Feature Article—Continuing Medical Education

Predicted versus measured energy expenditure by continuous,

online indirect calorimetry in ventilated, critically ill children
during the early postinjury period*
Jose Luis Vazquez Martinez, MD, PhD, Paloma Dorao Martinez-Romillo, MD, PhD,
Jesus Diez Sebastian, MD, Francisco Ruza Tarrio, MD, PhD

LEARNING OBJECTIVES
On completion of this article, the reader should be able to:
1. Identify key facets of the overfeeding syndrome in critically ill infants and children.
2. Identify the major formulas used to estimate energy expenditure in critically ill infants and children.
3. Determine energy expenditure using the Caldwell-Kennedy equation.
Dr. Vazquez Martinez, Dr. Martinez-Romillo, and Dr. Ruza Tarrio are the recipients of grant 96/0835 from the Fondo de Investigación
Sanitaria de la Seguridad Social. Dr. Diez Sebastian has disclosed that he has no financial relationships with commercial companies
pertaining to this educational activity.
Visit the Pediatric Critical Care Medicine Online website (www.pccmjournal.com) for information on obtaining continuing medical
education credit.

Objective: Compare the energy expenditure, predicted by anthropometric
equations, with that measured by continuous on-line indirect calorimetry in
ventilated, critically ill children during the early postinjury period.
Design: Prospective, clinical study.
Setting: Pediatric intensive care unit of a pediatric university hospital.
Patients: A total of 43 ventilated, critically ill children during the first
6 hrs after injury.
Interventions: An indirect calorimeter was used to continuously mea-
sure the energy expenditure for 24 hrs.
Measurements and Main Results: Clinical data collected were age,
gender, actual and ideal weight, height, and body surface. Nutritional
status was assessed by Waterlow and Shukla Index. Severity of illness
was determined by Pediatric Risk of Mortality, Physiologic Stability Index,
and Therapeutic Intervention Scoring System. Energy expenditure was
measured (MEE) by continuous on-line indirect calorimetry for 24 hrs.
Predicted Energy Expenditure (PEE) was calculated using the Harris-
Benedict, Caldwell-Kennedy, Schofield, Food and Agriculture/World
Health Organization/United Nation Union, Maffeis, Fleisch, Kleiber, Dreyer,
and Hunter equations, using the actual and ideal weight. MEE and PEE
were compared using paired Student’s t-test, linear correlation (r), intra-
class correlation coefficient (pI), and the Bland-Altman method. Mean
MEE resulted in 674 ⴞ 384 kcal/day. Most of the predictive equations
overestimated MEE in ventilated, critically ill children during the early
postinjury period. MEE and PEE differed significantly (p < .05) except
when the Caldwell-Kennedy and the Fleisch equations were used. r2
ranged from 0.78 to 0.81 (p < .05), and pI was excellent (>.75) for the
Caldwell-Kennedy, Schofield, Food and Agriculture/World Health Organi-
zation/United Nation Union, Fleisch, and Kleiber equations. The Bland-
Altman method showed poor accuracy; the Caldwell-Kennedy equation
was the best predictor of energy expenditure (bias, 38 kcal/day; preci-
sion, ⴞ179 kcal/day). The accuracy in the medical group was higher (pI
range, .71–.94) than in surgical patients (pI range, .18 –.75).
Conclusions: Predictive equations do not accurately predict energy
expenditure in ventilated, critically ill children during the early postinjury
period; if available, indirect calorimetry must be performed. (Pediatr Crit
Care Med 0; 0:●●●–●●●)
KEY WORDS: continuous indirect calorimetry; critically ill children; early
postinjury period; energy expenditure; nutrition; predictive equations

T he outcome of patients admit-
ted to the intensive care unit
(ICU) improves when their nu-
tritional requirements are ful-
filled. A precise evaluation of caloric re-
quirements is essential to set up
individualized support, thus avoiding com-
plications related to over- and underfeed-
ing. Overfeeding has been associated with
increased CO2 production, respiratory fail-
ure, hyperglycaemia, and fat deposits in the
liver, thus impairing recovery; underfeed-
ing can lead to malnutrition, muscle weak-
ness, and impaired immunity. In both
cases, mortality and morbidity are in-
creased (1).
Indirect calorimetry is a reliable and
noninvasive method that determines en-
ergy expenditure at bedside in critically
ill patients, including children, although
it is expensive, time-consuming, and re-

*See also p. 96.
Staff, Pediatric Intensive Care Unit, Hospital
Ramón y Cajal, Madrid, Spain; Associate Research
Doctor, Pediatric Intensive Care Unit, Hospital Infan-
til La Paz, Madrid, Spain (JLVM); Staff, Pediatric
Intensive Care Unit, Hospital Infantil La Paz, Madrid,
Spain (PDM-R); Staff, Research and Quality Unit,
Preventive Medicine Department, Hospital Universi-
tario La Paz, Madrid; Associate Professor, Univer-
sidad Autónoma de Madrid, Spain (JDS); and Chief, Pediátricos, Servicio de Pediatría, Hospital Ramón y
Pediatric Intensive Care Unit, Hospital Infantil La Cajal, 28034 Madrid, Spain. E-mail: jvazquez.
Paz, Madrid, Spain; Professor of Pediatrics, Univer- hrc@salud.madrid.org, dulari@wanadoo.es
sidad Autónoma de Madrid, Spain (FRT).
Copyright © 2004 by the Society of Critical Care
Supported, in part, by grant 96/0235 from the
Medicine and the World Federation of Pediatric In-
Fondo de Investigación Sanitaria de la Seguridad
tensive and Critical Care Societies
Social (FISS).
Address requests for reprints to: José Luis DOI: 10.1097/01.PCC.0000102224.98095.0A
Vázquez, MD, PhD, Unidad de Cuidados Intensivos

