Agreement between PiCCO pulse-contour analysis, pulmonal artery thermodilution and transthoracic thermodilution during off-pump coronary artery by-pass surgery

Authors

  • P. S. Halvorsen,

    Corresponding author
    1. 1The Interventional Centre, Departments of 2Anesthesiology, 3Cardio-Thoracic Surgeryand 4Biostatistics at Rikshospitalet-Radiumhospitalet University Hospitaland 5University of Oslo, Department Group of Clinical Medicine, Oslo, Norway
    Search for more papers by this author
  • 1 A. Espinoza,

    1. 1The Interventional Centre, Departments of 2Anesthesiology, 3Cardio-Thoracic Surgeryand 4Biostatistics at Rikshospitalet-Radiumhospitalet University Hospitaland 5University of Oslo, Department Group of Clinical Medicine, Oslo, Norway
    Search for more papers by this author
  • 1,2 R. Lundblad,

    1. 1The Interventional Centre, Departments of 2Anesthesiology, 3Cardio-Thoracic Surgeryand 4Biostatistics at Rikshospitalet-Radiumhospitalet University Hospitaland 5University of Oslo, Department Group of Clinical Medicine, Oslo, Norway
    Search for more papers by this author
  • 1,3 M. Cvancarova,

    1. 1The Interventional Centre, Departments of 2Anesthesiology, 3Cardio-Thoracic Surgeryand 4Biostatistics at Rikshospitalet-Radiumhospitalet University Hospitaland 5University of Oslo, Department Group of Clinical Medicine, Oslo, Norway
    Search for more papers by this author
  • 4 P. K. Hol,

    1. 1The Interventional Centre, Departments of 2Anesthesiology, 3Cardio-Thoracic Surgeryand 4Biostatistics at Rikshospitalet-Radiumhospitalet University Hospitaland 5University of Oslo, Department Group of Clinical Medicine, Oslo, Norway
    Search for more papers by this author
  • 1 E. Fosse,

    1. 1The Interventional Centre, Departments of 2Anesthesiology, 3Cardio-Thoracic Surgeryand 4Biostatistics at Rikshospitalet-Radiumhospitalet University Hospitaland 5University of Oslo, Department Group of Clinical Medicine, Oslo, Norway
    Search for more papers by this author
  • and 1,5 T. I. Tønnessen 1,2,5

    1. 1The Interventional Centre, Departments of 2Anesthesiology, 3Cardio-Thoracic Surgeryand 4Biostatistics at Rikshospitalet-Radiumhospitalet University Hospitaland 5University of Oslo, Department Group of Clinical Medicine, Oslo, Norway
    Search for more papers by this author

Per Steinar Halvorsen
The Interventional Centre
Rikshospitalet-Radiumhospitalet University Hospital
0027 Oslo
Norway
e-mail: per.steinar.halvorsen@rikshospitalet.no

Abstract

Background:  Haemodynamic instability during off-pump coronary artery bypass surgery (OPCAB) may appear rapidly, and continuous monitoring of the cardiac index (CI) during the procedure is advisable. With the PiCCO monitor, CI can be measured continuously and almost real time with pulse-contour analysis and intermittently with transthoracic thermodilution. The agreement between pulmonal artery thermodilution CI (Tpa), transthoracic thermodilution CI (Tpc) and pulse-contour CI (PCCI) during OPCAB surgery has not been evaluated sufficiently.

Methods:  In 30 patients scheduled for OPCAB surgery, a pulmonary artery catheter and a PiCCO catheter were inserted. At different time points during surgery, Tpa, Tpc and PCCI were compared. Measurements were performed after induction of anesthesia (T1), after pericardiothomy (T2), after grafting on the anterior (T3), posterior (T4) and lateral (T5) walls and after chest closure (T6). The PCCI was recalibrated at time point T2–T6.

