Am J Physiol Endocrinol Metab 302: E441–E448, 2012.
First published November 29, 2011; doi:10.1152/ajpendo.00290.2011.
Approaches for quantifying energy intake and %calorie restriction during
calorie restriction interventions in humans: the multicenter CALERIE study
Susan B. Racette,1 Sai Krupa Das,2 Manjushri Bhapkar,3 Evan C. Hadley,4 Susan B. Roberts,2
Eric Ravussin,5 Carl Pieper,7 James P. DeLany,5,6 William E. Kraus,7 James Rochon,3
and Leanne M. Redman5; The CALERIE Study Group
1
Washington University School of Medicine, St. Louis, Missouri; 2Jean Mayer US Department of Agriculture Human
Nutrition Research Center on Aging at Tufts University, Boston, Massachusetts; 3Duke Clinical Research Institute, Durham,
North Carolina; 4National Institute on Aging, Bethesda, Maryland; 5Pennington Biomedical Research Center, Baton Rouge,
Louisiana; 6University of Pittsburgh, Pittsburgh, Pennsylvania; and 7Duke University School of Medicine,
Durham, North Carolina
Racette SB, Das SK, Bhapkar M, Hadley EC, Roberts SB,
Ravussin E, Pieper C, DeLany JP, Kraus WE, Rochon J, Redman
LM; The CALERIE Study Group. Approaches for quantifying
energy intake and %calorie restriction during calorie restriction interventions in humans: the multicenter CALERIE study. Am J Physiol
Endocrinol Metab 302: E441–E448, 2012. First published November
29, 2011; doi:10.1152/ajpendo.00290.2011.—Calorie restriction (CR)
is a component of most weight loss interventions and a potential
strategy to slow aging. Accurate determination of energy intake and
%CR is critical when interpreting the results of CR interventions; this
is most accurately achieved using the doubly labeled water method to
quantify total energy expenditure (TEE). However, the costs and
analytical requirements of this method preclude its repeated use in
many clinical trials. Our aims were to determine 1) the optimal TEE
assessment time points for quantifying average energy intake and
%CR during long-term CR interventions and 2) the optimal approach
for quantifying short-term changes in body energy stores to determine
energy intake and %CR during 2-wk DLW periods. Adults randomized to a CR intervention in the multicenter CALERIE study underwent measurements of TEE by doubly labeled water and body
composition at baseline and months 1, 3, and 6. Average %CR
achieved during the intervention was 24.9 ⫾ 8.7%, which was
computed using an approach that included four TEE assessment time
points (i.e., TEEbaseline, months 1, 3, and 6) plus the 6-mo change in body
composition. Approaches that included fewer TEE assessments
yielded %CR values of 23.4 ⫾ 9.0 (TEEbaseline, months 3 and 6),
25.0 ⫾ 8.7 (TEEbaseline, months 1 and 6), and 20.9 ⫾ 7.1% (TEEbaseline,
month 6); the latter approach differed significantly from approach 1
(P ⬍ 0.001). TEE declined 9.6 ⫾ 9.9% within 2– 4 wk of CR
beginning and then stabilized. Regression of daily home weights
provided the most reliable estimate of short-term change in energy
stores. In summary, optimal quantification of energy intake and %CR
during weight loss necessitates a TEE measurement within the first
month of CR to capture the rapid reduction in TEE.
total energy expenditure; doubly labeled water; body energy stores
throughout the intervention. Furthermore, because physiological changes may differ depending on whether weight loss is
achieved by CR alone, exercise alone, or CR plus exercise, an
accurate assessment of EI is important to help distinguish
effects that are attributable to CR from those attributable to
exercise.
Many of the current options for estimating EI and %CR in
clinical trials have limited accuracy. Self-reported EI is
recognized to be inaccurate, with a bias toward underreporting, particularly among obese individuals (10, 27). Weight
change is also an imperfect quantitative indicator of EI and
%CR, particularly with interventions that combine CR and
exercise (23).
Average daily EI can be closely approximated from total
daily energy expenditure (TEE) when body weight is stable
(11, 27). TEE can be assessed using the doubly labeled water
(DLW) method (21), which is the most accurate method and
the gold standard for quantifying TEE and EI in free-living
individuals (9). When body weight is changing, EI is not equal
to TEE but can be approximated using the energy balance
equation: EI ⫽ TEE ⫹ change in body energy stores (23).
