Clinical assessment of cardiac performance in infants and children ...

Intensive Care Med (2005) 31:568–573
DOI 10.1007/s00134-005-2569-5
NEONATAL AND PEDIATRIC INTENSIVE CARE
Jonathan R. Egan
Marino Festa
Andrew D. Cole
Graham R. Nunn
Jonathan Gillis
David S. Winlaw
Clinical assessment of cardiac performance
in infants and children following cardiac
surgery
Received: 20 April 2004
Accepted: 20 January 2005
Published online: 15 February 2005
Springer-Verlag 2005
J. R. Egan ()) · J. Gillis
Paediatric Intensive Care Unit,
The Children’s Hospital at Westmead,
2145 Westmead, NSW, Australia
e-mail: jonathoe@chw.edu.au
Tel.: +61-2-98450000
Fax: +61-2-98453078
M. Festa
Paediatric Intensive Care Unit,
Guy’s Hospital,
St. Thomas’ Street, London, SE1 9RT, UK
A. D. Cole · G. R. Nunn · D. S. Winlaw
Adolph Basser Cardiac Institute,
The Children’s Hospital at Westmead,
2145 Westmead, NSW, Australia
Abstract Objective: To compare
clinical assessment of cardiac performance
with an invasive method of
haemodynamic monitoring. Design
and setting: Prospective observational
study in a 16-bed tertiary paediatric
intensive care unit. Patients
and participants: Infants and children
undergoing cardiopulmonary bypass
and surgical repair of congenital heart
lesions. Interventions: Based on
physical examination and routinely
available haemodynamic monitoring
in the paediatric intensive care unit,
medical and nursing staff assessed
cardiac index, systemic vascular resistance
index and volume status.
Clinical assessment was compared
with cardiac index, systemic vascular
resistance index and global end diastolic
volume index, obtained by
femoral artery thermodilution.
Measurements and results: A total of
76 clinical estimations of the three
parameters were made in 16 infants
and children undergoing biventricular
repair of congenital heart lesions.
Agreement was poor between clinical
and invasive methods of determining
all three studied parameters of cardiac
performance. Cardiac index was
significantly underestimated clinically;
mean difference was
0.71 l min Ÿ1 m Ÿ2 (95% range of
agreement €2.7). Clinical estimates
of systemic vascular resistance
(weighted k=0.15) and volume status
(weighted k=0.04) showed poor levels
of agreement with measured values
and were overestimated clinically.
There was one complication related
to a femoral arterial catheter
and one device failure. Conclusions:
Routine clinical assessment of parameters
of cardiac performance
agreed poorly with invasive determinations
of these indices. Management
decisions based on inaccurate clinical
assessments may be detrimental to
patients. Invasive haemodynamic
monitoring using femoral artery
thermodilution warrants cautious
further evaluation as there is little
agreement with clinical assessment
which is presently standard accepted
care in this patient population.
Keywords Paediatric · Cardiac
output · Thermodilution
Introduction
Clinical assessment of cardiac performance following
cardiac surgery is routinely relied upon in paediatric patients
and has become the basis for assessing new therapies
in this patient population [1]. However, the accuracy
of some aspects of clinical examination has been questioned.
Tibby et al. [2] showed a poor correlation between
clinical and invasive assessment of cardiac index. Other
parameters of cardiac performance commonly assessed
clinically, such as vascular resistance and volume status,
have not been compared with invasive methods in the
paediatric population, and similar inaccuracies may also
exist with assessment of these parameters.

569
Despite these concerns about clinical assessment alternative
approaches such as invasive haemodynamic
monitoring with pulmonary or femoral arterial thermodilution
(FATD) are not used widely in paediatrics. This
may relate to difficulties with central access, potential
complications or perhaps be due to a lack of proven
benefit in outcome attributable to their use in children [3].
Unless invasive haemodynamic monitoring is shown to be
of benefit in paediatric patients, its use is likely to remain
limited. However, outcome studies and further assessments
of invasive haemodynamic monitoring in children
are warranted only if clinical assessments of cardiac
performance in comparison are poor.
