British Journal of Anaesthesia 106 (4): 475–81 (2011)

Advance Access publication 31 December 2010 . doi:10.1093/bja/aeq372

CARDIOVASCULAR

Lack of agreement between pulmonary arterial

thermodilution cardiac output and the pressure recording
analytical method in postoperative cardiac surgery patients‡
H. Paarmann 1†, H. V. Groesdonk 1,3†, B. Sedemund-Adib 1, T. Hanke 2, H. Heinze 1, M. Heringlake 1*
and J. Schön 1
1
Department of Anesthesiology and 2 Department of Cardiac and Thoracic Vascular Surgery, University of Lübeck, Ratzeburger Allee 160,
D-23538 Lübeck, Germany
3
Department of Cardiac and Thoracic Vascular Surgery, University of Saarland, Homburg/Saar, Germany
* Corresponding author. E-mail: heringlake@t-online.de

Editor’s key points
† The pressure recording
analytical method (PRAM)
is a recent development
of uncalibrated pulse-
contour analysis-based
methods for CO
determination.
† 46 paired CO
measurements of PRAM-
and thermodilution-CO
were obtained in 23
patients after cardiac
surgery.
† Large differences
between PRAM-CO and
ThD-CO were found with
percentage error of 87%.
† PRAM-CO cannot replace
thermodilution-CO in
cardiac surgery patients.
Background. Pulse-contour analysis method (PCM) cardiac output (CO) monitors are
increasingly used for CO monitoring during anaesthesia and in the critically ill. Very
recently, several systems have been introduced that do not need calibration; among
them the pressure recording analytical method (PRAM). Sparse data comparing the
accuracy of the PRAM-CO with conventional thermodilution CO (ThD-CO) in cardiac
surgery patients are available.
Methods. In this prospective comparison study, paired CO measurements with a pulmonary
artery catheter and a PRAM monitoring set were obtained 20– 30 min apart (t1, t2) in 23
extubated patients on the first postoperative day after cardiac surgery. Data were
analysed by the Bland–Altman method.
Results. A total of 46 paired CO measurements (23 for each interval) were collected. The
Bland –Altman analysis showed a mean difference (bias) of 0.0 litre min21 and limits of
agreement (1.96 SD) of 4.53 to 24.54 litre min21 [upper 95% confidence interval (CI),
3.26 –5.80; lower 95% CI, 25.8 to 23.27]. The percentage error (1.96 SD/mean of the
reference method) was 87%.
Conclusions. These results question the reliability of the PRAM technology for the
determination of CO in postoperative cardiac surgery patients.
Keywords: cardiac surgery; haemodynamic monitoring; pressure recorded analytical
method; pulmonary arterial thermodilution; pulse-contour analysis
Accepted for publication: 24 October 2010

Establishing and maintaining adequate cardiac output (CO)
and oxygen delivery is a pivotal part of perioperative haemo-
dynamic management in patients undergoing cardiac
surgery.1 Traditionally, CO monitoring in this field has been
accomplished by the use of a pulmonary artery catheter
(PAC) and pulmonary arterial thermodilution (PA-ThD), an
approach that is still favoured by many clinicians.2
Arterial pulse-contour method (PCM) systems offer a
promising alternative method for the determination of CO.
Several of these monitoring systems are increasingly used
†
These authors contributed equally to this work.
‡
This article is accompanied by Editorial II.
during cardiac surgery.3 – 6 The pressure recording analytical
method (PRAM) is a relatively new technology that does
not require external calibration or adjustments for age,
gender, or anthropometric variables (height and weight).
The technique is based on the analysis of peripheral artery
waveform morphology and has been extensively described
elsewhere.7 8 It is of note that PRAM is recommended also
for use in patients treated with an intra-aortic balloon
pump (IABP),9 10 an unique feature in comparison with the
other commercially available PCM technologies.