Results:  Mean difference and the limits of agreements (percentage error) between Tpa and Tpc were: –0.14 ± 0.60 (22.0%) l/min/m2, between Tpa and PCCI: –0.07 ± 0.92 (33.5%) l/min/m2 and between Tpc and PCCI: 0.10 ± 1.00 (35.5%) l/min/m2. For changes in CI from one time point to the next (ΔCI), the limits of agreements between ΔCI Tpa and ΔCI Tpc were 0.04 ± 0.90 l/min/m2, between ΔCI Tpa and ΔCI PCCI: –0.02 ± 1.22 l/min/m2 and between ΔCI Tpc and ΔCI PCCI: –0.08 ± 1.32 l/min/m2.

Conclusion:  In OPCAB surgery, limits of agreement comparing thermodilution methods were smaller than comparing PCCI with thermodilution. Recalibration of PCCI is therefore advisable.

Hemodynamic instability is common during off-pump coronary artery bypass (OPCAB) surgery (1–3). Therefore, invasive cardiovascular monitoring with accurate cardiac index (CI) measurements is advisable during this procedure. Intermittent thermodilution measurements of CI with a pulmonary artery catheter may not detect rapid and transient hemodynamic changes. In steady-state conditions, a continuous thermodilution method measured by a modified pulmonary artery catheter has shown acceptable agreement with the intermittent bolus technique (4,5), but the continuous CI value is delayed by several minutes compared with real-time CI (6,7). Thus, deterioration may not be detected early enough. This makes continuous CI measured with the pulmonary artery catheter unsuitable for OPCAB surgery. Conversely the pulse-contour method of determining CI has gained interest because it reveals almost a beat-to-beat cardiac index (8,9). In several studies conducted in different clinical settings, good agreement has been shown between pulse-contour analysis and pulmonary artery thermodilution (10–14). In principle, the pulse-contour method estimates flow based on pressure. Different pulse-contour algorithms are used by different monitoring systems. The pulse-contour CI, measured with the PiCCO monitor, is based on an algorithm developed by Wesseling et al. (15). Stroke volume of the left ventricle is derived from the systolic part of the arterial pressure waveform and the patient’s individual aortic impedance. The Wesseling algorithm uses three properties of the aorta and the arterial system: the aortic impedance, the aortic compliance and the vascular resistance. They are all non-linear pressure-dependent properties of the arterial system and differ from patient to patient. Transthoracic thermodilution is therefore necessary to calculate the aortic impedance and to calibrate the pulse-contour analysis. During hemodynamic instability, the accuracy of the pulse-contour algorithm has been questioned and some authors recommend recalibrating the pulse-contour analysis if the systemic vascular resistance index (SVRI) or mean arterial pressure (MAP) change more than 20–50% (16,17), while others do not (11,18,19).

Although the pulse-contour method has gained popularity, the method has not been sufficiently evaluated during OPCAB surgery. In this study, we assess the agreement of the PiCCO pulse-contour CI with pulmonary artery thermodilution and transthoracic thermodilution CI. During surgery, comparisons of the methods are performed at six different time points.

Patients and methods

The Regional Ethics Committee approved the study and informed consent was obtained from all patients. Patients with unstable angina pectoris, arrhythmias, valvular heart disease, intracardiac shunts and severe peripheral arterial occluding diseases were excluded. All patients had an ejection fraction above 0.30.

Thirty patients (26 men, four women) with median age of 63 years (range 39–86), median body surface area (BSA) of 2.02 m (2) (range 1.72–2.56), New York Heart Association classification of angina pectoris (NYHA) of median 2.0 (range 1–4) and ejection fraction (EF) of median 0.72 (range 0.37–0.88) were enrolled in the study. Sixteen patients (53%) had had a previous myocardial infarction and 27 (90%) of the patients received pre-operative beta-blockers. The number of distal coronary anastomoses performed was median 3.0 (range 1–4). One patient was converted from OPCAB to on pump surgery before grafting as a result of severe myocardial hypertrophy and fragile coronary arteries. Surgery on the anterior wall was performed in 29 patients, on the posterior wall in 15 patients and on the lateral wall in 24 patients. Ventricular fibrillation occurred during reperfusion of the circumflex artery in one patient. He was successfully defibrillated and recovered without sequelae. Two patients had AV-block during occlusion of the right coronary artery. Both patients recovered spontaneously to sinus rhythm after reperfusion. There were no peri-operative deaths, myocardial or cerebral infarctions.