Although estimation of TEE by DLW during weight loss is
accurate and precise (3), measurement of changes in energy
stores is less accurate because the magnitude of changes in fat
mass (FM) and fat-free mass (FFM) is small relative to the
measurement error during 2-wk DLW periods (3). DLW periods longer than two isotope tracer half-lives (⬃2 wk in adults)
are not recommended because isotope enrichments become too
low to allow accurate quantification of isotope elimination
rates. Because DLW is generally used only before and at the end
of an intervention, assumptions must be made about TEE during
intervals between TEE assessments. Furthermore, the costs and
analytical requirements of the DLW method preclude its repeated use in many clinical trials. Therefore, our primary aim
was to determine which TEE assessment time points are
optimal for estimating energy intake and %CR during longterm CR interventions. A secondary aim was to determine the
optimal approach for quantifying short-term changes in body
energy stores during 2-wk DLW periods.
(CR), a reduction in energy intake (EI)
from habitual levels, is a key intervention in many clinical
weight loss trials and, more recently, as a potential means to
slow the aging process (8, 20). Determining a dose-response
relationship between CR and physiological changes requires
accurate and precise assessment of EI at baseline (BL) and
METHODS
Address for correspondence: S. B. Racette, Washington University School
of Medicine, Campus Box 8502, 4444 Forest Park Ave., St. Louis, MO 63108
(e-mail: [email protected]).
Participants. Participants included healthy, weight-stable, nonobese volunteers enrolled in phase 1 of the multicenter clinical trial
“CALERIE” (Comprehensive Assessment of Long-Term Effects of
Reducing Intake of Energy). The interventions were conducted from
CALORIC RESTRICTION
http://www.ajpendo.org
E441
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.6 on June 14, 2017
Submitted 13 June 2011; accepted in final form 28 November 2011
E442
QUANTIFYING ENERGY INTAKE AND %CR DURING WEIGHT LOSS
samples). Abundances of 2H and 18O in urine samples and in dilutions
of the isotope doses were measured in duplicate, using isotope ratio
mass spectrometry at the Pennington Biomedical Research Center.
Isotope elimination rates (kh and ko) were calculated using linear
regression, and CO2 production was calculated using the equations of
Schoeller (24), with the modifications of Racette et al. (16).
Because of the proximity of the BL1, BL2, and M1 TEE measurements, isotopic enrichment remained elevated at the beginning of the
BL2 and M1 TEE periods. Therefore, the predose urine samples from
BL1 were analyzed with the BL2 and M1 sample sets to determine
isotopic enrichment above baseline for the calculation of isotope
elimination rates. Likewise, the predose BL1 urine samples were used
to determine the time zero enrichment for the dilution space calculations, but the values were adjusted for the actual predose enrichments
measured at BL2 and M1. TEE was computed using respiratory
quotient values of 0.858 at baseline, 0.840 at M1, 0.851 at M3, and
0.860 at M6. These group respiratory quotient values were derived
from each subject’s reported diet composition at each time point and
measured changes in body composition between time points.
Resting metabolic rate was measured by indirect calorimety in the
morning after an overnight fast and an inpatient stay in the metabolic
research unit at each study site at baseline, M3, and M6; a subsample
also was assessed at M1.
Body composition. FM and FFM were measured at the beginning
and end of each 2-wk TEE measurement period by either dual-energy
X-ray absorptiometry, as described previously (17, 18), or air displacement plethysmography (BOD POD; Life Measurement, Concord, CA). The BOD POD measurements were made in duplicate
using methods described previously (14).
Body weight. All body weight measurements were made in the
morning in the fasted state after voiding. Two methods were used:
clinic weights and home weights. Clinic weights were measured each
week by research personnel at the study clinics using calibrated scales
Fig. 1. Study timeline and description of 9 approaches for computing average 6-mo energy intake and %calorie restriction (CR). TEE, total daily energy
expenditure; DLW, doubly labeled water; BL, baseline; M1, month 1; M3, month 3; M6, month 6.