Pediatr Crit Care Med 2004 Vol. 5, No. 1 19

quires trained personnel. Several meta-
bolic monitors have been validated in
vitro and in vivo (2, 3). However, it is
known that there are operational and
technical limitations in the use of the
indirect calorimetry in the ICU (high in-
spired oxygen concentrations and air
leaks). Moreover, ICU care (e.g., endotra-
cheal aspiration and daily toiletry) and
clinical patient conditions (hemodynamic
instability, agitation, and fever) can also
reduce the accuracy of indirect calorim-
etry when it is performed during a short
period of time, leading to contradictory
results. Thus, longer indirect calorimetry
studies, such as continuous on-line indi-
rect calorimetry performed for 24 hrs,
have shown that mechanically ventilated
adult patients are in an hypermetabolic
state during the early postinjury period
(4). Different conclusions have been
reached when other investigators have
performed shorter indirect calorimetry
studies (5–7). Metabolic response in chil-
dren seems to be faster (8). Some authors
have shown that after surgery is per-
formed on neonates, there is only a tran-
sient increase in resting energy expendi-
ture (9). However, other studies have
failed to show a rise in energy expendi-
ture during the early postinjury period in
infants or older children (10, 11).
When indirect calorimetry is not avail-
able, the energy expenditure is calculated
using predictive equations. Some authors
have modified the predictive formulas
based on anthropometric parameters
(gender, weight, and height) and adding
clinical stress factors (12, 13). However,
the accuracy of these modified equations
remains unknown for the early postinjury
period because patients under study were
in different phases of the stress response
with variable times elapsed since injury.
Furthermore, some patients were fasting,
whereas other patients were receiving
parenteral or enteral nutrition. All these
factors could lead to contradictory re-
sults.
There is recent interest in setting up
precocious nutrition, even in the imme-
diate postoperative period. This requires
assessment of energy expenditure during
this early postinjury phase. Letton et al.
(14) suggested that in the absence of cal-
orimetric measurements, predicted basal
metabolic rate should be used to estimate
caloric delivery during the early postop-
erative period, although, to our knowl-
edge, no studies exist that have moni-
tored the accuracy of predictive formulas
20
in critically ill children during the early
postoperative period.
The purpose of the present study was
to compare the accuracy of available
equations to assess nutritional demands
in ventilated, critically ill children during
the first 24 hrs after pediatric ICU admis-
sion. The objective was to contrast the
accuracy of predictive formulas with in-
direct calorimetry measurements for pre-
dicting energy expenditure during the
early postinjury period. This was carried
out by comparing the energy expenditure
values measured by continuous, 24-hr
indirect calorimetry (measured energy
expenditure, MEE) with values obtained
using the current anthropometric equa-
tions (predicted energy expenditure,
PEE).
MATERIALS AND METHODS
The study protocol was approved by the La
Paz Children’s Hospital Ethics Committee,
and informed consent was obtained from the
parents.
Patients. Mechanically ventilated, critically
ill children who were admitted to the pediatric
intensive care unit (PICU) at La Paz Children’s
Hospital, Madrid, Spain, were prospectively
studied within the first 6 hrs after injury. This
study was conducted between January 1997
and June 2000. Patients requiring mechanical
ventilation for ⬎24 hrs were included in the
study. All patients who were included were
ventilated by Servo (900C and 300, Siemens-
Elema, Solna, Sweden), and Ohmeda-Advent
(Louisville, KY) ventilators. Exclusion criteria
were ventilation with high fractional inspired
oxygen concentration (FiO2 ⬎0.6), the pres-
ence of significant air leaks (with significance
defined as the difference between inspired and
expired tidal volume amounting to ⬎10%;
both volumes are continuously measured by
the ventilator) (15–17), and the use of contin-
uous flow or administration of anesthetic
gases.
Data Collecting Procedures. The following
anthropometric parameters were identified:
gender, age, actual weight, ideal weight,
height, and body surface area. In cases in
which weighing the patient was impossible,
the weight informed by the parent was taken
into consideration. Ideal weight was defined as
the weight for 50th percentile of the actual
height of each patient (18). Body surface was
obtained with the Haycock method (19). Acute
and chronic nutritional status was evaluated
according to the Waterlow Nutritional Index
(20) and the Shukla Nutritional Index (21),
respectively. Other clinical data collected in-
cluded a) primary reason for admission and b)
time elapsed from injury to the initiation of
metabolic monitoring. The severity of illness
was assessed using the Pediatric Risk of Mor-
tality (22) and the Physiologic Stability Index
(23), and the amount of care was assessed
using the Therapeutic Intervention Scoring
System (24).
Predicted Energy Expenditure. The PEE
was calculated using both actual and ideal
weight in the following formulas: the Harris-
Benedict equations (25), the Caldwell-
Kennedy equation (26), the Schofield equa-
tions (27), the Food and Agriculture/World
Health Organization/United Nation Union
equations (28), the Maffeis equations (29), the
Fleisch equation (19), the Kleiber equation
(30), the Dreyer equation (31), and the Hunter
equation (32). Results are expressed in kilo-
calories per day. Table 1 reflects these equa-
tions.
We calculated the metabolic index accord-
ing to Weissman and Kemper: [(MEE ⫺ PEE)/
PEE ⫻ 100] (33). The metabolic index repre-
sents the relation between MEE and PEE,
expressed as a percentage; a negative value
(⬍0) means that PEE overestimated the en-
ergy expenditure, and a positive value (⬎0)
means that PEE underestimated the energy
expenditure. Other authors use the metabolic
index with the following criteria: whenever
MEE is ⬎110% of predicted metabolic rate of
healthy children (in our series, the Shofield
equations), patients are defined as hypermeta-
bolic, and when ⬍90%, they are defined as
hypometabolic (34).
Measured Energy Expenditure and Respi-
ratory Quotient. The MEE and respiratory
quotient (V̇CO2/V̇O2) was obtained by indirect
calorimetry, using a portable metabolic unit
(Deltatrac II, MBM-200, Datex, Helsinki, Fin-
land). The Deltatrac is an open-system indi-
rect calorimetry device. This metabolic moni-
tor has been widely validated and tested for
accuracy and reproducibility in vitro and in
vivo (3, 35), which enables its clinical use in
the PICU. It measures V̇O2 and V̇CO2 from gas
exchange measurements and calculates the
MEE and respiratory quotient. It is based on
the Weir equation (36): MEE (kcal/day) ⫽
3.941 V̇O2(L/min) ⫹ 1.106 V̇CO2(L/min) ⫺ 2.17
UN(g/day), where MEE is measured energy
expenditure, V̇O2 is oxygen consumption, V̇CO2
is dioxide carbon production, and UN is uri-
nary nitrogen. The error derived from ignor-
ing urinary nitrogen is ⬍2% (37). Therefore