To investigate this we performed an observational
study of paediatric patients following cardiac surgery.
Three commonly assessed parameters of cardiac performance
were assessed by two methods. Clinical assessment
of cardiac output, volume status and systemic vascular
resistance were compared with invasively obtained
values for these parameters. Our experience with an invasive
form of haemodynamic monitoring in paediatric
patients was also reported.
Materials and methods
Inclusion criteria were weight between 4 and 20 kg, an operation
requiring cardiopulmonary bypass to achieve a biventricular repair
and normal pre-operative renal function. Informed consent was
obtained prior to patient enrolment, and the ethics committee of
The Children’s Hospital at Westmead approved the study.
Clinical estimates of haemodynamic status were recorded in
ventilated post-operative patients in whom objective measurements
of cardiac index (CI), systemic vascular resistance index (SVRI)
and global end-diastolic volume index (GEDVI) were performed by
FATD (PiCCO, Pulsion Medical, Munich, Germany). Cardiac
output using FATD has been validated in ventilated infants and
children using indirect calorimetry and the Fick principle by Tibby
et al. [3].
A central venous line (CVL) and a femoral intra-arterial line
(22-G catheter with 1.3-F thermistor if less than 7 kg, 4-F integrated
catheter if greater than 7 kg) were inserted intraoperatively in all
patients following induction of anaesthesia. All patients received
phenoxybenxamine just prior to the commencement of cardiopulmonary
bypass. Following completion of the operation, prior to
returning to the paediatric intensive care unit (PICU) a trans-oesophageal
echocardiogram was performed.
Upon arrival in the PICU an initial thermodilution was performed
by one of three investigators (M.F., D.W., J.E.) who took no
part in clinical assessments. Thermodilution involved injection of a
small volume (1.5 ml + 0.15 ml/kg, up to 5 ml) of 3C normal
saline into the CVL. Arterial blood temperature was sensed by the
PiCCO thermistor within the femoral arterial line. Three thermodilutions
were performed in quick succession, with data later
averaged with the analysis. Data were downloaded directly to a
computer and analysed following the completion of the study.
Thermodilution was repeated at defined time points: 6, 18, 24, 30,
36 h post-operatively, then every 12 h until post-operative day 4
and then daily until the study ceased on post-operative day 7. Alternatively,
the study ceased when the arterial line containing the
thermistor was removed prior to discharge from the intensive care
unit.
At the same time that cardiac performance was measured by
thermodilution structured clinical assessments were performed by
medical and nursing staff. Assessments were recorded on a standardised
form and placed in a sealed envelope after completion. Staff
could access all typical clinical and laboratory data, but were blinded
to parameters derived from FATD. The digital readout of the PiCCO
monitor was covered following each thermodilution so that the continuous
readout was not visible. Clinical assessment required estimation
of three parameters of cardiac performance: (a) an exact value
for CI and to categorise CI, (b) SVRI and (c) intravascular volume
status. Cardiac index was categorised as low (5). This is the same
classification used previously by Tibby et al. [2]. SVRI was defined
as low (1600).
Shann [4] defined normal SVRI as 800–1600 dyn s Ÿ1 cm Ÿ5 m Ÿ2 ,and
levels above and below this normal range were considered as being
high and low respectively, for the purposes of the study. Intravascular
volume status was classified as hypovolaemic, euvolaemic or overloaded.
In later analyses this was compared with the invasively obtained
value for GEDVI, which has been used as a surrogate for
volume status [5, 6]. Normal GEDVI for a comparable paediatric
population, which included 48 patients, was defined as 390–590 ml/
m 2 [7], which was considered as euvolaemia for the purposes of
comparison. Hypovolaemia was defined for study purposes as a
GEDVI less than 390 ml/m 2 and an overloaded state that greater than
590 ml/m 2 . GEDVI was chosen rather than intrathoracic blood volume
index (ITBVI) as it was felt to conceptualise “filling” of the heart
and thus perhaps be closer to what clinicians were considering in their
assessment of volume status. ITBVI is closely related to GEDVI, and
the relationship between them is linear; both have been used in
comparison to “preload” or “volume status” in adult and paediatric
studies [5, 6, 8]. Both GEDVI and ITBVI have been shown to increase
in response to volume loading [5, 6].
