Clinical Nutrition xxx (xxxx) xxx

Contents lists available at ScienceDirect

Clinical Nutrition
journal homepage: http://www.elsevier.com/locate/clnu

Original article

Evaluating steady-state resting energy expenditure using indirect

calorimetry in adults with overweight and obesity
C.J. Popp a, *, M. Butler a, M. Curran a, P. Illiano a, M.A. Sevick a, b, D.E. St-Jules a
a
Department of Population Health, New York University, USA
b
Department of Medicine, New York University, USA

a r t i c l e i n f o s u m m a r y

Article history: Background: Determining a period of steady state (SS) is recommended when estimating resting energy
Received 6 May 2019 expenditure (REE) using a metabolic cart. However, this practice may be unnecessarily burdensome and
Accepted 2 October 2019 time-consuming in the research setting.
Aim: The aim of the study was to evaluate the use of SS criteria, and compare it to alternative approaches
Keywords: in adults with overweight and obesity.
Resting metabolic rate
Methods: In this cross-sectional, ancillary analysis, participants enrolled in a bariatric (study 1; n ¼ 13)
Energy expenditure
and lifestyle (study 2; n ¼ 51) weight loss intervention were included. Indirect calorimetry was per-
Methodology
Energy requirements
formed during baseline measurements using a metabolic cart for 25 min, including a 5-min stabilization
Weight loss period at the start. SS was defined as the first 5-min period with a coefficient of variation (CV) 10% for
both VO2 and VCO2 (hereafter REE5eSS). Body composition was measured using bioelectrical impedance
analysis in study 2 participants only. REE5eSS was compared against the lowest CV (REECVelowest), 5-min
time intervals (REE6e10, REE11e15, REE16e20, REE21e25), 4-min and 3-min SS intervals (REE4eSS and REE3
eSS), and time intervals of 6e15, 6e20 and 6e25 min (REE6e15, REE6e20, and REE6e25) using repeated
measures ANOVA and BlandeAltman analysis to test for bias, limits of agreement and accuracy
(±6% measured REE).
Results: Participants were 54 ± 13 years old, mostly women (75%) and had a BMI of 35 ± 5 kg/m2. Overall,
54/63 (84%) of participants reached REE5eSS, often (47/54, 87%) within the first 10-min (6e15 min).
Alternative approaches to estimating REE had a relatively low bias (16 to 13 kcals), narrow limits of
agreement and high accuracy (83e98%) when compared to REE5eSS, in particular, outperforming stan-
dard prediction equations (e.g., Mifflin St. Joer).
Conclusion: Indirect calorimetry measurements that utilize the 5-min SS approach to estimate REE are
considered the gold-standard. Under circumstances of non-SS, it appears 4-min and 3-min SS periods, or
fixed time intervals of atleast 5 min are accurate and practical alternatives for estimating REE in adults
with overweight and obesity. However, future trials should validate alternative methods in similar
populations to confirm these findings.
© 2019 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.

1. Introduction calorimetry (IC) to estimate REE based on the consumption of

The measurement of resting energy expenditure (REE) is
important for the study of conditions related to energy balance
(e.g., obesity). REE accounts for 60e70% of the variability in daily
energy expenditure [1].The gold-standard method for assessing
REE includes the use of a metabolic cart, which employs indirect
* Corresponding author. Department of Population Health, Center for Healthful
Behavior Change, New York University, 180 Madison Ave, New York, NY 10016, USA.
E-mail address: collin.popp@nyulangone.org (C.J. Popp).
oxygen and production of carbon dioxide. Metabolic carts are
common in both clinical and research settings, as they are more
accurate than prediction equations (i.e., Mifflin St. Joer), and
cheaper and easier to perform than direct calorimetry.
In 2015, the Academy of Nutrition and Dietetics (AND) published
a standard protocol for the measurement of IC when using a
metabolic cart [2].The protocol recommends that: 1) subjects rest
for 30-min prior to measurement, 2) the first 5-min of measure-
ment be discarded as a stabilization period, and 3) REE be calcu-
lated using the first 5-min steady state (SS) period. SS is defined as a

https://doi.org/10.1016/j.clnu.2019.10.002
0261-5614/© 2019 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.