we did not measure this parameter in our
patients.
We calibrated the metabolic monitor ac-
cording to the operator’s manual. Gas calibra-
tion was accomplished before each measure-
ment to assess the accuracy of V̇O2 and V̇CO2
analyzers with a bottle containing a gas mix-
ture of 95% O2 and 5% CO2, 30 mins after the
device was warmed up. Pressure calibration
was performed out every 6 months, setting the
pressure value according to the actual baro-
metric pressure. Flow calibration was per-
formed every 2 months by burning 5 mL of
ethanol as per the alcohol burning test set-up
(for specifications see Deltatrac operator’s
manual).
Pediatr Crit Care Med 2004 Vol. 5, No. 1
Table 1. Standard equations used to predict energy expenditure (PEE) in ventilated, critically ill tient’s needs compared with MEE. Variability
children during the early postinjury period is measured by calculating the SD of individual
differences for all patients. It represents the
Harris-Benedict equation (kcal/day) reproducibility with which one can predict
Males: 66.4730 ⫹ (5.0033 ⫻ height) ⫹ (13.7516 ⫻ weight) ⫺ (6.7550 ⫻ age) energy expenditure in an individual patient, so
Females: 655.0955 ⫹ (1.8496 ⫻ height) ⫹ (9.5634 ⫻ weight) ⫺ (4.6756 ⫻ age) the range of values corresponding to ⫾1 SD of
Caldwell-Kennedy equation (kcal/day): 22 ⫹ (31.05 ⫻ weight) ⫹ (1.16 ⫻ age)
the mean difference (bias) between both meth-
Schofield equations (Mj/day) (1 kcal ⫽ 4.186 kJ)
⬍3 yrs old ods represents the performance precision, or
Boy: (0.0007 ⫻ weight) ⫹ (6.349 ⫻ height) ⫺ 2.584 the consistency of the predictive formula.
Girl: (0.068 ⫻ weight) ⫹ (4.281 ⫻ height) ⫺ 1.730 All analyses were made using the Microsoft
3–10 yrs old Excel program (Version 97, Microsoft, Seattle,
Boy: (0.082 ⫻ weight) ⫹ (0.545 ⫻ height) ⫹ 1.736 WA), the Statistical Package for the Social
Girl: (0.071 ⫻ weight) ⫹ (0.677 ⫻ height) ⫹ 1.553 Science for Windows software package (SPSS
10–18 yrs old 8.0, 1995, SPSS) and the EPI-INFO 6.04 sta-
Boy: (0.068 ⫻ weight) ⫹ (0.574 ⫻ height) ⫹ 2.157 tistical package (EPI-INFO 6, 1994, Center for
Girl: (0.035 ⫻ weight) ⫹ (1.948 ⫻ height) ⫹ 0.837
FAO/WHO/UNU equations
Diseases Control, Atlanta, GA). Technical sup-
⬍3 yrs old port was provided by the Research and Quality
Boy (kcal/day): (60.9 ⫻ weight) ⫺ 54 Unit from La Paz Children’s Hospital.
Girl (kcal/day): (61 ⫻ weight) ⫺ 51
3–10 yrs old (1 kcal ⫽ 4.186 kJ)
Boy (kJ/day): (95 ⫻ weight) ⫹ 2071 RESULTS
Girl (kJ/day): (94 ⫻ weight) ⫹ 2088
10–18 yrs old A total of 43 patients (25 males and 18
Boy (kcal/day): (16.6 ⫻ weight) ⫹ (77 ⫻ height) ⫹ 572
Girl (kcal/day): (7.4 ⫻ weight) ⫹ (482 ⫻ height) ⫹ 217
females) were studied. All patients were
Maffeis equations (kJ/day) (1 kcal ⫽ 4.186 kJ) studied within the first 6 hrs after elective
Boy: (28.6 ⫻ weight) ⫹ (23.6 ⫻ height) ⫺ (69.1 ⫻ age) ⫹ 1287 or emergency admission to the PICU.
Girl: (35.8 ⫻ weight) ⫹ (15.6 ⫻ height) ⫺ (36.3 ⫻ age) ⫹ 1552 There were 35 surgical patients and eight
Fleisch equation (kcal/day) medical patients distributed as follows:
Males
1–12 yrs old: 24 ⫻ BSA ⫻ (54 ⫺ 0.885 ⫻ age) 30 (69.8%) after postoperative cardiac
13–19 yrs old: 24 ⫻ BSA ⫻ {42.5 ⫺ [0.643 ⫻ (age ⫺ 13)]} surgery, two (4.7%) after neurosurgical
Females injury, two (4.7%) after liver transplant,
1–10 yrs old: 24 ⫻ BSA ⫻ [54 ⫺ (1.045 ⫻ age)] one (2.3%) after thoracic surgery, and
11–19 yrs old: 24 ⫻ BSA ⫻ {42.5 ⫺ [0.778 ⫻ (age ⫺ 11)]}
Kleiber equation (kcal/day) PEE ⫽ 70 ⫻ weight0.75
eight (18.6%) with medical problems
Dreyer equation (kcal/day) (four sepsis, three respiratory illness, and
Males: weight1/2/(0.1015 ⫻ age0.1333) one non-neurosurgical brain injury). All
Females: weight1/2/(0.1127 ⫻ age0.1333) patients were on mechanical ventilation
Hunter equation (kcal/day) PEE ⫽ 22 ⫻ weight (controlled or intermittent mandatory
FAO/WHO/UNU, Food and Agriculture/World Health Organization/United Nation Union; BSA, body
ventilation, according to the routine
surface area in m2. management of our PICU). The range of
FiO2 that was used was 40% to 60%. Only
ten (23.2%) of the patients were venti-
We set the metabolic cart to Respiratory ten fractionated indirect calorimetry studies lated with uncuffed endotracheal tube,
Mode (all patients were on mechanical venti- and one long indirect calorimetry study (of but no patients had a significant air leak.
lation) and selected the desired flow range 1,440 minute-to-minute measurements) were None of the patients received enteral
according to the patient’s size. Then we con- performed. or parenteral nutritional support. They
nected the mixing chamber inlet of the mon- Statistical Analysis. Descriptive variables received glucosaline solutions for daily
itor to the expiratory outlet of the respirator are reported as mean ⫾ 1 SD, and the normal- requirements, and some also received
and the inspiratory sampling line in the in- ity of the distribution was confirmed using the fresh frozen plasma to correct hemato-
spiratory tubing of the respirator on the hu- Kolmogorov-Smirnov (Lilliefors modification)
logic disorders. Table 2 summarizes de-
midifier outlet. The monitor uses an algo- test. Paired Student’s t-tests were used to look
rithm to delete the artefacts caused by for significant differences between MEE and mographic and clinical characteristics as
fluctuating or severely unstable readings (e.g., PEE. A p value ⱕ.05 was considered statisti- well as metabolic data. The use of ideal
those created during airway suctions or arte- cally significant. Linear correlation coefficient weight in the predictive equations always
rial hypotension). We started metabolic mea- (r) determined the strength of the relationship increased the magnitude of PEE. Thirty-
surements, measuring the energy expenditure between MEE and PEE. Intraclass correlation one (72%) patients presented some de-
continuously during 24 hrs. This MEE value coefficient (pI) (39) was calculated to measure gree of acute or chronic malnutrition
represents the average of the 1,440 measure- the reproducibility of the analyzed methods. A (Fig. 1).
ments (one per minute) executed on each pa- pI value ⬎.75 means excellent reproducibility, The surgical patients had lower Water-
tient for 24 hrs. We calculated the intrasubject and ⬍.4 indicates poor reproducibility. We low Nutritional Index and lower Shukla
percentage variation of MEE (38). We also also determined the bias and the precision for
Nutritional Index values than the medical
performed ten short indirect calorimetry mea- predictive formulas vs. indirect calorimetry
surements that focused on the first hours after using the Bland and Altman (40) comparison group (89.08 ⫾ 13.1% vs. 94.28 ⫾ 11.6%
admission: 30, 60, 90, 120, 150, and 180 mins procedure. The mean difference between both and 84.48 ⫾ 13.4% vs. 96.31 ⫾ 17.1%,
after the start of indirect calorimetry measure- methods represents the performance bias. On respectively). The mean Pediatric Risk of
ments and at 4, 6, 18, and 24 hrs (all these average, the bias indicates how far off an in- Mortality score was seven, the mean
short measures were 30 mins). To summarize, dividual will be in predicting any given pa- Physiologic Stability Index score was six,