Following completion of the study clinical assessments were
compared to the corresponding measurements of cardiac performance
obtained with FATD. Data were analysed using SPSS version
10 (SPSS, Chicago, Ill., USA). Continuous data, available for
CI, were compared using a plot of mean vs. difference. The difference
was calculated by subtracting the CI assessed clinically
from that obtained invasively. The two-tailed Student’s t test was
used to determine whether this difference compared to zero was
significant. A p value of less than 0.05 was considered significant.
Invasive FATD numerical data were categorised according to the
above definitions prior to analysis. For categorical data the k statistic
as a measure of agreement was utilised. The level of agreement
between methods of assessment corresponds to the k value. A
value of 0–0.2 is poor, 0.2–0.4 fair, 0.4–0.6 moderate and a value of
1 indicates perfect agreement. The weighted k statistic was used to
allow for the presence of more than three categories, and discrepancies
of more than two categories contribute more heavily to the
statistic. Comparison was also made between the abilities of
nursing and medical staff of varying seniority to assess exact CI,
using Student’s t test. Invasive haemodynamic parameter data
satisfied assumptions for the normal distribution and were assessed
for accuracy and repeatability [9].
Results
Twenty children were enrolled into the study. Four patients
did not have data for comparison of methods of
determining cardiac performance (one because clinical
assessments were not made with the thermodilutions and
three lacked invasively obtained data). The resulting 16
patients had a median age of 15 months (interquartile
range 9.5–35) and median weight of 9 kg (7.3–13.6). No

570
Table 1 Cardiac lesions repaired
(AV atrioventricular,
ASD atrial septal defect, DORV
double-outlet right ventricle,
MAPCA major aortopulmonary
collateral artery, MPA main
pulmonary artery, MV mitral
valve, PA pulmonary artery,
RPA right pulmonary artery, RV
right ventricle, RVOT right
ventricular outflow tract, VSD
ventricular septal defect)
Cardiac lesion Operation No.
Complete atrioventricular septal defect AV septal defect repair 4
VSD Closure of ventricular septal defect 3
DORV, VSD, D-transposition Arterial switch, VSD closure, PA debanding 1
Tetralogy of Fallot
Repair of tetralogy of Fallot with transannular 2
patch
Mitral valve stenosis MV replacement 1
Pulmonary atresia, VSD, previous Complete repair of pulmonary atresia
1
unifocalisation of MAPCAs
with RV to PA conduit and closure of VSD
VSD with sub-pulmonary stenosis Closure of VSD with sub pulmonary resection 1
Secundum ASD Direct closure of ASD 1
Aortic valve stenosis and hypoplastic Aortic valve commisurotomy and RVOT 1
RVOT
patch augmentation
Pulmonary artery stenosis affecting
main pulmonary artery and right branch
pulmonary artery
Patch augmentation of MPA and RPA 1
patients were excluded on the basis of the post-operative
trans-oesophageal echocardiogram, with no residual shunt
or valvular insufficiency of haemodynamic significance
detected. The commonest lesion was atrioventricular
septal defect (n=4); the cardiac lesions and operation are
outlined in Table 1. In the three patients for whom there
was not invasively obtained data one had a fractured
thermistor wire, external to the patient, and another required
mechanical circulatory support with a left ventricular
assist device. The third, a 9-kg child having a
ventricular septal defect repair had a briefly ischaemic leg
following insertion of a 4-F femoral arterial line, but this
resolved with line removal. Otherwise no difficulties were
experienced with either the catheters or cardiac function
monitor, and no other complications occurred. All patients
survived to PICU discharge; the median length of
PICU stay was 3 days (interquartile range 2.7–4).