AJP-Endocrinol Metab • doi:10.1152/ajpendo.00290.2011 • www.ajpendo.org
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.6 on June 14, 2017
2002 to 2005 at three clinical sites: Pennington Biomedical Research
Center (Baton Rouge, LA), Tufts University (Boston, MA), and
Washington University School of Medicine (St. Louis, MO). Participant characteristics, inclusion and exclusion criteria, and the CR
interventions are described in detail elsewhere (1, 7, 17). Briefly,
eligible subjects with body mass index between 23.5 and 29.99 kg/m2
were randomized to one of several CR regimens, an exercise regimen,
a combination of CR and exercise, or to a control group for a 6- or
12-mo period. To allow comparison across study sites, the current
analysis includes only subjects enrolled in CR interventions without
an exercise component (n ⫽ 54), and the duration is limited to the first
6 mo of the intervention. Of these 54 participants, 40 met secondary
criteria for inclusion in the current analysis based on complete data for
TEE, body composition, and body weight at BL and months 1, 3, and
6. Each study was approved by the site’s Institutional Review Board.
All participants gave written informed consent before enrollment.
CR interventions. CR prescriptions at the three sites ranged from 16
to 30% (i.e., a 16 –30% reduction in EI relative to BL EI). The mean
prescribed %CR during the 6-mo CR interventions was 27.9 ⫾ 5.3%.
BL EI was assumed to equal BL TEE determined by the doubly
labeled water (DLW) method during two consecutive 2-wk periods.
Participants received intensive behavioral support and counseling to
facilitate adherence to the CR interventions and were advised not to
change their daily physical activity.
Energy expenditure. Total daily energy expenditure (TEE) was
measured by the DLW method (24) during two consecutive 2-wk
periods at BL and during 2-wk periods preceding the month 1, month
3, and month 6 time points (i.e., M1, M3, and M6) of the CR
intervention, as shown in Fig. 1. Two baseline urine samples were
collected at the clinical sites before an oral dose of doubly labeled
water was administered (0.20 g of H218O, 0.12 g of 2H2O/kg total
body water). Subjects voided 1.5 and 3 h after dosing; postdose urine
samples were collected at 4.5 h, 6 h, day 7 (2 samples), and day 14 (2
QUANTIFYING ENERGY INTAKE AND %CR DURING WEIGHT LOSS
Table 1. Participant characteristics at baseline
Age, yr
Weight, kg
Body mass index, kg/m2
%Fat mass
TEE, kcal/day
Resting metabolic rate,
kcal/day
Physical activity level*
All (n ⫽ 40)
Females (n ⫽ 30)
Males (n ⫽ 10)
38.8 ⫾ 8.5
79.3 ⫾ 10.0
27.6 ⫾ 1.6
35.0 ⫾ 7.7
2,761 ⫾ 479
38.6 ⫾ 8.1
75.6 ⫾ 7.5
27.6 ⫾ 1.7
38.4 ⫾ 5.3
2,559 ⫾ 295
39.6 ⫾ 10.3
90.5 ⫾ 8.1
27.6 ⫾ 1.6
24.8 ⫾ 3.8
3,369 ⫾ 410
1,550 ⫾ 251
1.79 ⫾ 0.19
1,464 ⫾ 190
1.76 ⫾ 0.18
1,811 ⫾ 232
1.87 ⫾ 0.20
Values are means ⫾ SD. TEE, total daily energy expenditure. *Physical
activity level is calculated as TEE/resting metabolic rate.
bined with short-term changes in energy stores (using short-term ⌬ES
approaches c, d, e, or f). Average 6-mo EI was calculated as a
time-weighted average that included all four EI values (i.e., baseline,
month 1, month 3, and month 6) and used the precise number of days
between time points for each participant.
For the eight TEE-based approaches, baseline EI was computed as
the average TEE from two consecutive baseline TEE assessment
periods; EI was assumed to equal TEE because subjects were weight
stable at baseline.
Quantifying %CR. Mean 6-mo %CR was calculated as the percentage decrease in EI relative to baseline EI for each of the nine EI
quantification approaches according to the following equation:
%CR(mean) ⫽ [1 ⫺ EI(mean)/EI(BL)] ⫻ 100.