Please cite this article as: Popp CJ et al., Evaluating steady-state resting energy expenditure using indirect calorimetry in adults with overweight
and obesity, Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.10.002
2 C.J. Popp et al. / Clinical Nutrition xxx (xxxx) xxx
period where the coefficient of variation (CV) is  10% for gas
exchange variables, the most common being the volume of oxygen
consumed (VO2) and the volume of carbon dioxide produced
(VCO2). Achieving SS is often regarded as necessary to obtain an
accurate measure of REE [3e7].
In the clinical (n of 1) setting, failure to achieve SS can be
problematic, as it indicates random measurement error that in-
creases the likelihood of over- or under-feeding for nutrition sup-
port. However, in a research setting, random measurement error
increases sample size requirements but should have minimal
impact on point estimates when using adequately powered sam-
ples [8].In this context, stringent REE protocols may be unneces-
sarily burdensome, time-consuming, and wasteful. Indeed, REE
measurement could easily take an hour per participant, and up to
25% of participants in prior studies fail to achieve SS [3,9e11].Cur-
rent recommendations are to repeat the IC measurement until SS is
achieved or discard the data points as invalid. Loss of these data
points may increase sample size requirements more than the
random error from including non-SS measurements and more
importantly, could lead to unknown systematic bias, as the factors
contributing to high CVs are not well understood. Therefore, the
aim of the study was to evaluate the use of SS criteria and compare
it to alternative approaches in adults with overweight and obesity.
2. Materials and methods
2.1. Study design
In this cross-sectional, ancillary analysis, data for this study were
collected during baseline measurement visits of 64 participants
enrolled in two ongoing weight loss studies. Study 1 participants
(n ¼ 13) were recruited for a prospective casecontrol study of bar-
iatric surgery, which included patients 18e75 years old that were
English or Spanish speaking, and either (1) diagnosed with type 2
diabetes mellitus (T2D) (or HbA1c  6.5%); (2) HbA1c <5.7%
measured within 6 months not currently on diabetic medication; or
(3) HbA1c between 5.7 and 6.4%. Exclusion criteria included: (1)
prior bariatric surgery; (2) insulin-dependent diabetes mellitus; (3)
recent use of non-steroidal anti-inflammatory medications (NSAIDs)
(<72 h); (4) chronic gastrointestinal disease (e.g., inflammatory
bowel disease); (5) autoimmune diseases (e.g., rheumatoid arthritis,
psoriasis); (6) prior myocardial infarction, percutaneous coronary
interventions, coronary angioplasty or stroke; (7) neoplastic disease
within 3 years; (8) use of immunosuppressive medications; (9)
steroid therapy or postmenopausal hormone therapy medication in
last 3 months (10) and chronic kidney disease (eGFR < 60 ml/min/
1.73 m2). Study 2 participants (n ¼ 51) were enrolled in the Personal
Diet Study, a randomized clinical weight loss trial in overweight
adults (BMI >27 kg/m2) with pre-diabetes (HbA1c 5.7e6.4%) and
early-stage T2D (HbA1c  8% treated with metformin and/or lifestyle
only). Details regarding the Personal Diet Study have been previ-
ously published [12].The work performed was carried out in accor-
dance with the Declaration of Helsinki. All participants signed an
informed consent prior to testing, and the Institutional Review
Board at New York University Langone Health approved testing
procedures.
3. Indirect calorimetry procedures
Measurement visits were conducted in a fasted state at the New
York University Langone Health Clinical and Translational Science
Institute (CTSI). Height was measured to the nearest 1 cm using a
portable stadiometer (SECA 213, Seca GmBH & Co. KG, Hamburg,
Germany), and body weight was measured in light clothing without
shoes to the nearest 0.1 kg using a Stow-A-Weigh scale (Scale-
Please cite this article as: Popp CJ et al., Evaluating steady-state resting energy expenditure using indirect calorimetry in adults with overweight
and obesity, Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.10.002
Tronix, Welch Allyn, Skaneateles, NY, USA). Prior to REE measure-
ment, participants were asked to confirm the following: (1) no
vigorous physical activity levels within last 48 h, (2) a minimum of
8 h fast, and (3) no use of caffeine or nicotine products that day. Fat
mass (FM) and fat-free mass (FFM) were assessed using bioelec-
trical impedance analysis (BIA; InBody 270, InBody, Inc. Cerritos,
CA, USA) in study 2 participants only. Participants were directed to
lay supine for 10 min, during which the flow meter and gas analyzer
were calibrated per the manufacturer's instructions. Open-circuit
IC was performed using a COSMED Quark RMR® metabolic cart
with flow-dilution canopy hood (COSMED, Rome, Italy). VO2 con-
sumption and VCO2 production were measured for 20 min
following a 5-min run-in stabilization period. The participants were
instructed to abstain from sleeping and talking during the test.
Room temperature (24.3 ± 3.6  C) and humidity (32.5 ± 17.5%) were