A total of 76 clinical estimations of CI, SVRI and
volume status were made by medical (56 occasions) and
nursing (20 occasions) staff.
Invasively measured CI showed a mean of
4.6 l min Ÿ1 m Ÿ2 (95% confidence interval 4.63€0.30, interquartile
range 3.8–5.2). In terms of the accuracy and
repeatability of invasive CI measurements, the measurement
error and 95% range was 0.26€0.52 l min Ÿ1 m Ÿ2 ,
and the coefficient of variation was 5.7%. A mean vs.
difference plot for absolute values for cardiac index obtained
clinically and invasively showed poor agreement
between clinically estimated and FATD measured values
for CI (95% range of agreement €2.7l/min; (Fig. 1).
Overall clinically estimated CI underestimated the invasively
obtained value; the mean difference was
0.71 l min Ÿ1 m Ÿ2 (95% confidence interval 0.37–1.04,
P

571
Fig. 2 Cardiac index (CI), invasive compared to clinical assessment.
This shows the categorical comparison of clinical assessment
with invasive assessment. For instance, the invasive assessment
category “normal-high” was clinically categorised as “low-normal”
12/40 and “low” 1/40. Note that for the lower CI groups that 11/20
(55%) were overestimated clinically
Fig. 3 Systemic vascular resistance index (SVRI), invasive compared
to clinical assessment. This shows the categorical comparison
of clinical assessment with invasive assessment of and shows a
trend towards overestimation of SVRI by clinical assessment. For
instance, in the invasive category “low,” 6/9 clinical assessments
were classified normal and 3/9 “low”
Fig. 4 Volume status, invasive compared to clinical assessment.
This shows the categorical comparison of clinical assessment with
invasive assessment of volume status by global end-diastolic volume
index (GEDVI) and a trend towards overestimation of volume
status by clinical assessment. For instance, in the invasive category
“hypovolaemic” 5/38 clinical assessments were correct whereas 33
were “euvolaemic,” or “overloaded”
A trend towards overestimation of SVRI by clinical assessment
was seen (Fig. 3), and there was poor agreement
with invasive assessments (k=0.12; Table 2).
Invasively measured GEDVI showed a mean of
427 ml/m 2 (95% confidence interval 427€38, interquartile
range 302–495). The measurement error and 95% range
was 45€88 ml/m 2 , with a coefficient of variation of 10%.
Volume status was overestimated clinically (Fig. 4) and
agreed poorly with invasive measures (k=0.09; Table 2).
Clinically 5 of 38 (13%) assessments in the hypovolaemic
state were correct. ITBVI showed a similar overestimation
to GEDVI when compared to volume status (k=0.09).

572
Discussion
This study used FATD as an invasive method of
haemodynamic assessment. This technique has been validated
in a comparable paediatric population following
cardiac surgery. Tibby et al. [3] validated FATD in ventilated
infants and children using indirect calorimetry and
the Fick principle, with good agreement (R 2 =0.994).
Similarly, Pauli et al. [10] demonstrated equivalence of
CI measurement by FATD and the Fick principle in
children. The femoral arterial route has been felt to offer
some benefits compared to pulmonary arterial (PA) lines
in terms of ease of insertion and a lower complication rate
[3]. PA lines are not often used in paediatric patients [3]
and are now primarily a research tool in this population
[11]. Nonetheless problems exist with the interpretation
of invasively obtained haemodynamic data, as numerous
studies in adult patients have demonstrated [12, 13, 14].
Importantly, despite the accuracy of invasive haemodynamic
data obtained with PA lines, an improvement in
outcome and cost benefit has not been shown and the use
of PA lines in adults is declining [15].
This study shows that clinicians were unable to categorise
parameters of cardiac performance correctly on the
basis of their clinical assessment. Clinical assessment was
inaccurate and agreed poorly with invasive determination
of CI, SVRI and volume status in this patient population.