Statistical analysis. Results are reported as means ⫾ SD unless
otherwise noted. Confidence intervals are presented for %CR during
the 6-mo intervention. Repeated-measures ANOVA using SAS Proc
Mixed was used to test for equality of means of EI and %CR across
the different approaches. A heterogeneous compound symmetry covariance structure was assumed to account for unequal variances for
the different approaches. We tested overall differences for all approaches vs. the reference as well as pairwise comparisons for each
approach with the reference. This analysis is considered exploratory,
so no adjustments were made for multiple comparisons. Agreement
between individual %CR values obtained by quantification approach
1 and approaches 2, 3, and 5 was assessed graphically by the method
of Bland and Altman. All analyses were performed with SAS software
version 9.2 (SAS Institute, Cary, NC).
RESULTS
As shown in Table 1, 40 participants with complete data
during 6 mo of CR were included in the analyses. The age
range was 24 –59 yr; all but three participants were classified as
overweight. Average body weight values at baseline, month 1,
month 3, and month 6 were 79.3 ⫾ 10.0, 77.2 ⫾ 9.7, 73.6 ⫾
9.1, and 70.7 ⫾ 8.8 kg, respectively. Average 6-mo weight
change was ⫺8.6 ⫾ 3.7 kg (range: ⫺3.3 to ⫺19.9 kg). Fat
mass comprised 61.3 ⫾ 43.8% of the lost weight at month 1,
82.6 ⫾ 23.3% at month 3, and 85.5 ⫾ 17.3% at month 6.
Figure 2 illustrates the relative changes in body weight, body
composition, TEE, and resting metabolic rate observed during
the 6-mo CR interventions. As shown, TEE declined 9.6 ⫾
9.9% (P ⬍ 0.0001) within 1 mo of CR beginning and then
stabilized. TEE values at baseline, month 1, month 3, and
Fig. 2. Changes in body weight, body composition, TEE, and resting metabolic
rate after 1, 3, and 6 mo of CR.
AJP-Endocrinol Metab • doi:10.1152/ajpendo.00290.2011 • www.ajpendo.org
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.6 on June 14, 2017
that were accurate to ⫾50 g. Home weights were measured and
recorded by participants in their homes, using calibrated scales provided by the study. Home weights were obtained daily for ⱕ4 wk,
beginning 1 wk before and ending 1 wk after each 2-wk TEE
measurement period.
Self-reported food intake. Participants received detailed instructions on how to measure and record all foods and beverages in daily
food diaries. Research dietitians reviewed these diaries with participants at the end of each recording period and analyzed them for
nutrient content using either Nutrition Data System for Research
(NDS-R; software versions 4.05, 4.06, and 5.0; Nutrition Coordinating Center, University of Minnesota, Minneapolis, MN) or Moore’s
Extended Nutrient Database (MENu; Pennington Biomedical Research Center, Baton Rouge, LA). EI was calculated from two
consecutive 6-day food records at baseline and 6-day food records at
months 1, 3, and 6 of the CR intervention, corresponding to the TEE
measurement periods. Only participants with food records for ⱖ6
days at baseline and ⱖ5 days at each subsequent time point were
included in this analysis.
Quantifying change in body energy stores. The average daily
change in energy stores (⌬ES) was computed from either body
composition or body weight using six different approaches. Longterm ⌬ES was computed in two ways: 1) change in body composition
from baseline to month 6 divided by the exact number of days
between measurements and 2) change in clinic weight from baseline
to month 6 divided by the exact number of days between measurements. Short-term ⌬ES was computed in four ways: 1) change in body
composition between TEE time points (i.e., BL to M1, M1 to M3, M3
to M6) divided by the number of days between time points, 2) change
in body composition during each 2-wk TEE assessment period divided by 14 days, 3) regression of weekly clinic weights during each
2-wk TEE assessment period, and 4) regression of daily home weights
for 4 wk surrounding each 2-wk TEE assessment. Changes in body
composition were converted to ⌬ES using 9.3 and 1.1 kcal/g as the
energy coefficients of FM and FFM, respectively: ⌬ES (kcal/day) ⫽
⌬FM (g/day) ⫻ 9.3 kcal/g ⫹ ⌬FFM (g/day) ⫻ 1.1 kcal/g. Changes in
body weight (g/day) were converted to ⌬ES (kcal/day) using the
energy coefficient 7.4 kcal/g: ⌬ES (kcal/day) ⫽ ⌬weight (g/day) ⫻
7.4 kcal/g.