& The Author [2010]. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved.
For Permissions, please email: journals.permissions@oup.com
BJA
Experimental data show an acceptable agreement of
PRAM-derived CO (PRAM-CO) in comparison with (ThD-CO)
and suggest that PRAM is a useful alternative for CO monitor-
ing.11 12 However, clinical comparative data about the accu-
racy of the PRAM system in comparison with standard
technologies are still sparse.8 – 10 13
The present study was designed to determine the
reliability of PRAM-CO in comparison with ThD-CO in a het-
erogeneous group of patients after cardiac surgery.
Methods
Patients
After approval by the local ethical committee and preopera-
tive written informed consent, 27 ASA consecutive physical
status III patients (13 females/14 males) were enrolled.
Four patients needing postoperative treatment with an
IABP were excluded from further analysis.
Mean age, height, and weight were 67 (range 51 –82) yr,
166 (154– 181) cm, and 71 (56– 115) kg, respectively.
All patients had undergone cardiac surgery with cardio-
pulmonary bypass and moderate hypothermia. The surgical
procedures were: coronary artery bypass grafting (CABG),
n¼12; CABG and aortic replacement, n¼6; CABG and mitral
valve reconstruction, n¼2; CABG and mitral valve replace-
ment, n¼2; and CABG and aortic valve replacement and
mitral valve reconstruction, n¼5.
Patients were studied on the first postoperative day after
cardiac surgery when treated in the intensive care unit
(ICU). Patients were awake, calm, and spontaneously breath-
ing during the measurements. CO was determined twice; the
first measurement cycle was performed without supplemen-
tal oxygen, the second measurement cycle was performed
after the optimization of oxygen delivery with 4 litre min21
supplemental oxygen via a face mask with the intention to
eliminate interfering instabilities of the oxygen delivery/
oxygen demand ratio.
No patients with cardiac arrhythmias were included in this
study; all patients were either in sinus rhythm or paced by
epicardial atrial pacing electrodes. Measurements were per-
formed during stable haemodynamic conditions achieved
by continuous infusions of 0.027 –0.073 mg kg21 min21 nor-
epinephrine, 1.9 –5.7 mg kg21 min21 dobutamine, and
0.22 –0.54 mg kg21 min21 milrinone.
Four patients were treated with an IABP.
Measurement of CO by thermodilution
All patients were routinely monitored by means of a radial
artery catheter, a central venous line, and a PAC (CCombo
744HF75) for semi-continuous and bolus determination of
CO connected to a Vigilance IIw monitor (Edwards Life-
sciences, Irvine, CA, USA). ThD-CO was calculated as the
mean of at least three separate measurements not deviating
more than 20% obtained over 3 min by injection of 10 ml of
48C NaCl 0.9% solution. The two determinations of CO were
performed 12 –16.5 h after surgery 20–30 min apart (t1
and t2, respectively, see above).
476
Paarmann et al.
Basic physical principles of PRAM
In contrast to other methods for the determination of CO
from the arterial pulse like the PiCCO monitor3 14 15 and
the LidCO system,4 16 but comparable with the FlowTrac
system,4 17 – 19 the PRAM is not only based on the analysis
of the systolic area under the curve of the arterial pulse
but also on the analysis of the complete systolic and diastolic
pulse wave. The PRAM technique does not require external
calibration or adjustments for age, gender, height, or
weight. Changes in the arterial pulse wave during the com-
plete cardiac cycle systolic pressure, dicrotic notch, diastolic
pressure, and additional waves that may result from pulse
wave reflection (termed: points of instability) are recorded
with a sampling rate of 1000 min21 and analysed according