maintained as best as possible, and ambient noise and lighting
were minimized. Gas exchange variables (VO2 and VCO2) were
measured every 10 s. REE was calculated from the Weir equation
and expressed in kilocalories per day (kcal/d) [13].Respiratory
quotient (RQ) was determined by dividing VCO2 by VO2. To assess
the validity of the test, the range of RQ was determined and
compared to the physiological range (0.67e1.3). As an external
reference, REE was also estimated using the Mifflin St. Jeor
prediction equation (REEMF) [14].
3.1. Steady state
After discarding the 5-min stabilization period, standard refer-
ence REE was calculated as the first 5-min measurement period in
which the CV for both VO2 and VCO2 are 10% (REE5eSS). In addition,
the first 5-min period with the lowest CV from REE was determined
(REECVelowest), as were shortened SS periods of 4-min (REE4eSS) and
3-min (REE3eSS). Similar to REE5eSS, the REE4eSS and REE3-SS periods
were determined by selecting the first consecutive 4-min and 3-min
period with a CV  10% for both VO2 and VCO2. Finally, REE esti-
mates were calculated based on 5-min intervals and their rolling
averages: REE6e10, REE11e15, REE6e20, REE21e25, REE6e15, REE6e20,
and REE6e25. Participants who did not reach REE5-SS during the
measurement had their REE estimated from the REE4eSS or REE3eSS.
3.2. Statistical analysis
Data are reported as means ± standard deviations (SD) unless
otherwise noted, and the level of significance was set at an alpha of
0.05. The ShapiroeWilk test was run to assess for normality for
continuous variables. Participant characteristics were compared
between by study (study 1 vs. study 2), and by individuals who
reached REE5eSS and those who did not (i.e., non-SS) using inde-
pendent samples, 2-sided t-tests for normally distributed variables,
KruskaleWallis tests for non-normally distributed variables, and
Chi-squared tests for categorical variables.
To examine whether the first 5 min of the IC measurement,
which is generally discarded as a stabilization period, we conducted
a paired samples t-test comparing estimates of REE from first 5 min
(REE0e5) with estimates from the 6e25 min time interval. To
examine how estimates of our primary outcomes changed over the
IC measurement period and to see whether these outcome trajec-
tories differed by whether participants reached a steady state, we
conducted several repeated measures ANOVAs. We specified one
within-subjects factor (time) and one between-subjects factor (a
grouping variable for those who reach REE5eSS and those who did
not). The repeated measures for these analyses were the 6e10 min,
11e15 min, 16e20 min, and 21e25 min time interval mean esti-
mates for REE, VO2 CV, and VCO2 CV. The ANOVAs were performed
testing the effect of time and interaction effect for those who reach
 C.J. Popp et al. / Clinical Nutrition xxx (xxxx) xxx
REE5eSS and those who did not (i.e., non-SS). The F statistics for
within-subjects and between-subjects effects were estimated using
the Greenhouse-Geisser method. The same procedures were also
performed to examine whether outcome trajectories differed by
study (study 1 v. study 2).
Finally, we conducted analyses examining how different mea-
surement intervals for estimating REE (e.g., REE6e10 and REE4eSS)
compared to the gold standard estimate of REE5eSS. Bias, limits of
agreement and the plot of bias against average measured REE (SS
and time intervals) and the reference measurement (REE5eSS) were
analyzed with Bland Altman plots using GraphPad Prism (version
7.05, GraphPad Software, San Diego, CA, USA). Data for the
comparisons in the Bland Altman plots came only from those par-
ticipants who met the REE5eSS criteria. The accuracy of the mea-
surement was determined by counting the individual differences
that were ±6% of REE for each individual, which was considered
clinically relevant. Normal within-person variability in resting en-
ergy expenditure among adults is in the order of 3e6% [15e17].The
mean difference of measured REE close to zero suggests a good
level of agreement between measurements. All statistical analyses
were conducted using SPSS (SPSS version 23, IBM, Armonk, NY,
USA) and GraphPad Prism.
4. Results
The final sample included 63 participants. One participant had
consumed a nicotine product prior to the REE measurement, and
their data were removed from the final analysis. Table 1 details
participant characteristics by study type. On average, those in study
1 were younger (P < 0.0001) and had a greater BMI (P ¼ 0.001)
compared to those in study 2.
The participants in the sample were 53 ± 13 years old, mostly
women (75%) and had a BMI of 35 ± 5 kg/m2. Of the 63 participants,
84% of participants (n ¼ 53) reached REE5eSS, and 87% (n ¼ 47)
reached REE5eSS within the first 10 min (6e15 min). There was no
significant difference between study 1 and 2 participants with regards
to reaching SS under the 5-, 4- or 3-min criteria (Supplemental
Material Table 1). FM was significantly different (P ¼ 0.047) be-
tween REE5eSS and those who did not reach SS (Table 1).
Of the 10 participants who did not meet the REE5eSS criteria, 60%