Pediatr Crit Care Med 2004 Vol. 5, No. 1 21

Table 2. Demographics and clinical characteris- day and a precision (SD of bias) of 179
tics of patients, with the average values of pre- kcal/day (Fig. 4).
dicted and measured energy expenditure (PEE Other Bland-Altman comparison re-
and MEE) in ventilated, critically ill children dur-
sults are shown in Table 4, likewise indi-
ing the early postinjury period
cating a range of mean bias and variabil-
Variables Mean ⫾ SD ity.
We also tested accuracy of the predic-
Clinical tive equations in surgical patients vs.
Age, yr 4.21 ⫾ 3.67 medical patients, using the intraclass
Weight, kg 15.94 ⫾ 11.97 correlation coefficient (Table 5). This co-
Ideal weight, kg 17.15 ⫾ 29.16
Height, cm 96.94 ⫾ 26.57 efficient was always higher in the group
Body surface area, m 2
0.624 ⫾ 0.32 of medical patients (pI range, .71–.94)
Time after admission, hrs 2.7 ⫾ 4.7 than in the surgical group (pI range, .18 –
PRISM score 7⫾5 .75). The Fleisch equation showed the
PSI score 6⫾5
TISS score 36 ⫾ 7
highest pI in the surgical group (pI ⫽ .73
Figure 1. Distribution of patients by the Water- and .75, using the actual or ideal weight,
Waterlow ratio, % 90.05 ⫾ 13.9
low and Shukla nutritional stages. Note that only respectively), and the Caldwell-Kennedy
Shukla ratio, % 86.69 ⫾ 15.4
PEE, kcal/day 12 patients presented a normal nutritional status.
equation, using actual and ideal weight,
Actual weight Acute malnutrition was present in 26 of 31 pa-
resulted in the highest pI in the medical
Harris-Benedict 848 ⫾ 256.4 tients, who also presented chronic malnutrition.
Caldwell-Kennedy 625.7 ⫾ 403.2 group (pI ⫽ .94 and .92, respectively).
Schofield 761.7 ⫾ 320.8
FAO/WHO/UNU 740.7 ⫾ 338.8 DISCUSSION
Maffeis 855.9 ⫾ 188.1
Fleisch 726.4 ⫾ 298.8 than surgical patients (570 ⫾ 194 kcal/ This study evaluated the accuracy of
Kleiber 536.8 ⫾ 285.4 day; 40 ⫾ 7 kcal/kg/day; n ⫽ 8), although
Dreyer 963.8 ⫾ 221.7 predictive equations to estimate the en-
Hunter 350.8 ⫾ 263.5
this was not statistically significant. The ergy expenditure in critically ill children
Ideal weight metabolic index outlined important vari- during the early postinjury period and
Harris-Benedict 862.6 ⫾ 252 ation, ranging from ⫺33.2% to ⫹80.7% compared these values with those mea-
Caldwell-Kennedy 666.4 ⫾ 387.5 in the equations when the actual weight
Schofield 775.6 ⫾ 318.2 sured using continuous 24-hr indirect
was used and from ⫺37.5% to ⫹85.7% calorimetry measurement. To the best of
FAO/WHO/UNU 798.8 ⫾ 305.6
Maffeis 865.2 ⫾ 185.3 when the ideal weight was used, although our knowledge, there are no previous
Fleisch 726.7 ⫾ 317.1 it indicated that most of the predictive studies validating the use of predictive
Kleiber 571.6 ⫾ 269.7 equations overestimated the energy ex-
Dreyer 1013.7 ⫾ 195.4
equations during the early postinjury pe-
penditure. Only the Caldwell-Kennedy riod for children in whom energy expen-
Hunter 375 ⫾ 253.6
MEE, kcal/day 673.8 ⫾ 384 and the Fleisch equations (using both diture was continuously measured
RQ 0.78 ⫾ 0.01 actual and ideal weight) presented a met- throughout the first 24-hr postinjury pe-
Schofield MI, % 88.5 abolic index range of ⫾10%. riod.
Intrasubject variation for MEE, We found statistically significant dif-
%
This study included 43 patients. The
ferences between MEE and PEE using all population was mostly children admitted
Medical patients 18.35
Surgical patients 14.94 predictive methods except for the Cald- after elective surgery because of the re-
well-Kennedy and Fleisch equations (ei- strictive inclusion and exclusion criteria
PRISM, Pediatric Risk of Mortality; PSI, Phi- ther with actual or ideal weight). We also used to achieve a homogeneous popula-
syologic Stability Index; TISS, Therapeutic Inter- found a good linear correlation (Table 3) tion. Because the period of time for pa-
vention Scoring System; FAO/WHO/UNU, Food resulting in the highest correlation coef- tient study initiation was short (median,
and Agriculture/World Health Organization/ ficient when the Hunter equation with
United Nation Union; RQ, respiratory quotient;
1 hr), we managed to study the energy
actual weight (r2 ⫽ .81; p ⫽ .001) and the expenditure during the early postinjury
MI, metabolic index.
Fleisch equation with ideal weight (r2 ⫽ period. We chose this well-defined period
.81; p ⬍ .001) were used. The best repro- of time because we wanted to ensure that
and the mean Therapeutic Intervention ducibility of formulas (measured by the all our patients were in the same postin-
Scoring System score was 36. intraclass correlation) was obtained when jury phase (it is known that the energy
The mean MEE was 673.8 ⫾ 384 kcal/ the Caldwell-Kennedy, Schofield, Food expenditure varies according to the stress
day (47 ⫾ 19 kcal/kg/day), and the mean and Agriculture/World Health Organiza- response phase (41)). Second, we wanted
respiratory quotient was 0.78 ⫾ 0.01. We tion/United Nation Union, Fleisch, and to assess nutritional requirements during
compared energy expenditure measured Kleiber equations were used, with the the early postinjury period because there
with short indirect calorimetry studies highest value once again corresponding is an increasing interest in knowing how
and continuous 24-hr indirect calorime- to the Caldwell-Kennedy equation (pI ⫽ to determine nutritional support needs in
try using paired Student’s t-test (Fig. 2). .89; p ⬍ .0001) (Fig. 3). critically ill patients during this time (42,
The mean intrasubject MEE variation However, the Bland-Altman compari- 43).
was 15.57 %, which was similar for med- son of MEE and PEE calculated with the All the evaluated equations are based
ical and surgical patients. Medical pa- Caldwell-Kennedy equation (the highest on anthropometric parameters that are
tients showed a higher MEE (926 ⫾ 734 intraclass correlation coefficient in our measurable in the PICU, although accu-
kcal/day; 54 ⫾ 3 kcal/kg/day; n ⫽ 35) study) showed a mean bias of ⫺38 kcal/ rately obtaining some measures in criti-