Quantifying EI. Nine approaches were evaluated for quantifying EI
and %CR, as outlined in Fig. 1. Approaches 1– 8 utilized TEE and
changes in body energy stores, whereas approach 9 was based on
self-reported dietary intake using food diaries. Four long-term approaches for determining average 6-mo EI and %CR were based on
average 6-mo TEE plus 6-mo change in body composition (using the
long-term ⌬ES approach as described above). Average 6-mo TEE was
calculated as a time-weighted average that included up to four TEE
assessment time points (i.e., baseline, months 1, 3, and 6) and used the
precise number of days between time points for each participant.
Four short-term approaches for determining average 6-mo EI and
%CR were based on EI values at baseline, month 1, month 3, and
month 6 computed from TEE assessments at those time points com-
E443
E444
QUANTIFYING ENERGY INTAKE AND %CR DURING WEIGHT LOSS
Table 2. Daily changes in body energy stores during CR interventions computed using 6 different approaches
Daily Change in Body Energy Stores, kcal/day
Approaches for Computing Change in Body Energy Stores
Long term
6-Mo ⌬body composition (approach a)
6-Mo ⌬weight (approach b)
Short term
⌬Body composition between time points (approach c)
2-Wk ⌬body composition (approach d)
2-Wk weekly clinic weight regression (approach e)*
4-Wk daily home weight regression (approach f)†
M1
M3
M6
BL to M6
⫺422 ⫾ 180
⫺388 ⫾ 163
⫺621 ⫾ 548
⫺891 ⫾ 470
⫺659 ⫾ 409
⫺657 ⫾ 331
⫺594 ⫾ 241
⫺669 ⫾ 432
⫺563 ⫾ 464
⫺403 ⫾ 233
⫺281 ⫾ 173
⫺268 ⫾ 613
⫺304 ⫾ 402
⫺232 ⫾ 335
⫺422 ⫾ 180
⫺615 ⫾ 307‡
⫺515 ⫾ 243‡
⫺424 ⫾ 181
Values are means ⫾ SD; n ⫽ 40 participants for approaches a, b, c, e, and f and n ⫽ 38 for approach d. See METHODS for calculations. CR, calorie restriction;
M1, month 1; M3, month 3; M6, month 6; BL, baseline. *Mean no. of body weight values included was 3.3 ⫾ 0.5. †Mean no. of body weight values included
was 18.7 ⫾ 2.9. ‡P ⬍ 0.01 for comparison with approach a.
and 24.9 ⫾ 8.7%, respectively. To explore the consequences of
reducing the number of TEE measurements, we computed EI
and %CR after excluding the month 1 and/or month 3 TEE
values. Exclusion of the month 3 TEE time point (approach 3)
did not compromise the results, yielding 6-mo EI and %CR
values that agreed most closely with approach 1. In contrast,
exclusion of both interim TEE measurements produced unacceptably high EI and, therefore, low %CR values relative to
approach 1. Bland-Altman plots depicting the difference in
individual %CR values between approach 1 and approaches 2,
3, and 4 are shown in Fig. 4.
Two additional long-term approaches for computing EI and
%CR were explored (results not included in Table 3). Use of
6-mo change in weight instead of body composition in approach 1 did not significantly affect EI or %CR (2,097 ⫾ 362
kcal/day and 23.7 ⫾ 8.3% CR, respectively). In contrast, using
only baseline TEE plus a 6-mo change in body composition
yielded unacceptably high EI (2,339 ⫾ 452 kcal/day) and, therefore, low %CR values (15.4 ⫾ 5.8%) relative to approach 1
because metabolic adaptation was not accounted for.
Short-term EI quantification approaches that utilized changes in
body composition between TEE time points (approach 5) or 4-wk
daily home weight regression (approach 8) agreed well with
approach 1 (Table 3). In contrast, approaches that relied on 2-wk
changes in body composition (approach 6) or 2-wk clinic weight
regression (approach 7) to compute average 6-mo EI and %CR
did not agree with the reference approach.
DISCUSSION
Fig. 3. Association between daily change in body energy stores computed from
body weight and daily change in body energy stores computed from fat mass
and fat-free mass. Body weight and body composition were measured at
baseline and after 6 mo of CR. ⽧Body weight.