Anesthesia

The patients were premedicated with 5–15 mg diazepam. An arterial catheter was placed in the radial artery and anesthesia was induced with 2–5 μg/kg fentanyl, 0.1 mg/kg diazepam, 2–5 mg/kg thiopental and 0.15 mg/kg cisatracurium. Anesthesia was maintained with sevoflurane and fentanyl. Vasoactive drugs were given to maintain MAP between 60 and 100 mmHg and atropine was given to maintain a heart rate above 45 beats/min.

A 7.5-F pulmonary artery catheter (ref: 774HF75; Edwards Lifesciences, Irvine, CA) was inserted via a 9.0-F triple lumen introducer (ref: M3L9FHSI; Edwards Lifesciences) in the right internal jugular vein and advanced to the wedge position guided by waveform and pressure analysis. The pulmonary artery catheter was connected to a Vigilance monitor (version 6.2; Edwards Lifesciences) with an inline temperature sensor (model 953 22) for intermittent pulmonary artery thermodilution CI. A 5-F arterial thermodilution catheter (PV 2025L20, PULSIOCATH; Pulsion Medical Systems, Munich, Germany) was inserted in the right femoral artery and connected to a pressure transducer (PV8115; PULSION Medical Systems) and the signals were computed in a PiCCO monitor (Pulsion Plus, software version 5.1, PULSION Medical Systems). A PiCCO inline injectate temperature sensor was fitted in the central venous line for transpulmonary thermodilution CI measurements. Randomly with respect to the respiratory cycle, three subsequent boluses of 10 ml of ice cold glucose 5% were injected in the proximal port of the pulmonary artery catheter and thermodilution CI for both the PiCCO and the pulmonary artery catheter were then simultaneously obtained. The injections were carried out by the same person to minimize variation, and the mean values of the thermodilution CI for the pulmonary artery catheter and PiCCO catheter were compared. The mean value of CI obtained by transthoracic thermodilution was used to calibrate the pulse-contour analysis.

Measurements

Intermittent pulmonary artery thermodilution CI and intermittent transthoracic thermodilution CI were measured after induction of anesthesia (T1), after pericardiotomy (T2), after grafting on the anterior wall (T3), posterior wall (T4), lateral wall (T5) and finally after chest closure (T6). Pulse-contour measurements were obtained at T2 to T6. Because the mean transthoracic thermodilution CI at T1 was used for initial calibration of the pulse-contour analysis, measurements and comparisons between the pulse-contour analysis and the thermodilution methods could not be performed at that time point. Before measurements were undertaken at T3 to T5, the heart was repositioned allowing it to recover for 5 min after manipulation or ischemia. Measurements were obtained without any surgical interruption and with the operating table in zero position. The recalibration period lasted typically 3 to 4 min and hemodynamic variables during this time interval were recorded. At each time point heart rate (HR), MAP and SvO2 immediately before and after calibration were compared.

All hemodynamic data were collected with the software program ICUPilot (CMA, Stockholm, Sweden) and the program was set to collect data every fifth second. The pulse-contour was recalibrated at T2 to T6 and the mean of the three last pulse-contour CI values immediately before calibration and the mean of the three successive intermittent thermodilution pulmonary artery CI and intermittent transthoracic thermodilution CI values at each time point were used when comparing the methods.

Statistical analysis

The variables which were normally distributed were analysed with multiple t-tests with Bonferroni’s correction. Wilcoxon’s rank sum test with Bonferroni’s correction was used for variables with skewed distributions. Because of the unequal number of patients at different time points, the analysis of variance for repeated measurements was not used. Measuring agreement between pairs of the three methods was assessed applying the Bland–Altman method. At the different time points each patient was measured three times. The standard deviations of the differences between the methods are therefore enlarged as described by Bland and Altman (20). Repeatability is expressed using the coefficient of variation [standard deviation (SD) divided by mean confidence interval (CI)] and computed with one-way analysis of variance. Values are presented as mean and SD except when otherwise noted. The statistical analysis was computed with SPSS (Version 12; SPSS Inc., Chicago, IL).