to the physical principles of perturbation, an analytic
approach to approximate a value that cannot be calculated
precisely. The theoretical and physical principles of this
method have been extensively described elsewhere.7 8
Measurement of CO by PRAM
A PRAM-CO MostCare monitoring set (Vygon GmbH & Co. KG,
Aachen, Germany) for continuous determination of CO was
connected to the standard ICU haemodynamic monitor
(Infinity, Dräger, Lübeck, Germany) and the system was set
up and started according to the instructions of the manufac-
turer. The average of three 1 min continuous registrations
with 1 min intervals between each was taken as the repre-
sentative result.
Statistical analysis
The sample size was adjusted to comparable comparison
studies.11 20 No formal power analysis was performed. A
total of 54 paired CO measurements (27 for each interval)
were collected. Patients requiring temporary support by an
IABP were excluded from the analysis, because of lack of
power in this small group of four patients. This exclusion
left 46 data pairs to be analysed.
Analyses were performed by MedCalc 11.0 for Windows.
After analysis for normal distribution by the Kolmogorov –
Smirnov test, data were analysed parametrically by Student’s
t-test. Pearson’s correlation coefficient was used to compare
PRAM-CO and ThD-CO. Further, the Bland–Altman statistics,
as well as percentage errors (1.96 SD/mean of reference
method) were calculated. A mean percentage error not
exceeding 30% was defined to indicate clinical useful
reliability of the PRAM-CO.21 Data are given as mean (SD); a
P-value of ,0.05 indicates statistical significance.
Results
The CO range was 2.6 –9.2 litre min21 for PA-ThD and 2.4– 12
litre min21 for PRAM.
A comparison of the measurements at t1 and t2 with a
t-test for paired samples revealed no significant differences
between ThD-CO (P¼0.305) and PRAM-CO (P¼0.280) or
other haemodynamic variables (Table 1). This finding
Lack of agreement between ThD-CO and PRAM-CO BJA
Discussion
Table 1 Haemodynamic variables without supplemental oxygen
(t1) and with supplemental oxygen (t2). SD, standard deviation; The results of the present study are clearly in contrast to
MAP, mean arterial pressure; CVP, central venous pressure; PAP, other studies using PA-ThD, direct-oxygen Fick method, and
pulmonary artery pressure; ThD-CO, thermodilution cardiac Doppler echocardiography as a comparator, all coming to
output; PRAM-CO, pressure recording analytical method cardiac
the conclusion that the PRAM technique is reliable for CO
output; Svo2 , mixed venous oxygen saturation; Sao2 , arterial
determinations during13 22 and after cardiac surgery10 12
oxygen saturation
and also during non-cardiac surgery.8 12 23 Romano and col-
t1 t2 Paired leagues studied the reliability of PRAM in a small group of
Mean SD Mean SD samples haemodynamically stable cardiac patients undergoing
t-test diagnostic right and left heart catheterization. In this
(P-value)
study, the pressure signal was recorded from an aortic cath-
MAP (mm Hg) 73.5 9.1 75.5 8.5 0.138 eter and not from a radial line; an issue that may have
CVP (mm Hg) 12.0 6.4 11.9 5.6 0.808 important effects on the pulse contour.8 Comparably, two
PAP mean (mm Hg) 25.7 8.6 26.3 9.9 0.546 other studies by the same group found a good agreement
ThD-CO (litre min21) 5.2 1.0 5.2 0.9 0.573 between PRAM-CO and PA-ThD-CO during13 and after
PRAM-CO (litre min21) 5.1 2.6 5.4 2.3 0.144 cardiac surgery.12 The number of patients was comparable
SvO2 (%) 64.8 7.2 69.3 5.3 0.002 with the size of the present study (n¼32 and 28, respect-
SaO2 (%) 95.8 2.1 98.3 1.6 ,0.0001 ively). Another study from Faltoni and colleagues9 compar-
ing PRAM-CO with ThD-CO in eight cardiac surgery patients
on IABP—available only as congress abstract—found good