This confirms a lack of agreement between clinical and
invasive determination of CI, as was shown previously by
Tibby et al. [2]. Furthermore, following our study the
clinician’s clinical inaccuracy extends to the assessment
of systemic vascular resistance and volume status, with
only 13% of hypovolaemic patients being detected clinically.
Difficulties with the clinical assessment of these
parameters have previously been suggested by studies
examining the value of capillary refill time and core-peripheral
temperature gradients, which have been shown to
inaccurately predict systemic vascular resistance [16, 17].
The universal use of phenoxybenzamine on bypass in
our patients could have made the clinical estimation of
cardiac index and volume status more challenging and
may have contributed to the lack of agreement between
assessment methods in our study. Phenoxybenzamine has
been used at our institution in order to provide a high CI
and vasodilated state which is felt to be more easily
managed than a low CI, vasoconstricted state. Consequently
few of the study patients had low CI measurements,
with the mean measured CI in the normal to high
range. Whilst this may limit the applicability of our study
to patients with low cardiac output states, it is important
to note that clinical assessment of cardiac output was also
inaccurate in the low CI range. In the instances when the
CI was less than the low-normal range, less than half of
the corresponding clinical assessments were correct, a
concerning tendency.
The study was limited by the small number of patients
and the small number of assessments conducted. Nevertheless
the direction of the disagreement between invasive
and clinical methods of assessment was noteworthy. CI
overall was underestimated, despite failing to detect the
majority with low CI. Systemic vascular resistance and
volume status were overestimated clinically. The failure
to assess volume status correctly was marked with just
13% of “hypovolaemic” and 11% of “overloaded” patient
assessments being categorised correctly. The normal
values for GEDVI obtained from the literature [7], although
from similarly sized patients, were not from patients
with volume loaded hearts as were some of our
post-operative patients; thus the apparent overestimation
of volume status assessed clinically may relate to differences
between the reference and study populations.
Two of the 20 patients experienced technical problems
related to FATD. One patient had an ischaemic leg and
one thermistor wire fractured external to another patient.
Difficulties with FATD in the paediatric population are
rarely described in the literature. However, femoral arterial
ischaemia is a recognised problem. In a prospective
review of 330 arterial cannulations only 3 (1%) ischaemic
limbs were observed [18], as opposed to our 5% incidence.
In a series of ten neonates being monitored by
FATD, one 4.3-kg baby (10% incidence) developed an
ischaemic leg on day 2 of FATD monitoring, which resolved
with removal of the line [5]. The ability to detect
adverse incidents in our study was limited by the small
number of patients involved; nevertheless limb ischaemia
is a potential complication of this method of assessment
and may be a factor in perceived reluctance to use this
technology, especially in neonates. Apart from this issue
the FATD technique was found to be straightforward.
An increasing number of parameters of cardiac performance
when assessed clinically have now been shown
to agree poorly compared with their invasive determination
in the paediatric population. The inaccuracy of
clinical assessment for three parameters of cardiac performance
is troubling as this is accepted standard care.
Particularly concerning is the detection of only 45% of
the more critical patients with a CI less than the lownormal
range and only 13% of the hypovolaemic patients.
Nonetheless mortality following biventricular cardiac repairs
in children is low, and clinical assessment deserves
some credit. Invasive haemodynamic monitoring in the
form of FATD does have potential complications. Although
the improved accuracy could potentially translate
into better outcomes, this has not been the case with PA
lines in adults, and further evaluation of FATD in the
paediatric population should be undertaken cautiously.
Acknowledgements These findings were previously presented at
the 4th World Congress of Pediatric Intensive Care, 8–12 June
2003, Boston, USA [19].