Results

Hemodynamics

Hemodynamic changes during surgery (T1 to T6) are depicted in Table 1. The most pronounced change in hemodynamics was seen after opening and closure of the thorax. There was no significant difference in vasoactive medication between any time points, although surgery on the lateral wall of the heart (T5) required more vasoactive medication (data not shown).

Table 1. 
Hemodynamics and cardiac index measured with the different methods at time point T1–T6.
 T1T2T3T4T5T6
  • Data are presented as mean (SD). HR, heart rate; MAP, mean arterial pressure; SVRI, systemic vascular resistance index; SvO2, mixed venous saturation; Tpa, thermodilution pulmonary artery catheter; Tpc, transthoracic thermodilution with the PiCCO catheter; PCCI, pulse-contour CI. Significant differences between the methods at different time points were observed for Tpa, Tpc at T4 and T6.

  • *

    P < 0.01.

HR (beats pr. min.)52 (7)69 (14)67 (10)66 (9)74 (15)69 (12)
MAP (mmHg)70 (11)84 (13)78 (10)76 (14)75 (10)68 (10)
SVRI (dynes • sec/cm5/m2)2314 (700)1922 (428)1929 (399)1848 (389)1868 (457)1679 (418)
SvO2 %75 (7)78 (6)76 (6)74 (4)77 (5)68 (6)
Tpa (l/min/m2)2.2 (0.4)2.9 (0.7)2.9 (0.7)2.7 (0.6)*2.7 (0.5)2.4 (0.6)*
Tpc (l/min/m2)2.4 (0.5)3.0 (0.7)3.0 (0.6)2.9 (0.7)*2.8 (0.5)2.8 (0.6)*
PCCI (l/min/m2)3.0 (0.6)2.9 (0.6)2.8 (0.7)2.9 (0.5)2.6 (0.6)

Cardiac index

The coefficients of variation for the repeated thermodilution measurements at each time point were 5.5% for the pulmonal artery thermodilution method and 6.5% for the transthoracic thermodilution method. For the pulse-contour method, the three repeated values obtained immediately before recalibration were used for calculation of the coefficients of variation. These values were achieved during a time frame of 15 s. Thus, the computed coefficients of variation were only 0.6% for the pulse-contour method.

At T4 and T6, there were significant changes between CI measured by the pulmonary artery and transthoracic thermodilution (Table 1). No significant differences were observed between the pulse-contour method and the thermodilution methods. The limits of agreements expressed by mean difference ± 2SD (percentage error) between pulmonary artery thermodilution and transthoracic thermodilution were: –0.14 ± 0.60 (22.0%) l/min/m2, between pulmonary artery thermodilution and pulse-contour: –0.07 ± 0.92 (33.5%) l/min/m2 and between transthoracic thermodilution and pulse-contour CI: 0.10 ± 1.00 (35.5%) l/min/m2. The comparison between the methods assessed by limits of agreements with their corresponding 95% CI intervals and percentage errors is shown in Table 2 and in Fig. 1(A–C). As seen from these results, the closest agreement is observed between the two thermodilution methods. Table 2 and Fig. 2(A–C) depict the agreements between the three methods expressed with delta CI (ΔCI) defined by a change in CI from one time point to the next. The best agreement in ΔCI is observed between the two thermodilution methods.

Table 2. 
The mean difference and the limits of agreements of cardiac index (l/min/m2) and delta cardiac index (l/min/m2) between the methods for all time points.
 nMean diff. (95% CI)Limits of agreement
Lower (95% CI)Upper (95% CI)
  1. Tpc, transthoracic thermodilution CI with the PiCCO catheter; Tpa, thermodilution CI with the pulmonary artery catheter; PCCI, pulse-contour CI with the PiCCO catheter. Δ, change in cardiac index from one time point to the next.

Tpa vs.Tpc141−0.14 (−0.19: −0.09)−0.74 (−0.90; −0.58)0.46 (0.30; 0.62)
Tpa vs. PCCI107−0.07 (−0.16; 0.02)−0.99 (−1.22; −0.76)0.85 (0.62; 1.08)
Tpc vs. PCCI1070.10 (−0.00; 0.20)−0.90 (−1.14; −0.66)1.10 (0.86; 1.34)
ΔTpa vs. ΔTpc1410.04 (−0.01; 0.10)−0.81 (−0.67; −0.95)0.9 (0.76; 1.04)
ΔTpa vs. ΔPCCI107−0.02 (−0.1; 0.06)−1.4 (−1.02; −1.46)1.2 (0.98; 1.42)
ΔTpc vs. ΔPCCI107−0.08 (−0.17; 0.01)−1.40 (−1.20; 1.60)1.24 (1.04; 1.44)
Figure 1.