(n ¼ 6) met the criteria for REE4eSS or REE3eSS. During the 6e15 min
time interval, 54% and 57% of participants achieved REE4eSS and
REE3eSS, respectively, indicating overlap between the SS-based
methods. Only 6% of our sample did not meet the 5-, 4 or 3-min
SS criteria. The range for RQ values for all participants during the
entire duration of the REE measurement (6e25 min) was 0.65e1.2,
which is slightly outside the physiological RQ range (0.67e1.3)
[18].The individual with an RQ of 0.65 had an RQ during the REE5eSS
period of 0.69, within the normal physiological limits.
Table 1
Participant characteristics by study and steady state.
All (n ¼ 63) Study 1 (n ¼ 13) Study 2 (n ¼ 50)
Age (yr) 54 ± 13 40 ± 12 57 ± 12
Gender (% Female) 75 9 (69) 38 (76)
Weight (kg) 98 ± 17 109 ± 11 9 ± 17
Height (cm) 166 ± 8 166 ± 7 166 ± 9
BMI (kg/m2) 35 ± 5 40 ± 3 34 ± 5
Body Fat (%) e e 41 ± 8
FFM (kg) e e 55 ± 12
FFM (%) e e 58 ± 8
FM (kg) e e 39 ± 11
FM (%) e e 41 ± 8
Values are Mean ± SD unless otherwise noted; SS ¼ steady state; REE5ess ¼ 5-min SS; REE4ess ¼ 4-min SS; REE3ess ¼ 3-min SS; BMI ¼ body mass index; FM ¼ Fat mass; FFM¼
Fat-free mass; REE ¼ resting energy expenditure; Body composition analysis was performed in only study 2 participants; p-values bolded were significant; p  0.05; FFM, FM,
and BMI were not normally distributed, KrushkaleWallis Test was performed on these variables.
Please cite this article as: Popp CJ et al., Evaluating steady-state resting energy expenditure using indirect calorimetry in adults with overweight
and obesity, Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.10.002
3
Table 2 compares the SS and time interval periods. The first
5-min of measurement differed from the subsequent 6e25 min
time interval in mean VO2, VCO2, and REE (P < 0.05), confirming
current recommendations to discard this stabilization period
[2,3,9,11].The repeated measures ANOVA comparing REE between
5-min measurement intervals indicates there was a significant
effect of time (F ¼ 2.784, P ¼ 0.047) but no interaction effect of time
and the between-subjects factor (reaching REE5eSS versus those
who did not; F ¼ 0.21, P ¼ 0.87). However, the effect of time was
clinically insignificant as the mean difference between REE6e10 and
REE21e25 was 26 kcal/d. There were no significant effects of time or
interaction effects with the between-subjects factor for CV esti-
mates of VO2 and VCO2. Furthermore, there was no effect of time
(F ¼ 2.122, P ¼ 0.105) and no interaction between time and study
type (study 1 vs. study 2; F ¼ 0.272, P ¼ 0.829). The time each in-
dividual achieved REE5eSS during the duration of the measurement
is depicted in Fig. 1. REE5eSS appears to overlap with the 4-min and
3-min SS periods (Supplemental Figs. 1 and 2).
Figure 2 depicts the BlandeAltman Plots for the SS, 5-min seg-
ments, and time intervals. Mean bias is represented as a solid line,
95% limits of agreement by short dashed lines, and the ±6% of in-
dividual REE (108 ± 22 kcal) by thick-dashed lines. The analysis
demonstrates low bias and narrow limits of agreement between
REE5eSS and the REE4eSS (Table 3) with all but one participant falling
within 6% of REE5eSS. The use of the REE4eSS produced limits of
agreement that were 4.3% below and 5.1% above the REE5eSS, making
REE4eSS a suitable alternative to estimate REE. Similar results were
seen between REE5eSS and the REE3eSS, but with larger limits of
agreement at 5.9% below and 6.7% above the REE5eSS. Over time, the
5-min time intervals became less precise and accurate, with wider
limits of agreement. The time intervals (i.e., REE6e10, REE6e15) had
low bias and narrow limits of agreement. The REE6e10 had limits of
agreement 4.4% below and 5.8% above, whereas REE6e15 had limits of
agreement 5.3% below and 5.6% above the REE5eSS. REE estimated
from the Mifflin equation (REEMF: 1632 ± 278 kcal/d) had the lowest
precision, widest limits of agreement and lowest accuracy (22%)
compared to the other measures.
As there were significant differences in baseline characteristics
(e.g. age and BMI) between the study 1 and study 2 samples, we
conducted sensitivity analyses evaluating whether achievement of
REE5eSS was associated with study cohort. Firstly we conducted a
binary logistic regression to determine the effects of age, BMI and
study on the odds of reaching REE5eSS. The logistic regression
model explained 7.5% (Nagelkerke R [2]; p ¼ 0.377) of the variance
in REE5eSS. Age (p ¼ 0.73), BMI (p ¼ 0.19) and study (p ¼ 0.89) were
not significantly associated with reaching REE5eSS. We also con-
ducted a ChieSquare test comparing the frequency of achieving
REE5eSS between study 1 and study 2 but found no significant
difference (study 1 N(%) ¼ 69; study 2 N(%) ¼ 82; p ¼ 0.311).
p-value REE5ess (n ¼ 53) non-SS (n ¼ 10) p-value
<0.0001 53 ± 13 52 ± 15 0.815
0.250 73 80 0.183
0.004 96 ± 17 106 ± 12 0.084
0.966 165 ± 8 168 ± 11 0.351
0.001 35 ± 6 37 ± 3 0.211
e 41 ± 8 44 ± 6 0.314
e 55 ± 12 59 ± 13 0.434
e 59 ± 8 55 ± 6 0.321
e 38 ± 12 46 ± 5 0.047
e 41 ± 8 44 ± 6 0.321
4 C.J. Popp et al. / Clinical Nutrition xxx (xxxx) xxx