supported the combination of the measurements at the two agreement between both methods, and the authors con-
time points in one analysis of a total of 46 data pairs. clude that the PRAM technique may also be suitable for
The correlation between ThD-CO and PRAM-CO was CO determination in patients treated with an IABP. Zangrillo
r¼0.313 [P¼.049; 95% confidence interval (CI), 0.0 –0.57], and colleagues10 recently showed a significant correlation
as shown in Figure 1. and clinically acceptable agreement with a percentage
The Bland –Altman analysis showed a mean difference error of 30% between PRAM cardiac index (CI) and ThD-CI
(bias) of 0.0 (95% CI, 20.74 to 0.74) litre min21 and limits measurement in 32 haemodynamically unstable patients
of agreement (1.96 SD) of 4.53 to 24.54 litre min21 (upper with IABP and/or high dose of inotropic drugs after cardiac
95% CI, 3.26– 5.80; lower 95% CI, 25.8 to 23.27) (Fig. 2). surgery. However, in this study, measurements with arti-
The percentage error (1.96 SD/mean of the reference facts caused by under- or overdamping of the arterial
method) was 87%. pressure curve were excluded without further definition of
The distribution of the data pairs in the scatter diagram criteria on the signal quality. Good agreement between
(Fig. 1) suggested a growing deviation between the the PRAM technique and PA-ThD was found in a pig
methods in the higher CO range. After stratification by the model, but the special anatomical conditions in the
median of the reference method (median ThD-CO¼5.1 litre animal model used may have had a relevant impact on
min21), a second analysis for the ‘high’ and ‘low’ CO range CO values.11
was performed. The reason for the discrepancy between the previous
Patients with higher and lower CO did not differ in mean work8 – 10 12 13 22 23 on the PRAM technology and our study
arterial pressure [ThD-CO≥5.1 litre min21: mean MAP¼73.6 is not clear. The sample size was comparable with previous
(SD 8.1) mm Hg, ThD-CO,5.1 litre min21: mean MAP¼76.1 studies and is of the typical size used for clinical comparison
(SD 9.4) mm Hg; P¼0.353], the amount of norepinephrine, studies. The population described is representative of the
dobutamine, or milrinone. typical patients treated after cardiac surgery in the ICU;
In the group with median ThD-CO,5.1 litre min21, the most of them needing minor to moderate inotropic and
correlation was r¼0.548 (P¼0.022; 95% CI, 0.1–0.81) vasopressor support. This is a typical condition where a non-
(Fig. 3A). In the Bland –Altman analysis, the bias was 20.90 invasive monitoring would be preferable to the invasive PAC.
litre min21 (95% CI, 21.86 to 0.05) and limits of agreement Further, all measurements were performed according to the
were 3.07 to 24.87 litre min21 (upper 95% CI, 1.43 –4.75; instructions of the manufacturer.
lower 95% CI, 26.56 to 23.24). The percentage error was Various factors may lead to erroneous ThD-CO measure-
93% (Fig. 4A). ments in patients after cardiac surgery, that is, acute temp-
In the group with median ThD-CO≥5.1 litre min21, the erature changes during rewarming in the immediate period
correlation was r¼0.233 (P¼0.283; 95% CI, 20.2 to 0.6) after cardiopulmonary bypass24 25 and tricuspid regurgita-
(Fig. 3B). In the Bland – Altman analysis, the bias was 0.29 tion.25 26 Although the first aspect is not applicable in the
litre min21 (95% CI, 20.68 to 1.25) and limits of agreement presented study with patients more than 12 h after
were 5.08 to 24.49 litre min21 (upper 95% CI, 3.40 –6.74; surgery, tricuspid regurgitation was ruled out by visual
lower 95% CI, 26.16 to 22.82). The percentage error was inspection of the central venous pressure curve and post-
78% (Fig. 4B). operative transoesophageal echocardiography.