22 Pediatr Crit Care Med 2004 Vol. 5, No. 1

 expressing them in kcal/day, in like man-
ner to the original predictive equations,
and not in kcal/kg/day or kcal/m2/day,
which would imply the generation of a
new variable (energy expenditure/body
size) different from the original measure
(energy expenditure).
This study shows that usual predictive
methods fail to reliably estimate energy
expenditure in ventilated, critically ill
children during the early postinjury pe-
riod. Indirect calorimetry was the only
reliable way to determine energy expen-
diture for individual patients in this pop-
ulation. Few studies have measured en-
ergy expenditure during the early
postinjury period, and none have vali-
dated the predictive equations during this
period in critically ill children. Most re-
ports have studied energy expenditure in
later postinjury stages, usually critically
ill children with a length of stay in the
PICU ⬎1 day (4, 5, 15, 44). Aiming to
improve the accuracy of the MEE, we
performed continuous on-line indirect
Figure 2. Daily variation of the energy expenditure (EE) measured by short indirect calorimetry studies
during the first 24 hrs postinjury. calorimetry for 24 hrs because there are
no established standards that define the
period during which the test must be
Table 3. Comparison between predicted and measured energy expenditure using 24-hr indirect performed (45, 46). Shorter indirect cal-
calorimetry: Paired t-test and linear correlation orimetry studies could be suitable for
clinical caring (47), although not for val-
Paired Mean Differencesa idation studies because energy expendi-
Equation (kcal/day) (mean ⫾ SD) p Value r2 p Value ture can change during the day (38).
We did not add stress-related or level
Harris-Benedict 162.9 ⫾ 236.5 .001 .62 ⬍.0001
IdealHarris-Benedict 177.3 ⫾ 240.8 .001 .6 ⬍.0001
of activity-related factors to the equa-
Caldwell-Kennedy ⫺39.12 ⫾ 185.4 NS .79 ⬍.0001 tions. When other authors did so, the
IdealCaldwell-Kennedy 0.45 ⫾ 188.8 NS .75 ⬍.0001 energy expenditure was overestimated,
Schofield 96.74 ⫾ 186 .01 .75 ⬍.0001 showing unacceptable results (12). We
IdealSchofield 109.2 ⫾ 189.1 .005 .75 ⬍.0001 have only modified the original equations
FAO/WHO/UNU 82.7 ⫾ 200.8 .038 .72 ⬍.0001
IdealFAO/WHO/UNU 133.7 ⫾ 198.2 .001 .72 ⬍.0001 by using the ideal weight. There is little
Maffeis 181.4 ⫾ 232.8 ⬍.0001 .77 ⬍.0001 evidence in the literature supporting the
IdealMaffeis 190.5 ⫾ 234.7 ⬍.001 .75 ⬍.0001 idea that ideal weight improves the PEE
Fleisch 58.6 ⫾ 199.7 NS .7 ⬍.0001 (48). In the present study, the use of ideal
IdealFleisch 61.4 ⫾ 169.1 NS .81 ⬍.0001
Kleiber ⫺130.5 ⫾ 178.9 .001 .79 ⬍.0001
weight resulted in a higher PEE value.
IdealKleiber ⫺96.7 ⫾ 185 .01 .79 ⬍.0001 This was due to the fact that many pa-
Dreyer 296.5 ⫾ 219 ⬍.0001 .75 ⬍.0001 tients suffered some degree of malnutri-
IdealDreyer 343.2 ⫾ 229.5 ⬍.0001 .75 ⬍.0001 tion, and, therefore, their ideal weight
Hunter ⫺317.7 ⫾ 180.5 .001 .81 .001 was always higher than their actual
IdealHunter ⫺293.6 ⫾ 210.1 ⬍.0001 .79 ⬍.0001
weight, leading to a higher PEE. How-
NS, not significant; FAO/WHO/UNU, Food and Agriculture/World Health Organization/United ever, the ideal weight did not signifi-
Nation Union. cantly improve the accuracy of the for-
a
Paired t-test. mulas.
Several authors have found different
results when comparing MEE and PEE
cally ill patients can be difficult. We not Hunter equations). We hypothesized that (6, 49 –53). These contradictory data
only analyzed specific pediatric equations because formulas for adults take age and could be due to the fact that the studied
(Caldwell-Kennedy, Schofield, Food and weight into account, the same could also patients were not homogeneous, so they
Agriculture/World Health Organization/ be suitable for infants. However, our re- were in different stress phases. Moreover,
United Nation Union, and Fleisch equa- sults did not support this hypothesis, and some of them were fasting, whereas oth-
tions) but also formulas that were origi- adult predictive formulas are not suitable ers were receiving nutritional support.
nally based on an adult population for children. It is important to mention On the whole, our results show a lack of
(Harris-Benedict, Kleiber, Dreyer, and that we always compared the values by agreement even though good linear and