Table 2
Comparison of steady state and time intervals.

REE5eSS REE4eSS REE3eSS REECVelow REE0e5 REE6e10 REE11e15 REE16e20 REE21e25 REE6e15 REE6e20 REE6e25

n/% SS 53/84 58/92 58/92 e e e e e e e e e

REE (kcal/d) 1796 ± 374 1796 ± 371 1794 ± 376 1808 ± 380 1880 ± 400 1819 ± 324 1802 ± 384 1797 ± 374 1793 ± 374 1811 ± 332 1806 ± 330 1803 ± 328
VO2 (ml/min) 258 ± 54 258 ± 54 258 ± 54 260 ± 55 269 ± 58 261 ± 49 260 ± 56 259 ± 55 258 ± 55 260 ± 50 260 ± 50 260 ± 50
VCO2 (ml/min) 217 ± 49 217 ± 50 219 ± 52 219 ± 49 233 ± 51 222 ± 40 218 ± 49 217 ± 50 215 ± 48 220 ± 41 219 ± 41 218 ± 40
CV VO2 (%) 6±2 6±2 6±2 7±3 13 ± 7 8±5 9±4 8±4 9±5 9±4 9±5 9±5
CV VCO2 (%) 8±2 8±2 7±2 9±5 14 ± 7 10 ± 7 10 ± 5 10 ± 6 10 ± 6 11 ± 6 11 ± 6 11 ± 6
RQ 0.84 ± 0.1 0.84 ± 0.1 0.85 ± 0.1 0.85 ± 0.1 0.87 ± 0.1 0.85 ± 0.1 0.85 ± 0.1 0.84 ± 0.1 0.84 ± 0.1 0.85 ± 0.1 0.85 ± 0.1 0.84 ± 0.1

Mean ± SD; those who met the 4-min and 3-min were not the same participants despite the same n. Agreement by short-dashed lines, and the ±6% of individual REE (108 kcal)
by thick-dashed lines.

5. Discussion the REE4eSS period may be a superior alternative, as it was more

The goal of an IC measurement is to provide an accurate rep-
resentation of 24-hr REE in an abbreviated time period. Current REE
measurement standards recommend a 30-min resting period prior
to measurement, discarding the first 5-min and achieving a SS
period of 5-min with a CV  10% for gas exchange parameters.
Although these protocols have the support of major nutrition
organizations (e.g., A.S.P.E.N., AND) they are graded as weak evi-
dence (Grade III) [2].As the strict measurement requirements may
be a barrier to research, we evaluated their necessity in a sample of
individuals who were predominantly obese.
The AND recommend a 30-min resting period prior to
measuring REE, which is based on the assumption that there is an
increase in energy expenditure associated with travel and move-
ment for measurement. Our results support prior research [9,11]
suggesting the 30-min resting period is unnecessary in free-living
adults with overweight and obesity. The current results indicate
that VO2 CV and VCO2 CV do not significantly change over time,
meaning that additional measurement time may not increase the
likelihood of reaching SS in overweight and obese adults. Reducing
the resting period to 10-min may be a suitable option as a means of
reducing the time required to measure REE. We previously found
that young, healthy adults who were required to travel to the
testing site were able to reach a 5-min SS period within 10-min
without prior rest [3].Frankenfield and Coleman [19] found that
after a 300-m walk, REE SS was achieved by the tenth minute, and
by 20 min, 95% of subject met their SS threshold.
As expected, a minority of individuals (~15%) failed to meet the
5-min SS criteria. The use of shorter SS periods (4- and 3-min)
increased the percentage of individuals who satisfied SS. However,
Fig. 1. The time each individual achieved REE5eSS during the duration of the mea-
surement. The 5-min intervals starting a minute 6 are listed on the x-axis.
accurate when compared to REE3eSS. Shorter SS periods of 4- and
3-min may serve as alternative estimates of REE. Reeves et al. [7]
found that a 4-min SS period was within the clinically meaning
range (±2%), but 3-min SS fell outside and may underestimate REE
despite low bias. Borges et al. [11] showed that 4- and 3-min SS
periods within a 10-min window allowed for a high percentage of
individuals to meet SS. However, these methods overestimated REE
(96e102 kcal/d). In our study, the 4-min SS interval may be used in
place of the 5-min SS. Importantly, these protocols use overlapping
time periods, so 4-min REE contains 24/30 (80%) of the same data
points as 5-min REE in the vast majority of individuals.
Our results confirm prior reports that some individuals fail to
satisfy the 5-min SS criteria over a 20-min measurement period.
The role of excess body weight as a predictor of non-SS was noted
by Irving et al. [9], suggesting that extreme fat free mass (FFM) and
fat mass (FM) at higher BMIs may affect the ability to reach SS.
While FFM accounts for over 80% of the variance in REE [20], it was
not significantly different between REE5eSS and non-SS groups. FM
was significantly greater among non-SS participants, but the dif-
ferences were non-significant when expressing FM as a percent of
total body weight. Bosy-Westphal et al. [21] found that absolute FM
up to 40% of total mass accounts for greater variance in REE, but
FM>40% suggests a lower impact. Future research should assess SS
in individuals with class 3 obesity (BMI >40 kg/m2) and the relative
contribution of body composition measured using a three or four
compartment model on the variance of REE.
The Mifflin St. Joer equation is commonly used in clinical and
research practice as the standard estimate for REE when IC is not
performed [22].Previously, Frankenfield [23] found that in in-
dividuals with obesity (BMI >30 kg/m2), the Mifflin equation was
unbiased (95% CI: 50 to 15 kcal/day) when compared to measured
REE with IC, and had an accuracy rate of 48%, defined as the per-
centage of estimates 5% of measured REE. In our study, the Mifflin
equation showed the greatest bias, widest limits of agreement, and
an accuracy of 22% when compared to the gold standard REE5eSS.
Assessing predicted and measured REE in a large cohort of obese
individuals (n ¼ 1851), Marra et al. [24] demonstrated that pre-
diction equations, like Mifflin St. Jeor, can be used for estimating
REE at the population level (BMI <40 kg/m2), however, these
equations are inaccurate for clinical (n of 1) use. Therefore, we
suggest for those who do not meet SS to estimate REE from the
6e15 min interval rather use a prediction equation as the 6e15 min
time interval estimate of REE were the most closely related to the
gold-standard REE5eSS. Practitioners and researchers may also find
the use of a time interval more practical, as some metabolic carts
(i.e., Parvo Medics) do not calculate CV intervals for SS in real-time.
Current recommendations suggest non-SS individuals repeat
the IC measurement at a later date until SS is achieved [9,11].While
these recommendations may be viable within a clinical setting;
they may be problematic in a research setting as they are burden-
some for participants and costly to research. Moreover, repeating