573
References
1. Hoffman TM, Wernovsky G, Atz AM,
Kulik TJ, Nelson DP, Chang AC,
Bailey JM, Akbary A, Kocsis JF,
Kaczmarek R, Spray TL, Wessel DL
(2003) Efficacy and safety of milrinone
in preventing low cardiac output syndrome
in infants and children after
corrective surgery for congenital heart
disease. Circulation 107:996–1002
2. Tibby SM, Hatherill M, Marsh MJ,
Murdoch IA (1997) Clinicians’ abilities
to estimate cardiac index in ventilated
children and infants. Arch Dis Child
77:516–518
3. Tibby SM, Hatherill M, Marsh MJ,
Morrison G, Anderson D, Murdoch IA
(1997) Clinical validation of cardiac
output measurements using femoral
artery thermodilution with direct Fick
in ventilated children and infants. Intensive
Care Med 23:987–991
4. Shann F (2001) Drug doses, 11th edn.
Collective
5. Schiffmann H, Erdlenbruch B, Singer
D, Singer S, Herting E, Hoeft A, Buhre
W (2002) Assessment of cardiac output,
intravascular volume status and extravascular
lung water by transpulmonary
indicator dilution in critically ill neonates
and infants. J Cardiothorac Vasc
Anesth 16:592–597
6. Michard F, Alaya S, Zarka V, Bahloul
M, Richard C, Teboul JL (2003) Global
end-diastolic volume as an indicator of
cardiac preload in patients with septic
shock. Chest 124:1900–1908
7. Kozlik-Feldmann R, Konert M, Freund
M, Netz H (1998) Normal values for
distribution volumes of less invasive
circulation monitoring by double indicator
measurement in paediatric intensive
care. Z Kardiol 87:762
8. Sakka SG, Ruhl CC, Pfeiffer UJ,
Beale R, McLuckie A, Reinhart K,
Meier-Hellmann A (2000) Assessment
of cardiac preload and extravascular
lung water by single transpulmonary
thermodilution. Intensive Care Med
26:180–187
9. Peat J, Mellis C, Williams K, Xuan W
(2001) Health science research: a
handbook of quantitative methods. Allen
and Unwin, Crows Nest–
10. Pauli C, Fakler U, Genz T, Hennig M,
Lorenz H-P, Hess J (2002) Cardiac
output determination in children:
equivalence of the transpulmonary
thermodilution method to the direct
Fick principle. Intensive Care Med
28:947–952
11. Ravishankar C, Tabbutt S, Wernovsky
G (2003) Critical care in cardiovascular
medicine. Curr Opin Pediatr 15:443–
453
12. Richard C, Warszawski J, Anguel N,
Deye N, Combes A. Harnottd D,
Boulain T, Lefort Y, Fartoukh M, Baud
F, Boyer A, Brochard L, Teboul J-L
(2003) Early use of the pulmonary artery
catheter and outcomes in patients
with septic shock and acute respiratory
distress syndrome. JAMA 290:2713–
2720
13. Fowler RA, Cook DJ (2003) The arc of
the pulmonary artery catheter. JAMA
290:2732–2734
14. Sandham JD, Hull RD, Brant RF,
Knox L, Pineo GF, Doig CJ, Laporta
DP, Viner S, Passerini L, Devitt H,
Kirby A, Jacka M (2003) A randomized,
controlled trial of the use of
pulmonary-artery catheters in high-risk
surgical patients. N Engl J Med 348:5–
14
15. Morgan TJ (2004) Life without the PA
catheter. Crit Care Resuscitation 6:9–12
16. Tibby SM, Hatherill M, Murdoch IA
(1999) Capillary refill and core-peripheral
temperature gap as indicators of
haemodynamic status in paediatric intensive
care patients. Arch Dis Child
80:163–166
17. Butt W, Shann F (1991) Core-peripheral
temperature gradient does not predict
cardiac output or systemic vascular
resistance in children. Anaesth Intensive
Care 19:84–87
18. Smith-Wright DL, Thomas PG, Lock
JE, Egar MI, Fuhrman BP (1984)
Complications of vascular catheterization
in critically ill children. Crit Care
Med 12:1015–1017
19. Egan J, Festa M, Winlaw D, Cole A,
Gillis J, Nunn G (2003) Comparison of
clinical and invasive assessments of
cardiac performance following paediatric
cardiac surgery. Pediatr Crit Care
Med 4 [Suppl]:A101