Pediatr Crit Care Med 2004 Vol. 5, No. 1 23

 and in two patients, it was ⬎379 kcal/day
(which represented 53% of MEE). This
represents poor precision; therefore, we
consider that the Caldwell-Kennedy
equation could be suitable for the overall
population, although it could lead to an
error that is not clinically tolerable in
some individual cases.
White et al. (17) studied 100 venti-
lated, critically ill children in different
stress response phases. They compared
the MEE obtained through short indirect
calorimetry studies with PEE estimated
using the Harris-Benedict, Schofield, and
Food and Agriculture/World Health Orga-
nization/United Nation Union equations.
In like manner to White et al., we proved
that predictive equations fail; we found
Figure 3. Linear and intraclass correlation coefficient of measured and predicted energy expenditure. poor agreement and poor precision with
Reproducibility was excellent when the Caldwell-Kennedy (C-K), Schoefield (Sch), Fleisch, Food and the Bland-Altman method. They also con-
Agriculture/World Health Organization/United Nation Union (FAO), and Kleiber equations were used. cluded that standard predictive equations
H-B, Harris-Benedict.
should be abandoned, and recommended
designing a new predictive equation that
takes into account clinical variables that
affect energy expenditure, such as body
temperature and the primary reason for
PICU admission. We have not validated
this equation in our series because it was
not designed for predicting energy expen-
diture during the early postinjury period.
Our patients were hypometabolic (the
MEE was lower than the predicted meta-
bolic rate of healthy children), as was
demonstrated by the metabolic index per-
taining to the Schofield equations. This is
consistent with other published data (5,
6, 15–17). Several factors can explain this
hypometabolism such as initial resuscita-
tion, mechanical ventilation (which re-
duces the cost of breathing) (54), admin-
istration of analgesia, sedation and
neuromuscular blockade (17, 55), and
malnutrition. Many of these conditions
were present in our patients.
The accuracy of the equations with the
type of injury (surgical vs. medical) was
also tested. The result showed a lack of
Figure 4. The Bland-Altman plot comparison between predicted energy expenditure using the Cald- reproducibility in the surgical group.
well-Kennedy equation and measured energy expenditure using continuous indirect calorimetry. A However, their nutritional stages were
lack of agreement exists between individual predicted energy expenditure (PEE) and measured energy very different, with the medical patients
expenditure (MEE) values, which could result in a considerable error in some individuals cases. PEE having a better nutritional state. This
and MEE are presented in kcal/day. Bias (mean difference between PEE and MEE), ⫺38 kcal/day; could partly explain the higher accuracy
precision (⫾1 SD of bias), ⫾179 kcal/day. of predictive equations in the medical
group: predictive methods have been de-
rived from well-nourished subjects; when
intraclass correlations existed between derived from direct calorimetry studies in malnutrition becomes severe, the energy
MEE and PEE. We found the Caldwell- healthy subjects. Its use is restricted to expenditure prediction is impaired. The
Kennedy equation to be the best predic- the first year of live. Nevertheless, the Caldwell-Kennedy equation was the best
tor of energy expenditure in ventilated, Bland-Altman plot for PEE pertaining to predictor of energy expenditure in medi-
critically ill children in the early postin- the Caldwell-Kennedy equation had a cal patients, whereas the Fleisch equation
jury period. This equation was formu- wide scatter around the mean. Seven pa- showed the highest precision in surgical
lated by Harris and Benedict in 1919 and tients (16.2%) had a bias ⬎179 kcal/day, patients. This would be consistent with

24 Pediatr Crit Care Med 2004 Vol. 5, No. 1

Table 4. Bland-Altman method comparison between measured and predicted energy expenditurea

P
Actual Weight Ideal Weight redictive equations
Equations Mean Bias Precision Mean Bias Precision do not accurately
Harris-Benedict 162.9b 237 177.3b 241 predict energy ex-
Caldwell-Kennedy ⫺37.7 180 0.4 189
Schofield 96.7b 186 109.2b 189 penditure in ventilated, criti-
FAO 82.7b 201 133.8b 198
Maffeis 181.4b 233 190.6b 235 cally ill children during the
Fleisch 108.7 250 61.1 170
Kleiber ⫺130.5b 179 ⫺96.7b 185 early postinjury period. These
Dreyer 296.5b 220 343.2b 230
Hunter ⫺317.8b 181 ⫺291.3b 189
children were frequently hy-
FAO, Food and Agriculture/World Health Organization/United Nation Union.
a
Mean bias and precision in kcal/day for predictive equations by using actual and ideal weight;
pometabolic during this phase
b
significantly different from indirect calorimetry, p ⬍ .05 (paired t-test). of stress response; if available,
indirect calorimetry must be
Table 5. Intraclass correlation coefficient of predictive energy expenditure between the diagnostic
groups performed.
Actual Weight Ideal Weight

Surgical Medical Surgical Medical

Equations (n ⫽ 35) (n ⫽ 8) (n ⫽ 35) (n ⫽ 8) early postinjury period, as other authors
have already shown for later stress re-
Harris-Benedict 0.495 0.724 0.448 0.716 sponse phases in critically ill children.
Caldwell-Kennedya 0.669 0.942 0.734 0.920 A technical limitation is that hemody-
Schofielda 0.691 0.884 0.658 0.880
FAO/WHO/UNUa 0.735 0.860 0.595 0.859 namic and respiratory stability are re-
Maffeisb 0.359 0.716 0.333 0.714 quired to perform an indirect calorimetry
Fleischa 0.736 0.842 0.752 0.911 study (56). This is especially important
Kleibera 0.626 0.852 0.678 0.849 when metabolic studies are performed
Dreyerb 0.182 0.738 0.097 0.715
Hunterb 0.256 0.744 0.265 0.745
only for a short time. When steady state is
not achieved, the metabolic monitor
a
The reproducibility changes from mild in surgical patients to excellent in medical patients; bthe shows wide and abrupt changes in meta-
reproducibility changes from poor in surgical patients to mild in medical patients. bolic readings. These erroneous measure-
Note that the magnitude of the intraclass correlation coefficient increases in the medical group, ments can be overcome by selecting Ar-
resulting in a better reproducibility. The use of the ideal weight in the equations produced variable tefact Suppression Mode in the monitor
results. set-up and by performing longer indirect
calorimetry studies, such as continuous
24-hr indirect calorimetry (57).
the hypothesis that weight is a major It is worth mentioning some limita- Finally, it is possible that not all ad-
determining factor of energy expendi- tions of our study. First, the medical mitted PICU patients with a medical con-
ture, not only in healthy subjects but also group was small. However, surgical and dition were actually within 6 hrs postin-
in ventilated, critically ill children. medical critically ill children were a ho- jury because the first 6 hrs postinjury is
We analyzed the energy expenditure mogeneous group pertaining to gender, not necessarily the same as the first 6-hr
and respiratory quotient variation age, weight, and height, which are the post-PICU admission. In contrast, both
throughout the first 24 hrs postinjury by major determining factors of energy ex- periods coincide in the surgical patients.
comparing long and short indirect calo- penditure. It is also known that the type
rimetry results. The paired differences of injury can influence the metabolic re- CONCLUSIONS
were not significant for energy expendi- sponse. However, early PICU manage- Predictive equations do not accurately
ture except for the first 150 mins. How- ment with similar therapeutic ap- predict the energy expenditure in venti-
ever, despite statistical significance, these proaches also plays a role in the lated, critically ill children during the
differences were small (about 45 kcal/day) metabolic stress response (53). Our pa- early postinjury period. These children
and, thus, have no clinical relevance. We tient population (medical and surgical) were frequently hypometabolic during
have also measured the respiratory quo- shared similar clinical and therapeutic this phase of stress response. Our find-
tient throughout the day, which did not conditions (e.g., early postinjury phase, ings support the routine use of indirect
significantly change during the 24 hrs. severity of illness, mechanical ventila- calorimetry, even including short indi-
This means that short indirect calorime- tion, sedated, and inotropic support), rect calorimetry studies, for this patient
try studies are suitable for calculating the which made them a homogeneous group population in those institutions that can
nutritional requirements during the first large enough to support the statement perform indirect calorimetry. This will
24 hrs postinjury that prediction equations fail during the allow the start of precocious nutrition,