477
BJA Paarmann et al.

10
Pearson‘s r=0.313, P=0.049

ThD-CO (litre min–1)

6

2 4 6 8 10 12 14
PRAM-CO (litre min–1)

Fig 1 Correlation between ThD-CO and PRAM-CO. ThD-CO, thermodilution cardiac output; PRAM-CO, pressure recorded analytical method.

6
+1.96 SD
4 4.5
ThD-CO–PRAM-CO (litre min–1)

Mean
0
0.0

–2

–4 –1.96 SD
–4.5
–6

2 4 6 8 10 12
Average of ThD-CO and PRAM-CO (litre min–1)

Fig 2 The Bland –Altman analysis comparing ThD-CO with PRAM-CO. ThD-CO, thermodilution cardiac output; PRAM-CO, pressure recorded
analytical method.

Additionally, the timing,25 27 injectate temperature,20 25
and the number of averaged bolus thermodilution mea-
surements25 may influence the precision of PA-ThD-CO
determination. However, all measurements were randomly
distributed throughout the respiratory cycle, 48C cold saline
was used as an injectate solution, and thus, it is unlikely
that differences between our and previous studies8 – 13 22
may be explained by imprecision of the reference method.
A possible explanation may be that all measurements were
determined from radial and therefore peripheral arterial lines.
It is well known that the pulse wave in the radial artery differs
from the central aortic and the femoral arterial pulse wave,14
an effect that has been shown to alter the reliability of other
PCM-CO monitors28 and may at least in part explain some dis-
crepancies of the presented to previous works. In the present
study, in a subgroup of patients with high CO, we found no

478
Lack of agreement between ThD-CO and PRAM-CO BJA

A 7
Pearson‘s r =0.548, P =0.022

ThD-CO (litre min–1)

5

2 3 4 5 6 7 8 9
PRAM-CO (litre min–1)

B 8
Pearson‘s r =0.233, P =0.283
ThD-CO (litre min–1)

2 4 6 8 10 12 14
PRAM-CO (litre min–1)

Fig 3 Correlation between ThD-CO and PRAM-CO after stratification by median thermodilution cardiac output. (A) ThD-CO,5.1 litre min21.
(B) ThD-CO≥5.1 litre min21. ThD-CO, thermodilution cardiac output; PRAM-CO, pressure recorded analytical method.

significant correlation between PRAM-CO and ThD-CO. As the
groups did not differ in the amount of inotropic agents or
vasopressors received, this lack of reliability in the subgroup
cannot be attributed to a change in arterial compliance due
to excessive vasopressor doses.
A methodological limitation is the application of the
Bland –Altman analysis for two repeated measures.29
However, this approach is appropriate for exploration of
data; therefore, the presented results have to be regarded
as descriptive rather than confirmatory.
In conclusion, the aforementioned factors do not suffi-
ciently explain the profound lack of agreement between
ThD-CO and PRAM-CO, with even no significant correlation
between the two methods in the range of higher CO. In
search for a minimally invasive tool for the low-to-medium
risk cardiac surgery patient, we chose a situation with more
or less stable haemodynamic conditions for the present
study. The immediately postoperative situation would have
involved various confounding factors as rapidly changing
arterial pressures due to changes in peripheral vascular
resistance, influence of rewarming, and agitation in awaken-
ing. Measurements in this period would have needed a much
greater sample size to allow reliable conclusions. However,
the lack of interchangeability of the ThD-CO and PRAM-CO

479
BJA Paarmann et al.

A 4
+1.96 SD
3
3.1

ThD-CO–PRAM-CO (litre min–1)

2

0
Mean
–1
–0.9
–2

–3

–4
–1.96 SD
–5
–4.9
–6

3 4 5 6 7 8
Average of ThD-CO and PRAM-CO
withThD-CO<5.1 (litre min–1)

B 6
+1.96 SD
5.1
4
ThD-CO–PRAM-CO (litre min–1)

Mean
0 0.3

–2

–4 –1.96 SD
–4.5
–6

3 4 5 6 7 8 9 10 11
Average of ThD-CO and PRAM-CO
withThD-CO > 5.1 (litre min–1)

Fig 4 The Bland –Altman analysis comparing ThD-CO with PRAM-CO after stratification by median cardiac output. (A) ThD-CO,5.1 litre min21.
(B) ThD-CO≥5.1 litre min21. ThD-CO, thermodilution cardiac output; PRAM-CO, pressure recorded analytical method.

