromuscular blockade using atracurium, the trachea was intubated and the lungs mechanically ventilated. Taking into account the patient's size and the type of surgical procedure, a 4.5 Fr two-lumen or 5.5 Fr three-lumen central venous catheter was inserted either by the subclavian or internal jugular venous route until the tip lay in the superior vena cava. When steady-state values for continuously measured ScvO2 were obtained, the PediaSat system (SPediaSatcvO2) Vigileo device was calibrated in vivo in accordance with the manufacturer's instructions. Only SQI values of 2 or less were accepted. If SQI was higher than 2, the position of the optical probe was corrected until the desired SQI values were reached, or the probe was removed and the patient excluded. Hb value was updated on the Vigileo device when there was a significant change (>1.8 g dl−1), according to the user manual.
Measurements were performed after reaching steady-state conditions. Blood samples were drawn from the distal lumen of the central venous catheter for assessment of ScvO2 (SCO-OXcvO2) and haemoglobin (Hb) concentration by co-oximetry at the following time points: in noncardiac and cardiac surgery without CPB approximately every 60 min; and in cardiac surgery using CPB, sampling took place approximately 30 min after induction of anaesthesia but before CPB, every 60 min after CPB began and 60 min after CPB. Blood samples were analysed using multiwavelength haemoximetry (GEMOPL; Instrumentation Laboratory, Lexington, Massachusetts, USA). Pulse rate, mean arterial pressure (MAP), central venous pressure (CVP) and body temperature were recorded. Patients' characteristics (age, sex, weight, height, BMI and body surface area) were noted.
A power calculation revealed that a sample size of 20 patients would have 80% power to detect a difference of at least 1 SD between SPediaSatcvO2 and SCO-OXcvO2 using a paired t-test with a 0.05 two-sided significance level. Data are expressed as mean (+SD) for continuous parametric variables or median (range) for nonparametric variables. Agreement between SPediSatcvO2 and SCO-OXcvO2 was assessed by Bland–Altman analysis.8 Linear regression analysis was performed to compare SPediSatcvO2 and SCO-OXcvO2 and the SPediSatcvO2 and SCO-OXcvO2 with SCO-OXcvO2 differences. To assess the independent influence of age, body temperature, pulse rate, MAP and CVP, multiple regression analysis was used. Sensitivity/specificity of ΔSPediSatO2 between two consecutive readings to indicate a fall or increase in SCO-OXO2 was calculated. SPSS version 16.1 (SPSS Inc., Chicago, USA) was used from the hospital resources for this purpose.
Of 28 children and adolescents aged from 0.6 to 19.0 years scheduled for major surgery, 26 were included. One child had to be excluded because of detection failure of the PediaSat system and another because the surgery was prematurely terminated when only one measurement had been performed. There were no catheter-related complications. Epidemiological data are expressed in Table 1.
In total, 142 paired simultaneous measurements of SPediaSatcvO2 and SCO-OXcvO2 were obtained [4–10 per patient (median 6)]. SPediaSatcvO2 values ranged from 57 to 98% (median 80%) and SCO-OXcvO2 values ranged from 57.1 to 95.8% (median 78.7%). Hb values ranged from 5.3 to 10.6 g dl−1 (median 9.6 g dl−1). Interparticipant correlation between SPediaSatcvO2 and SCO-OXcvO2 values was poor with r2 equal to 0.28 (P < 0.0001; Fig. 1). Overall, SPediSatcvO2 slightly overestimated SCO-OXcvO2 (mean bias +2.6%), but limits of agreement (LOA: ±2 SD of bias) were unacceptably high with −14.4 and +19.6% (Fig. 2). Bland–Altman analysis of SPediaSatcvO2 and SCO-OXcvO2 for repeated measurements resulted in a mean bias of 6.2% with LOA between – 6.3 and +18.7%. Sensitivity and specificity of SPediaSatcvO2 to indicate a fall or rise of SCO-OXcvO2 between two subsequent measurements were only 0.42 and 0.24, respectively. For significant changes of SCO-OXcvO2 (ΔSCO-OXcvO2 >5%), the sensitivity did not change, but specificity decreased to 0.18 (Fig. 3).