Pediatr Crit Care Med 2004 Vol. 5, No. 1 25

even in the early postinjury period, thus
avoiding underfeeding and overfeeding.
Only the Caldwell-Kennedy equation can
correctly predict the overall energy ex-
penditure, although it is not suitable for
all patients because of the wide range of
individual measurements.
ACKNOWLEDGMENT
We thank the nurses and medical staff
of the Pediatric Intensive Care Unit at La
Paz Children’s Hospital, Madrid, Spain,
for collaboration in the performance of
this study.
REFERENCES
1. Briassoulis G, Zavras N, Hatzis T: Malnutri-
tion, nutritional indices, and early enteral
feeding in critically ill children. Nutrition
2001; 17:548 –557
2. Tissot S, Delafosse B, Bertrand O, et al: Clin-
ical validation of the Deltatrac monitoring
system in mechanically ventilated patients.
Int Care Med 1995; 21:149 –153
3. Weyland W, Weyland A, Fritz U, et al: A new
paediatric metabolic monitor. Intensive Care
Med 1994; 20:51–57
4. Peerless R, Epstein CD, Martin JE, et al:
Oxygen consumption in the early post-injury
period: Use of continuous, on-line indirect
calorimetry. Crit Care Med 2000; 28:
395– 401
5. Gebara BM, Gelmini M, Sarnaik A: Oxygen
consumption, energy expenditure, and sub-
strate utilization after cardiac surgery in
children. Crit Care Med 1992; 20:1550 –1554
6. Chwals WJ, Lally KP, Wooley MM, et al: Mea-
sured energy expenditure in critically ill in-
fants and young children. J Surg Res 1988;
44:467– 472
7. Phillips RP, Ott L, Young B, et al: Nutritional
support and measured energy expenditure of
the child and adolescent with head injury.
J Neurosurg 1987; 67:846 – 851
8. Imura K, Okada A: Perioperative nutrition
and metabolism in pediatric patients. World
J Surg 2000; 24:1498 –1502
9. Jones MO, Pierro A, Hashim A, et al: Postop-
erative changes in resting energy expendi-
ture and interleukin 6 level in infants. Br J
Surg 1994; 81:536 –538
10. Shanbhogue RLK, Lloyd DA: Absence de hy-
permetabolism after operation in the new-
born infant. JPEN 1992; 16:333–336
11. Powis MR, Smith K, Rennie M, et al: Effect of
major abdominal operations on energy and
protein metabolism in infants and children.
J Pediatr Surg 1998; 33:49 –53
12. Coss-Bu JA, Jefferson LS, Walding D, et al:
Resting energy expenditure in children in a
pediatric intensive care unit: Comparison of
Harris-Benedict and Talbot predictions with
indirect calorimetry values. Am J Clin Nutr
1998; 67:74 – 80
13. Verhoeven JJ, Hazelzet JA, van der Voort E,
26
et al: Comparison of measured and predicted
energy expenditure in mechanically venti-
lated children. Intensive Care Med 1998; 24:
464 – 468
14. Letton RW, Chwals WJ, Jamie A, et al: Early
postoperative alterations in infant energy use
increase the risk of overfeeding. J Pediatr
Surg 1995; 30:988 –993
15. Thureen PJ, Philips RE, De Marie MP, et al:
Technical and methodologic considerations
for performance of indirect calorimetry in
ventilated and nonventilated preterm in-
fants. Crit Care Med 1997; 25:171–180
16. Briassoulis G, Venkataraman S, Thompson
AE: Energy expenditure in critically ill chil-
dren. Crit Care Med 2000; 28:1166 –1172
17. White MS, Shepherd RW, McEniery JA: En-
ergy expenditure in 100 ventilated, critically
ill children: Improving the accuracy of pre-
dictive equations. Crit Care Med 2000; 28:
2307–2312
18. Rodríguez Hierro F: Obesidad. In: Argente
Oliver J, Carrascosa Lezcano A, Gracia
Bouthelier R, et al (Eds). Tratado de Endo-
crinología Pediátrica y De laq Adolescencia.
Editores Médicos SA. 1st Edición. Madrid,
1995, pp 1307–1310
19. Fleisch A: Le mètabolisme basal standard et
se détermination au moyen du “Metabocal-
culator”. Helvetica Medica Acta 1951; 18:
23– 45
20. Warteloo JC: Classification and definition of
protein-calorie malnutrition. BMJ 1972;
3:566 –569
21. Shukla A, Forsyth HA, Anderson CHM, et al:
Infantile overnutrition in the first year of life:
A field study in Dudley, Worceshire. BMJ
1972; 4:507–515
22. Pollack MM, Ruttimann UE, Getson PR: The
Pediatric Risk of Mortality (PRIMS) score.
Crit Care Med 1988; 16:1110 –1116
23. Yeh TS, Pollack MM, Ruttiman UE, et al:
Validation of a physiologic stability index
(PSI) for use in critically ill infants and chil-
dren. Pediatr Res 1984; 18:445– 451
24. Cullen DJ, Civetta JM, Briggs BA, et al: Ther-
apeutic intervention scoring system: a
method for quantitative comparison of pa-
tient care. Crit Care Med 1974; 2:57– 60
25. Harris JA, Benedict FG: Standard basal me-
tabolism constants for physiologists and cli-
nicians: A biometric study of basal metabo-
lism in man. Carnegie Institute Publication
of Washington 279. Philadelphia, JB Lippin-
cott, 1919, pp 223–250
26. Caldwell MD, Kennedy-Caldwell C: Normal
nutritional requirements. Surg Clin North
Am 1981; 61:489 –506
27. Schofield WN: Predicting basal metabolic
rate: new standards and review of previous
work. Hum Nutr Clin Nutr 1985; 39C(Suppl
1):5– 41