even under stable conditions does not encourage further
studies in more complex haemodynamic situations. On the
basis of our data, we cannot make statements on reliability
of the PRAM-CO under mechanical circulatory support.
Further investigation is needed validating the use of
PRAM-CO in patients with IABP support and other devices.
Keeping in mind the limitation of the presented study,
that is, the small sample size, our findings suggest that the
underlying algorithm in the PRAM technology does not suffi-
ciently reproduce the CO measured with ThD in haemodyna-
mically stable patients after cardiac surgery. Consequently, at
least until further modifications of the algorithm are
available, the routine use of the PRAM technology for the
determination of CO in cardiac surgery patients cannot be
recommended. Ideally, future investigations should deter-
mine in a larger group of patients whether using a more
central artery for recording the arterial pressure curve
might improve the accuracy of the PRAM method.
Conflict of interest
The author M. Heringlake receives horonaria for lectures
and scientific support from Edwards Lifesciences, Irvine, CA,
USA.

480
Lack of agreement between ThD-CO and PRAM-CO
Funding
The present study was supported by Vygon GmbH & Co. KG,
Aachen, Germany, by providing the pressure recording
analytical method monitoring system.
References
1 Heringlake M, Heinze H, Misfeld M, Klaus S, Bechtel M. Goal-
directed hemodynamic optimization in high-risk cardiac surgery
patients: a tale from the past or a future obligation? Minerva
Anestesiol 2008; 74: 251 –8
2 Kastrup M, Markewitz A, Spies C, et al. Current practice of hemo-
dynamic monitoring and vasopressor and inotropic therapy in
post-operative cardiac surgery patients in Germany. Acta Anaes-
thesiol Scand 2007; 3: 347 –58
3 Buhre W, Weyland A, Kazmaier S, et al. Comparison of cardiac
output assessed by pulse contour analysis and thermodilution
in patients undergoing minimally invasive direct coronary artery
bypass grafting. J Cardiothorac Vasc Anesth 1999; 13: 437–40
4 Button D, Weibel L, Reuthebuch O, Genoni M, Zollinger A, Hofer CK.
Clinical evaluation of the FloTrac/Vigileo system and two estab-
lished continuous cardiac output monitoring devices in patients
undergoing cardiac surgery. Br J Anaesth 2007; 99: 329–36
5 De Castro V, Goarin JP, Lhotel L, Mabrouk N, Perel A, Coriat P. Com-
parison of stroke volume (SV) and stroke volume respiratory vari-
ation (SVV) measures by the axillary artery pulse-contour method
and by aortic Doppler echocardiography in patients undergoing
aortic surgery. Br J Anaesth 2006; 97: 605 –10
6 Missant C, Rex S, Wouters PF. Accuracy of cardiac output
measurements with pulse contour analysis (PulseCO) and
Doppler echocardiography during off-pump coronary bypass
grafting. Eur J Anaesthesiol 2008; 25: 243 –8
7 Bernstein DP. Pressure pulse contour-derived stroke volume and
cardiac output in the morbidly obese patient. Obes Surg 2008;
8: 1015– 21
8 Romano SM, Pistolesi M. Assessment of cardiac output from sys-
temic arterial pressure in humans. Crit Care Med 2002; 30:
1834– 41
9 Faltoni A, Franchi F, Bigio E, et al. Cardiac output thermodilution
and the MostCare system in patients on intra-aortic balloon
pump. Crit Care 2009; 13(Suppl. 1): P225
10 Zangrillo A, Maj G, Monaco F, et al. Cardiac index validation using
the pressure recording analytical method in unstable patients.
J Cardiothorac Vasc Anesth 2010; 24: 265 –9
11 Scoletta S, Romano SM, Biagioli B, Capannini G, Giomarelli P.
Pressure recording analytical method (PRAM) for measurement
of cardiac output during various haemodynamic states. Br J
Anaesth 2005; 95: 159 –65