For SPediaSatcvO2 and SCO-OXcvO2 values in all patients at the first time point of blood sampling following anaesthesia, the mean bias and LOA were 0.2%, and −5.9 and +6.3%, respectively. Similar agreement was found in the subgroup analysis of SPediaSatcvO2 and SCO-OXcvO2 sampled in cardiac surgery patients before starting the CPB with a mean bias of −1.9% and LOA of −4.6 and +0.8% (Table 2). The data of noncardiac surgery as well as cardiac surgery patients with and without CPB showed only a moderate agreement. Subgroup analysis is given in Table 2. Multiple regression analyses revealed that the mean bias of SPediaSatcvO2 and SCO-OXcvO2 was not influenced by body temperature, pulse rate, MAP and CVP, but age independently influenced the mean bias (P = 0.22). Subgroup analysis of patients less than and greater than 10 years showed a mean bias of +4.2 and +0.3%, and LOA of −24.7/+23.1% and −12.5/+13.1%, respectively.
The present study examined the extent to which central venous oxygen saturation and its changes can be reliably measured with the PediaSat system in children and adolescents undergoing major surgery. The main finding was that central venous oxygen saturation was not reliably estimated, nor were significant changes (ΔScvO2 >5%) reliably detected by the PediaSat system. Agreement between SPediaSatcvO2 and SCO-OXcvO2 was only acceptable for the first sample size drawn after anaesthesia had been completed.
The clinical application of ScvO2 as a surrogate for tissue oxygenation has generated interest in recent years.5 Continuous measurement of ScvO2 is, therefore, of high interest in critically ill patients, in patients undergoing major surgery and particularly in neonates and infants undergoing major noncardiac and cardiac surgery. Several clinical and laboratory studies have compared oxygen saturation values by co-oximetry.9–11 Scuderi et al.10 compared three pulmonary artery oximetry catheters and reported acceptable agreement with mixed venous oxygen saturation measured by co-oximetry (95% confidence limits: 2.47 and 3.36%, respectively). Burchell et al.12 interpreted a mean bias of −0.57% and LOA of ±3.76% as acceptable.
The PediaSat system investigated in this trial is suitable for neonates, infants and children, as three different sizes of the oximetry catheters are available. It is easy to handle, does not require any additional central venous access and allows in-vivo calibration. Compared with standard central venous catheters, its catheters are stiffer with an increased outer diameter. The necessity of adjusting to significant changes in Hb may also be considered a disadvantage.
Recently, Ranucci et al.3 reported excellent correlation and good agreement of continuously measured ScvO2 values using the PediaSat system with those measured by co-oximetry in 30 neonates and paediatric patients undergoing cardiac surgery. Each simultaneous measurement per patient was performed before, during and after the installation of CPB. The best agreement with a mean bias of zero and a precision of ±5.84% was found before starting the CPB, and the agreement of the data pairs collected during and after CPB worsened only slightly. Similar results were found by Liakopoulos et al.13 in five pigs and in 16 paediatric patients undergoing cardiac surgery. They reported an excellent agreement with a mean bias of nearly 0% and precision of ±4.4% in both. Additionally, in this study, the fibreoptically detected ScvO2 correlated better with changes in cardiac index compared with routine haemodynamic variables. They concluded that the PediaSat system provided accurate continuous ScvO2 monitoring.
In contrast with the above, the current investigation found the interparticipant correlation between SPediaSatcvO2 and SCO-OXcvO2 values to be poor, with r2 equal to 0.28 and an unacceptable agreement. Using the Bland–Altman analysis for repeated measurements, SPediSatcvO2 overestimated SCO-OXcvO2 and the LOA were still high. In only four (two noncardiac and two cardiac) of the 26 included patients was acceptable agreement between SPediaSatcvO2 and SCO-OXcvO2 found (data not shown). Subgroup analysis of SPediaSatcvO2 and SCO-OXcvO2 values sampled following anaesthesia for all patients, and those sampled in the subgroup of cardiac surgery patients before installation of the CPB showed good agreement. These findings may be explained by good detection conditions during long-term haemodynamic stability before consecutive surgical or mechanical disturbances. Interestingly, the SPediaSatcvO2 and SCO-OXcvO2 values ranged from 69.6 to 88.0%, which approximates to the calibration values. This is in accordance with Huber et al.,6 who reported best agreement of fibreoptically detected SO2 close to the calibration point of 70% in an in-vitro study. In all other subgroups, such as noncardiac surgery, cardiac surgery without CPB, during CPB and at the end of CPB, the agreement was only moderate. One reason might be brief consecutive surgical, mechanical or haemodynamic disturbances, which would require repeated calibration of the PediaSat system or possible correction of the catheter position. In common with Ranucci et al.,3 the current investigation found signal disturbances (SQI >2) occurred in the cardiac surgery with CPB group considerably more frequently than in the other subgroups, particularly during surgical manipulation of the superior vena cava and during CPB, mostly caused by juxtaposition of the tip of the catheter to the vessel wall. Under these conditions, even in the absence of a poor SQI measurement, bias cannot be excluded. Another explanation for the poorest agreement between SPediaSatcvO2 and SCO-OXcvO2 during CPB may be the interference of the ambient light of the operating theatre with the fibreoptic probe.