Please cite this article as: Popp CJ et al., Evaluating steady-state resting energy expenditure using indirect calorimetry in adults with overweight
and obesity, Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.10.002
 C.J. Popp et al. / Clinical Nutrition xxx (xxxx) xxx 5

Fig. 2. Bland Altman Plot analysis comparing REE5eSS of the reference period to (A) 4-min (B) 3-min (C) 6e10 min (D) 11e15 min (E) 16e20 min (F) 21e25 min (G) 6e15 min (H)
6e20 min (I) 6e25 min, and (J) REE estimated from Mifflin SteJeor equation. The mean bias is represented by a solid line, 95% limits of agreement by short-dashed lines, and the
±6% of individual REE (108 kcal) by thick-dashed lines.

Please cite this article as: Popp CJ et al., Evaluating steady-state resting energy expenditure using indirect calorimetry in adults with overweight
and obesity, Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.10.002
6 C.J. Popp et al. / Clinical Nutrition xxx (xxxx) xxx

Table 3
Bias and limits of agreement between reference SS, shortened SS and time intervals.

Mean Bias or Difference, kcal 95% Limits of Agreement, range kcal, (% within) % within ±6% REE

REE4eSS 7 ± 43 77 to 92 (96) 98

REE3eSS 7 ± 58 107 to 121 (96) 98
REECVelow 5 ± 89 169 to 181 (96) 85
REE6e10 13 ± 48 81 to 107 (94) 97
REE11e15 7 ± 70 144 to 129 (96) 91
REE16e20 10 ± 71 150 to 130 (94) 83
REE21e25 16 ± 69 151 to 119 (92) 89
REE6e15 3 ± 51 96 to 102 (94) 96
REE6e20 1 ± 52 103 to 100 (96) 94
REE6e25 5 ± 52 106 to 96 (94) 92
REEMF 182 ± 212 598 to 234 (94) 23

n ¼ 53 participants; REE5eSS was the reference.