Bland-Altman plots, showing the mean difference and limilts of agreement with corresponding 95% confidence intervals (dotted lines) between the different methods. A: Pulmonary artery thermodiluation CI (Tpa) vs Transthoracic thermodilution CI (Tpc). B: Pulmonary artery thermodilution CI (Tpa) vs Pulse-contour CI (PCCI). C: Transthoracic thermodilution CI (Tpc) vs Pulse-contour (PCCI).

Figure 2.

Bland-Altman plots, showing the mean difference and limits agreement with corresponding 95% confidence intervals (dotted lines) between the different methods for the ΔCI; change in CI from one time point to the next. A: ΔPulmonary artery thermodilution CI (ΔTpa) vs ΔTransthoracic thermodilution CI (ΔTpc). B: ΔPulmonary artery thermodilution CI (ΔTpa) vs ΔPulse-contour CI (ΔPCCI). C: ΔTransthoracic thermodilution CI (ΔTpc) vs ΔPulse-contour CI (ΔPCCI).

Table 3 lists the agreement for each time point during surgery. The agreement at each time point was better between the thermodilution methods than between the pulse-contour method and the thermodilution methods.

Table 3. 
Agreement in cardiac index (l/min/m2) between the methods at each time point assessed with the mean difference and the limits of agreement.
 nMean diff. (95% CI)Limits of agreement
Lower (95% CI)Upper (95% CI)
  1. Tpa, pulmonal artery thermodilutin CI; Tpc, transthoracic thermodilution CI and PCCI, pulse-contour CI.

Tpa vs.TpcT127−0.11 (−0.24; 0.02)−0.76 (−0.91; −0.61)0.54 (0.39; 0.69)
Tpa vs.TpcT227−0.14 (−0.28; 0.00)−0.88 (−1.05; −0.70)0.60 (0.42; 0.77)
Tpa vs.TpcT328−0.12 (−0.26; 0.02)−0.86 (−1.03; −0.69)0.62 (0.45; 0.79)
Tpa vs.TpcT414−0.22 (−0.40; −0.04)−0.90 (−1.12; −0.68)0.46 (0.24; 0.68)
Tpa vs.TpcT518−0.06 (−0.25; 0.13)−0.85 (−1.08; −0.62)0.73 (0.50; 0.96)
Tpa vs.TpcT627−0.32 (−0.49; −0.15)−1.20 (−1.40; −0.99)0.56 (0.35; 0.76)
Tpa vs. PCCIT2240.04 (−0.29; 0.37)−1.57 (−1.98; −1.17)1.65 (1.25; 2.06)
Tpa vs. PCCIT3260.03 (−0.21; 0.27)−1.21 (−1.51; −0.92)1.27 (0.98; 1.57)
Tpa vs. PCCIT412−0.08 (−0.33; 0.17)−0.96 (−1.27; −0.65)0.80 (0.49; 1.11)
Tpa vs. PCCIT519−0.21 (−0.50; 0.08)−1.48 (−1.84; −1.13)1.06 (0.71; 1.42)
Tpa vs. PCCIT626−0.15 (−0.36; 0.06)−1.20 (−1.45; −0.95)0.90 (0.65; 1.15)
Tpc vs. PCCIT2240.10 (−0.25; 0.45)−1.63 (−2.06; −1.19)1.83 (1.39; 2.26)
Tpc vs. PCCIT3260.13 (−0.11; 0.37)−1.09 (−1.38; −0.79)1.35 (1.05; 1.64)
Tpc vs. PCCIT4120.14 (−0.25; 0.53)−1.22 (−1.70; −0.74)1.50 (1.02; 1.98)
Tpc vs. PCCIT518−0.15 (−0.43; 0.13)−1.34 (−1.68; −1.00)1.04 (0.70; 1.38)
Tpc vs. PCCIT6270.20 (−0.06; 0.46)−1.16 (−1.48; −0.84)1.56 (1.24; 1.88)

Hemodynamic stability, assessed by variation in MAP, HR and SvO2, during the different recalibration periods was also obtained. A significant change was only seen for MAP at T2 (data not shown).