28. World Health Organization: Energy and pro-
tein requirements: Report of a joint FAO/
WHO/UNU expert consultation. WHO Tech-
nical Report Series No. 724. Geneva, World
Health Organization, 1985, p 71
29. Maffeis C, Schutz Y, Micciolo R, et al: Resting
metabolic rate in six- to ten-year-old obese
and non-obese children. J Pediatr 1993; 122:
556 –562
30. Kleiber M: The fire of the life. Huntington,
NY, Robert E Krieger, 1975, pp 179 –222
31. Sherman HC: Chemistry of Food and Nutri-
tion. Eighth Edition. New York, MacMillan,
1952
32. Hunter DC, Jaksic T, Lewis D, et al: Resting
energy expenditure in the critically ill: Esti-
mations versus measurement. Br J Surg
1988; 75:875– 878
33. Weissman C, Kemper M: Assessing hyperme-
tabolism and hypometabolism in the postop-
erative critically ill patient. Chest 1992; 102:
1556 –1571
34. Coss-Bu JA, Jefferson LS, Walding D, et al:
Resting energy expenditure and nitrogen bal-
ance in critically ill pediatric patients on me-
chanical ventilation. Nutrition 1998; 14:
649 – 652
35. Joosten KF, Jacobs FI, van Klaarwater E, et
al: Accuracy of an indirect calorimeter for
mechanically ventilated infants and children:
the influence of low rates of gas exchange
and varying FIO2. Crit Care Med 2000; 28:
3014 –3018
36. Weir JB: New methods for calculating meta-
bolic rate with special reference to protein
metabolism. J Physiol 1979; 109:1–9
37. Simonson DC, DeFronzo RA: Indirect calo-
rimetry: methodological and interpretative
problems. Am J Physiol 1990; 258:
E399 –E412
38. Smyrnios NA, Curley FJ, Shaker KG: Accu-
racy of 30-minute indirect calorimetry stud-
ies in predicting 24-hour energy expenditure
in mechanically ventilated, critically ill pa-
tients. JPEN 1997; 21:168 –174
39. Rosner B: The intra-class correlation coeffi-
cient. In: Rosner B (Ed). Fundamentals of
Biostatistics. Fourth Edition. Washington,
DC, 1995, pp 517–550
40. Bland JM, Altman DG: Statistical methods
for assessing agreement between two meth-
ods of clinical measurement. Lancet 1986;
1:307–310
41. Lavery GG, Glover P: The metabolic and nu-
tritional response to critical illness. Curr
Opin Crit Care 2000; 6:233–238
42. Zaloga GP: Enteral nutrition in the critically
ill. In: Chernow B (Ed). The Pharmacologic
Approach to the Critically Ill Patient. Third
edition. Baltimore, Williams & Wilkins,
2000, pp 1034 –1050
43. Heyland DK: Nutritional support in the crit-
ically ill patient: A critical review of the evi-
dence. Crit Care Clin 1998; 14:423– 440
44. Selby AM, McCauley JC, Schell DN, et al:
Indirect calorimetry in mechanically venti-
lated children: A new technique that over-
comes the problem of endotracheal tube
leak. Crit Care Med 1995; 23:365–370
45. Abdulrazzaq YM, Brooke OG: Respiratory
metabolism in preterm infants: The mea-
surement of oxygen consumption during
prolonged periods. Peditr Res 1984; 18:
928 –931
Pediatr Crit Care Med 2004 Vol. 5, No. 1
46. Bell EF, Rios GR, Wilmoth PK: Estimation of
24-h energy expenditure from shorter mea-
surement periods in premature infants. Pe-
diatr Res 1986; 20:646 – 649
47. Gudinchet F, Schutz Y, Micheli JL, et al:
Metabolic cost of growth in very low-birth-
weight infants. Pediatr Res 1982; 16:
1025–1030
48. Ireton-Jones CS, Turner WW: Actual or ideal
body weight: which should be used to predict
energy expenditure. J Am Diet Assoc 1991;
91:193–195
49. Tilden SJ, Watkins S, Tong TK, et al: Measured
energy expenditure in pediatric intensive care
patients. Am J Dis Child 1989; 143:490– 493
50. Molnár D, Jeges S, Erhardt E, et al: Measured
and predicted resting metabolic rate in obese
Pediatr Crit Care Med 2004 Vol. 5, No. 1
and nonobese adolescents. J Pediatr 1995;
127:571–577
51. Thompson MA, Bucolo S, Quirk P, et al:
Measured versus predicted resting energy ex-
penditure in infants: A need for reappraisal.
J Pediatr 1995; 126:21–28
52. Witte MK: Metabolic measurements during
mechanical ventilation in the pediatric in-
tensive care unit. Respir Care Clin N Am
1996; 2:573–586
53. Flancbaum L, Choban PS, Sambucco S, et
al: Comparison of indirect calorimetry, the
Fick method, and prediction equations in
estimating the energy requirements of
critically ill patients. Am J Clin Nutr 1999;
69:461– 466
54. Savino JA, Dawson J, Agarval NR, et al: The
metabolic cost of breathing in critical sur-
gery patients. J Trauma 1985; 12:
1126 –1133
55. Barton RG, Craft WB, Mone MC, et al:
Chemical paralysis reduces energy expen-
diture in patients with burns and severe
respiratory failure treated with mechanical
ventilation. J Burn Care Rehabil 1997; 18:
461– 468
56. Brandi LS, Bertolini R, Santini L, et al: Effects
of ventilator resetting on indirect calorimetry
measurement in the critically ill surgical pa-
tient. Crit Care Med 1999; 27:531–539
57. McClave SA, Spain DA, Skolnick JL, et al:
Achievement of steady state optimizes results
when performing indirect calorimetry. JPEN
2003; 27:16 –20
27