the REE measurements do not guarantee an individual will reach SS.
Bader et al. [17] found that within-person CV did not change after
repeated measures of IC, regardless of the measurement protocol
(time interval or SS). The variability in gas exchange parameters
may depend on a patient's physiological status, or it may be a result
of physiological reactivity associated with the IC measurement (e.g.,
connecting of the canopy, measurement environment) [25].The use
of SS cutoffs ensures that within-person random error is limited.
However, it is not clear that this has a large impact on population
estimates. Given this lack of certainty, the results of the current
study suggest that researchers may use alternative methods to es-
timate REE (such as REE4eSS and REE6e10) rather than re-evaluating
participants in the hope of achieving REE5eSS status. However, re-
sults from the current study and use of alternative methods for
estimating REE should be validated in a similar population using a
direct calorimetry or 24-hr indirect calorimetry.
There are several limitations to our study that require consid-
eration. First, our trial included mostly women. Second, our results
can only be generalized to middle-age to older adults with obesity.
Third, we lacked an external reference of true REE for validation
(e.g., direct calorimetry), and so had to use the REE5eSS as the
reference value. Finally, as participants were free-living, we were
unable to control the time of fasting and physical activity prior to
the IC measurement. Therefore, we had to rely on participant self-
report of these factors. Future trials are needed to compare SS and
time interval methods to indirect calorimetry measured in a whole-
room calorimeter over a 24-hr period.
In this study, we evaluated alternative REE protocols in adults
with overweight and obesity, including shorter SS periods, and
estimates without defined SS parameters. We conclude SS periods
of 4-min and 3-min are accurate and practical methods for
estimating REE in a research setting. However, IC measurements
that utilize the 5-min SS approach to estimate REE are considered
the gold-standard. We also conclude that under conditions of non-
SS or with the use of a metabolic cart that does not display CV in
real-time, time intervals (e.g., 6e10, 6e15 min) may be used to
estimate REE. We recommend a simplified protocol that includes
10-min of resting, a 5-min stabilization period, and 10-min mea-
surement window. This abbreviated IC protocol allows for multiple
IC measurements to be performed within a reasonable time period
without compromising the accuracy of estimating REE. However,
future trials should validate our findings and recommendations in
similar populations in order to confirm alternative methods are
accurate when estimating REE.
Statement and funding sources
Study 2 was registered at ClinicalTrials.gov:NCT03336411. This
work was supported by Grant 17SFRN33590133 from the American
Heart Association.
Statement of authorship
CJ Popp contributed to the conceptualization, data curation,
formal analysis, investigation, methodology, project administra-
tion, resources, software, supervision, validation, visualization and
writing (original draft, review, and editing); DE St-Jules contributed
to the conceptualization, investigation, methodology, supervision,
validation, visualization and writing (original draft, review, and
editing); M Butler contributed to the formal analysis, investigation,
validation, and writing (review and editing). P Illiano and M Curran
were responsible for data curation, resources, software and writing
(review and editing); MA Sevick was responsible for the funding
acquisition and writing (review and editing) of the paper.
Conflict of interest
CJP is a sport nutrition consultant for Renaissance Periodization,
LLC.
Acknowledgments
We thank the participants who graciously gave their time and
effort into participating. We also want to thank the staff at the
Clinical and Translational Science Institute at NYU Langone Health
for their assistant and effort with the data collection.
Appendix A. Supplementary data
Supplementary data related to this article can be found at
https://doi.org/10.1016/j.clnu.2019.10.002.
References
[1] Donahoo WT, Levine JA, Melanson EL. Variability in energy expenditure and
its components. Curr Opin Clin Nutr Metab Care 2004;7(6):599e605. http://
www.ncbi.nlm.nih.gov/pubmed/15534426.
[2] Fullmer S, Benson-Davies S, Earthman CP, Frankenfield DC, Gradwell E, Lee PS,
et al. Evidence analysis library review of best practices for performing indirect
calorimetry in healthy and nonecritically ill individuals. J Acad Nutr Diet
2015;115(9):1417e46. https://doi.org/10.1016/j.jand.2015.04.003. e2.
[3] Popp CJ, Tisch JJ, Sakarcan KE, Bridges WC, Jesch ED. Approximate time to
steady-state resting energy expenditure using indirect calorimetry in young,
healthy adults. Front Nutr 2016;3:49. https://doi.org/10.3389/
fnut.2016.00049.
[4] Borges JH, Langer RD, Cirolini VX, P ascoa MA, Guerra-Júnior G, Gonçalves EM.
Minimum time to achieve the steady state and optimum abbreviated period
to estimate the resting energy expenditure by indirect calorimetry in healthy
young adults. Nutr Clin Pract 2016;31(3):349e54. https://doi.org/10.1177/
0884533615627268.
[5] Sanchez-Delgado G, Alcantara JMA, Ortiz-Alvarez L, Xu H, Martinez-Tellez B,
Labayen I, et al. Reliability of resting metabolic rate measurements in young adults:
impact of methods for data analysis. Clin Nutr 2018;37(5):1618e24. https://
doi.org/10.1016/j.clnu.2017.07.026.
[6] McClave S, Spain D, Skolnick J, Lowen CC, Kieber MJ, Wickerham PS, et al.
Achievement of steady state optimizes results when performing indirect