Discussion

This is the first study on the agreement between pulse-contour analysis, pulmonary artery thermodilution and transthoracic thermodilution during multiple vessel OBCAB surgery. We found clinically acceptable mean differences and limits of agreements between the methods. The best agreement was seen between the two thermodilution methods. When compared with the thermodilution methods, the limits of agreement for the pulse-contour analysis is wider, which indicates that recalibration of the pulse-contour analysis is of advantage during OPCAB surgery.

When comparing a new method for assessing the cardiac index to a ‘gold standard’ method such as the pulmonary artery thermodilution, a clinically acceptable agreement is often based on a subjective clinical judgment (21). In order to define objective criteria, Critchley et al. proposed that an acceptance of a new method should rely on limits of agreement of up to ± 30%, for the methods to be used interchangeably (22). In our study, we used repeated measurements and calculated the coefficient of variation of 5.5% for the pulmonary artery and 6.5% for the transthoracic thermodilution measurements. This is comparable to other reports (8,9). When the limits of agreement are adjusted, considering the repeated measurements, as suggested by Critchley et al., our calculated agreement between pulmonary artery thermodilution and transthoracic thermodilution measurements is clinically acceptable with limits of agreement of ± 22%.

We found less agreement between the pulse-contour analysis and the two thermodilution methods with limits of agreement ± 33.5% and ± 35.5%. Thus, if one should strictly follow the recommendations by Critchley et al., the pulse-contour method is not interchangeable with the two thermodilution methods. Nevertheless, before concluding on this, one should also consider the clinical situation in which the methods are being used. The clinical setting of OPCAB surgery is special as hemodynamic instability is common and rapid pump failure may appear (3,23). Intermittent and even continuous thermodilution methods may not detect these changes. Therefore, these methods are unsuitable in OPCAB surgery and the advantage of obtaining beat-to-beat CI in OPCAB surgery is obvious. Thus, in this clinical situation, these methods are not in principle interchangeable. However, reliance of the pulse-contour method has to be founded on documentation of acceptable agreement between the pulse-contour analysis and the gold standard method; otherwise incorrect decisions may occur. Our observed limits of agreements of ± 33.5% and ± 35.5% are close to the clinically acceptable limits recommended in the paper by Critchley et al. Given the rapid hemodynamic changes during the OPCAB procedure, we argue that the pulse-contour method is preferable and that the agreement between the thermodilution methods is clinically acceptable.

To some extent our results differ from the other two studies reporting on the agreement between pulse-contour and thermodilution measurements in OPCAB surgery (8,9). In these studies, the reported limits of agreement between thermodilution measurements and pulse-contour analysis are closer (0003 ± 1.26 l/min and 0.1 ± 0.84 l/min, respectively). We found limits of agreements of ± 1.86 l/min and ± 2.0 l/min when converting our indexed values to l/min. Methodological differences and patient selection may explain some of the discrepancy seen between these studies. Goedje et al. compared transthoracic thermodilution cardiac output with the average of the pulse-contour values immediately before and after calibration. This improves the agreement because the latter value has been automatically recalibrated by the thermodilution measurements. In our study, we only compared the pulse-contour value immediately before the calibration with the mean values of the following thermodilution measurements. In our opinion, this reflects the problem more precisely. In a study including pediatric patients scheduled for heart surgery, Mahajan et al. used the same method of comparing pulse-contour with thermodilution as in our study (24). In addition, they repeated the analysis using simultaneous values of the recalibrated pulse-contour analysis which improved their results from 0.10 ± 1.94 to 0.06 ± 1.26 l/min/m2.

In our study, patients with multivessel disease were included. In the other two studies, patients with only anterior wall surgery were included. As the hemodynamic alternations during anterior wall surgery are less (10–20% change in CI and SVRI) compared with changes seen when grafting on the posterior and lateral walls (45–50% change in CI and SVRI), this may also account for the different results compared with our study (2,3). In addition, differences in the amount of inotropes and fluid used during surgery may also have contributed to the discrepancy.