Furthermore, the slope of the agreement between SPediaSatcvO2 and SCO-OXcvO2 suggests near linear dependency of SCO-OXcvO2 with overestimation in the low range and underestimation in the high range (Fig. 2). This near linear dependency was also found in all subgroup analyses (data not shown). These findings are in agreement with the results of several clinical and laboratory studies5,6,14,15 that have investigated the accuracy and precision of fibreoptic-measured ScvO2 of various devices. Most recently, in an in-vitro trial of the PediaSat system,4 we reported only an acceptable agreement of SPediaSatcvO2 and SCO-OXcvO2 values with a considerable overestimation of SCO-OXcvO2 values below 70%. The near linear dependency of the mean bias of SPediaSatcvO2 and SCO-OXcvO2 possibly indicates a systematic error in the PediaSat calibration algorithm. But this hypothesis is yet unproven. Surprisingly, whereas the sensitivity and specificity of SPediaSatcvO2 and SCO-OXcvO2 in our previous laboratory investigation were excellent, in the current trial, sensitivity and specificity of SPediSatcvO2 for significant changes of SCO-OXcvO2 (ΔSCO-OXcvO2 >5%) were only poor. For these results, we currently have no adequate explanation. Spenceley et al.1 evaluated the PediaSat system in 19 children after cardiac surgery and reported a good correlation between SPediaSatcvO2 and SCO-OXcvO2 values with r2 of 0.64, but the agreement was only moderate with a mean bias of +1.1% and precision of 16.9%. Their results led to the recommendation that the PediaSat system should only be used to provide an accurate trend of continuous ScvO2 within the physiological range. An imprecision of up to 5% may be considered as acceptable at SvO2 values of 60% in situations in which therapeutic interventions are needed.
The age of the patients significantly influenced the mean bias of SPediaSatcvO2 and SCO-OXcvO2 and data analysis of patients less than and greater than 10 years showed a change in the agreement. Bias was nearly 0 and the LOA halved compared with results in patients less than 10 years. A possible explanation for this finding could be the smaller vessels in smaller children with interference of reflected light from the vessel walls. Another reason could be that the blood samples for co-oximetry measurement were drawn from afferent small veins in children less than 10 years and, thus, faulty oxygen saturation values were determined. Nevertheless, the agreement of SPediaSatcvO2 and SCO-OXcvO2 in patients greater than 10 years was only moderate.
Our study has some limitations. In contrast to the few previously published studies investigating the PediaSat system, the patient population included in this trial was heterogeneous with a majority undergoing cardiac surgery. Hence, varying numbers of data pairs per patient were collected. However, the precision of fibreoptically measured ScvO2 should not be dependent on or restricted to a particular patient population. Second, the catheter position was not controlled by transoesophageal echocardiography (TEE) and, hence, borderline catheter malposition could not be detected. Errors may also have occurred in both intermittent blood and continuous gas analysis, affecting the agreement. Furthermore, the catheter values may also drift,6 requiring more frequent repeated recalibration manoeuvres. Finally, simple human errors in the sampling method cannot be fully excluded and might be responsible for outliers in the data. Preanalytical errors in whole-blood sampling occur and typically are caused by air bubbles in the sample, inhomogeneous samples, haemolysis, improper storage of the samples, clots, dilution from physiological saline in arterial lines and uneven ventilation.16 However, the results of this investigation suggest serious technical and clinical limitations in the current version of the PediaSat system when used during major surgery. The best agreement of SPediaSatcvO2 and SCO-OXcvO2 values was from samples taken near to the time of calibration, which might be explained by an incorrect calibration algorithm. This assumption is further supported by the finding of a near linear dependency of the bias of SPediaSatcvO2 and SCO-OXcvO2 on SCO-OXcvO2. Based on the data of the current investigation, for clinical use, this version of the PediaSat system would require recalibration significantly more frequently than Huber et al.6 have recommended, but then its use is associated with a poor cost–benefit ratio. In addition, the catheters are stiffer and thicker than nonfibreoptic catheters with the possible risk of trauma and thrombosis. In younger children and infants, small vessels might cause reflection of light from the vessel wall, and catheter dislocation leading to wedging of the catheter is not uncommon. Of greater importance is the small total blood volume in children less than 10 years, leading to greater variation in Hb concentration during sudden short-term haemodynamic disturbances that need volume resuscitation.