12 Romagnoli S, Romano SM, Bevilacqua S, et al. Cardiac output by
pulse contour: reliability under hemodynamic derangements.
Interact Cardiovasc Thorac Surg 2009; 8: 642 –6
13 Giomarelli P, Biagioli B, Scolletta S. Cardiac output monitoring by
pressure recording analytical method in cardiac surgery.
J Cardiothorac Surg 2004; 26: 515 –20
14 Baba T, Goto T, Yoshitake A, Shibata Y. Radial artery diameter with
increased femoral to radial arterial pressure gradient during car-
diopulmonary bypass. Anesth Analg 1997; 85: 252 –8
BJA
15 Della Rocca G, Costa MG, Pompei L, Coccia C, Pietropaoli P. Con-
tinuous and intermittent cardiac output measurement: pulmon-
ary artery catheter versus aortic transpulmonary technique. Br J
Anaesth 2002; 88: 350–6
16 Roding G, Prasser C, Keyl C, Liebhold A, Hobbhahn J. Continuous
cardiac output measurement: pulse contour analysis vs thermo-
dilution technique in cardiac surgical patients. Br J Anaesth 1999;
82: 525–30
17 Comton FD, Zukunft B, Hoffmann C, Zidek W, Schaefer JH. Per-
formance of a minimally invasive uncalibrated cardiac output
monitoring system (Flowtrac/Vigileo) in haemodynamically
unstable patients. Br J Anaesth 2008; 100: 451–6
18 Mayer J, Boldt J, Beschmann R, Stephan A, Suttner S. Uncali-
brated arterial pressure waveform analysis for less-invasive
cardiac output determination in obese patients undergoing
cardiac surgery. Br J Anaesth 2009; 103: 185– 90
19 Chatti R, de Rudniki S, Marque S, et al. Comparison of two versions
of the Vigileo-FloTrackTM system (1.03 and 1.07) for stroke volume
estimation: a multicentre, blinded comparison with oesophageal
Doppler measurements. Br J Anaesth 2009; 102: 463–9
20 Elkayam U, Berkley R, Azen S, Weber L, Geva B, Henry WL. Cardiac
output by thermodilution technique. Effect of injectate’s volume
and temperature on accuracy and reproducibility in the critically
ill patient. Chest 1983; 84: 418– 22
21 Critchley LA, Critchley JA. A meta-analysis of studies using bias
and precision statistics to compare cardiac output measurement
techniques. J Clin Monit Comput 1999; 15: 85 –91
22 Romano SM, Scoletta S, Olivotto I, et al. Systemic arterial wave-
form analysis and assessment of blood flow during extracorpor-
eal circulation. Perfusion 2006; 21: 109–16
23 Calamandrei M, Mirabile L, Muschetta S, Gensini GF, De Simone L,
Romano SM. Assessment of cardiac output in children: a com-
parison between the pressure recording analytical method
and Doppler echocardiography. Pediatr Crit Care Med 2008; 9:
310– 2
24 Bazaral MG, Petre J, Novoa R. Errors in thermodilution cardiac
output measurements caused by rapid pulmonary artery temp-
erature decreases after cardiopulmonary bypass. Anesthesiology
1992; 77: 31–7
25 Bendjelid K, Schütz N, Stuter PM, Romand JA. Continuous cardiac
output monitoring after cardiopulmonary bypass: a comparison
with bolus thermodilution measurement. Intensive Care Med
2006; 32: 919– 22
26 Heerdt PM, Pond CG, Blessios GA, Rosenbloom M. Inaccuracy of
cardiac output by thermodilution during acute tricuspid regurgi-
tation. Ann Thorac Surg 1993; 53: 706–8
27 Jansen JRC, Versprille A. Improvement of cardiac output esti-
mation by the thermodilution method during mechanical venti-
lation. Intensive Care Med 1986; 12: 71– 9
28 Mayer J, Boldt J, Schöllhorn T, Röhm KD, Menigstu AM,
Suttner S. Semi-invasive monitoring of cardiac output by a
new device using arterial pressure waveform analysis: a com-
parison with intermittent pulmonary artery thermodilution in
patients undergoing cardiac surgery. Br J Anaesth 2007; 98:
176– 82
29 Myles PS, Cui J. Using the Bland–Altman method to measure
agreement with repeated measures. Br J Anaesth 2007; 99:
309– 11
481