During hemodynamic instability, reliance of the pulse-contour analysis is being questioned and the results on the agreement between pulse-contour and thermodilution measurements differ considerably in the literature (11,13,16–19). Thus, there is no consensus regarding when and how often recalibration of the pulse-contour analysis should be performed.

In the post-operative period after on-pump heart surgery, Rodig et al. reported a mean difference and limits of agreement of 0.33 ± 2.99 l/min. At this time, sedation was turned off and MAP and SVRI varied the most. Recalibration of the pulse-contour analysis was recommended after administration of vasoactive drugs, recovery from anesthesia or if SVRI changed more than 50%. Godje et al. reports limits of agreement of ± 2.4 l/min between pulse-contour analysis and thermodilution. These limits were judged clinically acceptable during hemodynamic instability and recalibration was not found necessary even after inotropes in different doses were given. This conclusion was based on a subjective judgement and obviously one could question if the reported limits really are clinically acceptable.

Felbinger et al. assessed agreement in ΔCI between pulse-contour analysis and thermodilution during rapid fluid administration. The observed limits of agreement were ± 0.52 l/min/m2 between the pulse-contour and thermodilution measurements. Recalibration was not required in this clinical setting. In their study, the change in CI was minor with a mean change of 15% and the measurement period was short (mean 15 min). In addition, no records on ΔMAP and ΔSVRI were presented. Therefore, the authors, in our opinion, should not make a firm conclusion regarding recalibration.

Our study indicates that recalibration of the pulse-contour is advisable during OPCAB surgery. Although no significance was observed in mean differences between the pulse-contour analysis and the thermodilution methods at T2–T6, wider limits of agreements were depicted at all time points when compared with the corresponding limits of agreement ranges between the thermodilution methods. Thus, the narrower limits of agreement observed between pulmonal artery and transthoracic thermodilution imply that the pulse-contour analysis should be recalibrated. By performing recalibrating of the pulse-contour analysis, the transthoracic thermodilution CI and the pulse-contour CI will be the same. Consequently, the pulse-contour analysis would be closer to the gold standard method (pulmonal artery thermodilution). In addition, our results expressed by the limits of agreement of the ΔCI values also support the hypothesis that recalibration of the pulse-contour analysis is recommendable, as the closest agreement is seen between the ΔCI thermodilution values. Considering this and the fact that the OPCAB procedure implies rapid hemodynamic changes, we recommend recalibrating after opening and closure of the thorax and after grafting on the different heart walls.

Limitations of the study

During the recalibration periods, the pulse-contour CI value and the mean CI values of pulmonary artery thermodilution and the transthoracic thermodilution were obtained at slightly different time points (3–4 min difference). Thus, hemodynamic changes during the recalibration periods might have influenced the agreement between the pulse-contour analysis and the thermodilution methods. In contrast, the agreement between the two thermodilution methods may be closer as the measurements were performed simultaneously in the recalibration periods. Because we found hemodynamically stable recalibration periods and acceptable coefficient of variation for the thermodilution methods, the alterations in circulation during the recalibration period does not fully explain the diverging results on the agreement.

Another limitation of our study is the lack of a control group in which the pulse-contour analysis was not recalibrated during surgery. Although our results shows that recalibration of the pulse-contour analysis is advisable and improves the agreement towards the gold standard method (pulmonary artery thermodilution), we do not know to what extent recalibration effects the agreement.

Another limitation is the difference in the coefficient of variation for the three methods, which to some degree influences the calculation of the limits of agreements. The coefficient of variation is greater for the thermodilution methods than for the pulse-contour method. Hence, the limits of agreement seen between the thermodilution methods are wider than the corresponding limits of agreements between the thermodilution methods and the pulse-contour analysis. In the present study, this does not affect our conclusions.

Conclusion

This study shows clinically acceptable agreement between the PiCCO pulse-contour analysis, pulmonary artery and transthoracic thermodilution in OPCAB patients. Recalibration of the pulse-contour analysis is advisable and improves the agreement.

Ancillary