Please cite this article as: Popp CJ et al., Evaluating steady-state resting energy expenditure using indirect calorimetry in adults with overweight
and obesity, Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.10.002
 C.J. Popp et al. / Clinical Nutrition xxx (xxxx) xxx
calorimetry. J Parenter Enteral Nutr 2003;27(1):16e20. https://doi.org/
10.1177/014860710302700116.
[7] Reeves MM, Davies PSW, Bauer J, Battistutta D. Reducing the time period of
steady state does not affect the accuracy of energy expenditure measurements
by indirect calorimetry. J Appl Physiol 2004;97(1):130e4. https://doi.org/
10.1152/japplphysiol.01212.2003.
[8] Kwak SG, Kim JH. Central limit theorem: the cornerstone of modern statistics.
Korean J Anesthesiol 2017;70(2):144e56. https://doi.org/10.4097/
kjae.2017.70.2.144.
[9] Irving CJ, Eggett DL, Fullmer S. Comparing steady state to time interval and
nonesteady state measurements of resting metabolic rate. Nutr Clin Pract
2017;32(1):77e83. https://doi.org/10.1177/0884533616672064.
[10] Olejnik LA, Peters EN, Parrott JS, Marcus AF, Brody RA, Hand RK, et al.
Abbreviated steady state intervals for measuring resting energy expenditure
in patients on maintenance hemodialysis. J Parenter Enteral Nutr 2017;41(8):
1348e55. https://doi.org/10.1177/0148607116660981.
[11] Borges JH, Guerra-Júnior G, Gonçalves EM. Methods for data analysis of
resting energy expenditure measured using indirect calorimetry. Nutrition
2019;59:44e9. https://doi.org/10.1016/j.nut.2018.07.015.
[12] Popp CJ, St-Jules DE, Hu L, Ganguzza L, Illiano P, Curran M, et al. The rationale
and design of the personal diet study, a randomized clinical trial evaluating a
personalized approach to weight loss in individuals with pre-diabetes and
early-stage type 2 diabetes. Contemp Clin Trials 2019;79:80e8. https://
doi.org/10.1016/j.cct.2019.03.001.
[13] WEIR JBDB. New methods for calculating metabolic rate with special refer-
ence to protein metabolism. J Physiol 1949;109(1e2):1e9. http://www.ncbi.
nlm.nih.gov/pubmed/15394301.
[14] Mifflin MD, St Jeor ST, Hill LA, Scott BJ, Daugherty SA, Koh YO. A new pre-
dictive equation for resting energy expenditure in healthy individuals. Am J
Clin Nutr 1990;51(2):241e7. https://doi.org/10.1093/ajcn/51.2.241.
[15] Weststrate JA. Resting metabolic rate and diet-induced thermogenesis: a
methodological reappraisal. Am J Clin Nutr 1993;58(5):592e601. https://
doi.org/10.1093/ajcn/58.5.592.
[16] Haugen HA, Melanson EL, Tran ZV, Kearney JT, Hill JO. Variability of measured
resting metabolic rate. Am J Clin Nutr 2003;78(6):1141e4. https://doi.org/
10.1093/ajcn/78.6.1141.
Please cite this article as: Popp CJ et al., Evaluating steady-state resting energy expenditure using indirect calorimetry in adults with overweight
and obesity, Clinical Nutrition, https://doi.org/10.1016/j.clnu.2019.10.002
7
[17] Bader N, Bosy-Westphal A, Dilba B, Müller MJ. Intra- and interindividual
variability of resting energy expenditure in healthy male subjects - biological
and methodological variability of resting energy expenditure. Br J Nutr
2005;94(5):843e9. http://www.ncbi.nlm.nih.gov/pubmed/16277790.
[18] McClave SA, Snider HL. Invited review: use of indirect calorimetry in clinical
nutrition. Nutr Clin Pract 1992;7(5):207e21. https://doi.org/10.1177/
0115426592007005207.
[19] Frankenfield DC, Coleman A. Recovery to resting metabolic state after
walking. J Am Diet Assoc 2009;109(11):1914e6. https://doi.org/10.1016/
j.jada.2009.08.010.
[20] Ravussin E, Lillioja S, Anderson TE, Christin L, Bogardus C. Determinants of 24-
hour energy expenditure in man. Methods and results using a respiratory
chamber. J Clin Investig 1986;78(6):1568e78. https://doi.org/10.1172/
JCI112749.
[21] Bosy-Westphal A, Müller MJ, Boschmann M, Klaus S, Kreymann G,
Lührmann PM, et al. Grade of adiposity affects the impact of fat mass on
resting energy expenditure in women. Br J Nutr 2009;101(4):474e7. http://
www.ncbi.nlm.nih.gov/pubmed/19230079.
[22] Madden AM, Mulrooney HM, Shah S. Estimation of energy expenditure
using prediction equations in overweight and obese adults: a systematic
review. J Hum Nutr Diet 2016;29(4):458e76. https://doi.org/10.1111/
jhn.12355.
[23] Frankenfield DC. Bias and accuracy of resting metabolic rate equations in non-
obese and obese adults. Clin Nutr 2013;32(6):976e82. https://doi.org/
10.1016/j.clnu.2013.03.022.
[24] Marra M, Cioffi I, Sammarco R, Montagnese C, Naccarato M, Amato V, et al.
Prediction and evaluation of resting energy expenditure in a large group of
obese outpatients. Int J Obes 2017;41(5):697e705. https://doi.org/10.1038/
ijo.2017.34.
[25] Cadena-Me ndez M, Escalante-Ramírez B, Azpiroz-Leehan J, Infante-
Vazquez O. VO2 and VCO2 variabilities through indirect calorimetry instru-
mentation. SpringerPlus 2013;2(1):688. https://doi.org/10.1186/2193-1801-
2-688.