On the basis of the findings in the current study, the new PediaSat system when used in major paediatric surgery did not reliably reflect the central venous oxygen saturation measured by co-oximetry and cannot replace repeated invasive ScvO2 assessments in the clinically relevant range of ScvO2. Additionally, our data suggest that it does not fulfil the requirements of an intraoperative trend monitor in this heterogeneous major surgery population. Before introducing the system into routine management, further investigations are necessary using a revised PediaSat system.
1 Spenceley N, Krahn G, Skippen PW, et al.
Evaluation of a pediatric central venous oximetry catheter in critically ill children. Pediatr Crit Care Med
2 Spenceley N, Skippen P, Krahn G, et al
. Continuous central venous saturation monitoring in pediatrics: a case report. Pediatr Crit Care Med 2008; 9:13–16.
3 Ranucci M, Isgro G, De La Torre T, et al
. Continuous monitoring of central venous oxygen saturation (Pediasat) in pediatric patients undergoing cardiac surgery: a validation study of a new technology. J Cardiothorac Vasc Anesth 2008; 22:847–852.
4 Baulig W, Spielmann N, Zaiter H, et al
. In-vitro evaluation of the PediaSat continuous central venous oxygenation monitoring system. Eur J Anaesthesiol 2010; 27:289–294.
5 Molnar Z, Umgelter A, Toth I, et al
. Continuous monitoring of ScvO(2) by a new fibre-optic technology compared with blood gas oximetry in critically ill patients: a multicentre study. Intensive Care Med 2007; 33:1767–1770.
6 Huber D, Osthaus WA, Optenhofel J, et al
. Continuous monitoring of central venous oxygen saturation in neonates and small infants: in vitro evaluation of two different oximetry catheters. Paediatr Anaesth 2006; 16:1257–1261.
7 Bozkurt A, Rosen A, Rosen H, et al
. A portable near infrared spectroscopy system for bedside monitoring of newborn brain. Biomed Eng Online 2005; 4:29.
8 Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1:307–310.
9 Mahutte CK, Jaffe MB, Sasse SA, et al
. Relationship of thermodilution cardiac output to metabolic measurements and mixed venous oxygen saturation. Chest 1993; 104:1236–1242.
10 Scuderi PE, MacGregor DA, Bowton DL, et al
. A laboratory comparison of three pulmonary artery oximetry catheters. Anesthesiology 1994; 81:245–253.
11 Janvier G, Guenard H, Lomenech AM. In vitro accuracy of three blood O2
saturation optic catheter systems. Intensive Care Med 1994; 20:480–483.
12 Burchell SA, Yu M, Takiguchi SA, et al
. Evaluation of a continuous cardiac output and mixed venous oxygen saturation catheter in critically ill surgical patients. Crit Care Med 1997; 25:388–391.
13 Liakopoulos OJ, Ho JK, Yezbick A, et al
. An experimental and clinical evaluation of a novel central venous catheter with integrated oximetry for pediatric patients undergoing cardiac surgery. Anesth Analg 2007; 105:1598–1604, table of contents.
14 Baulig W, Dullenkopf A, Kobler A, et al
. Accuracy of continuous central venous oxygen saturation monitoring in patients undergoing cardiac surgery. J Clin Monit Comput 2008; 22:183–188.
15 Baulig W, Dullenkopf A, Hasenclever P, et al
. In vitro evaluation of the CeVOX continuous central venous oxygenation monitoring system. Anaesthesia 2008; 63:412–417.
16 Hedberg P, Majava A, Kiviluoma K, et al
. Potential preanalytical errors in whole-blood analysis: effect of syringe sample volume on blood gas, electrolyte and lactate values. Scand J Clin Lab Invest 2009; 69:585–591.