Using data compiled from a multicenter study of CR (CALERIE),
we tested various approaches for determining energy intake
and %CR during a 6-mo period of CR. The results of these
analyses have important clinical applicability for weight loss
studies and other trials in which energy intake is restricted
relative to the weight maintenance energy requirement, such as
CALERIE. Three important observations of the current study
are that 1) optimal quantification of average energy intake and
%CR during a weight loss intervention necessitates a TEE
measurement within the first 2– 4 wk to capture the rapid
reduction in daily energy expenditure that occurs early during
CR, 2) the optimal approach for determining free-living energy
intake and %CR at a single point in time during active weight
loss includes a TEE measurement combined with regression of
daily weights for 4 wk surrounding the TEE assessment period,
and 3) body weight is a valid substitute for body composition
AJP-Endocrinol Metab • doi:10.1152/ajpendo.00290.2011 • www.ajpendo.org
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.6 on June 14, 2017
month 6 averaged 2,761 ⫾ 479, 2,482 ⫾ 416, 2,471 ⫾ 454, and
2,447 ⫾ 410 kcal/day, respectively. Changes in resting metabolism accounted for 25.8% of the decrease in TEE at month 1,
23.0% at month 3, and 43.0% at month 6.
A comparison of six approaches for computing changes in
body energy stores is shown in Table 2. Average ⌬ES based on
a 6-mo change in body composition (approach a) was ⫺422 ⫾
180 kcal/day. As shown in Fig. 3 for individual subjects, there
was a strong relationship between ⌬ES computed from the
6-mo change in body composition and ⌬ES computed from the
6-mo change in body weight (r2 ⫽ 0.877, intercept ⫽ 30.0,
slope ⫽ 0.85). Of the short-term methods employed, regression
of daily home weights over 4 wk (approach f, ⫺424 ⫾ 181
kcal/day) yielded a ⌬ES estimate that agreed very well with
approach a. In contrast, 2-wk changes in body composition
(approach d) and a 2-wk clinic weight regression (approach e)
produced the most variable ⌬ES results that differed significantly (P ⬍ 0.01) from approach a.
EI and %CR results computed according to the nine quantification approaches are shown in Table 3. Approach 1 serves
as the reference approach and includes all four TEE assessment
time points and a 6-mo change in body composition; average
6-mo EI and %CR were computed to be 2,063 ⫾ 373 kcal/day
Approaches for Quantifying Energy Intake and %CR
Average
2,042 ⫾ 374
1,862 ⫾ 393§
1,955 ⫾ 426§
2,045 ⫾ 395
1,981 ⫾ 352§
2,166 ⫾ 444
2,178 ⫾ 773
2,142 ⫾ 620
2,215 ⫾ 602
1,988 ⫾ 396
M6
2,761 ⫾ 479
2,761 ⫾ 479
2,761 ⫾ 479
2,761 ⫾ 479
2,325 ⫾ 570
1,877 ⫾ 446
1,794 ⫾ 440
1,909 ⫾ 631
2,068 ⫾ 463
2,021 ⫾ 371
M3
2,063 ⫾ 373
2,108 ⫾ 394
2,062 ⫾ 383
2,182 ⫾ 408§
1,861 ⫾ 643
1,611 ⫾ 558
1,823 ⫾ 517
1,825 ⫾ 458
1,900 ⫾ 392
M1
2,761 ⫾ 479
2,761 ⫾ 479
2,761 ⫾ 479
2,761 ⫾ 479
BL‡
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.6 on June 14, 2017
25.7 ⫾ 9.2
32.8 ⫾ 11.7§
29.0 ⫾ 11.4§
25.8 ⫾ 8.5
11.5 ⫾ 19.9§
24.9 ⫾ 8.7
23.4 ⫾ 9.0
25.0 ⫾ 8.7
20.9 ⫾ 7.1§
Average
%CR
22.7, 28.6
28.8, 36.7
25.3, 32.6
23.1, 28.5
4.4 ⫾ 18.6
22.2, 27.8
18.6, 23.2
22.2, 27.7
95% CI
Values are means ⫾ SD; n ⫽ 40 participants for approaches 1–5 and 7, n ⫽ 36 for approach 6, and n ⫽ 33 for approach 9. See METHODS for description. CI, confidence interval. †TEE by doubly labeled
water. ‡BL energy intake was assumed to equal baseline TEE for approaches 1– 8; prescribed %CR averaged 27.9 ⫾ 5.3 during the CR interventions. §P ⬍ 0.01 for comparison with approach 1.
Long-term approaches
6-Mo TEE(BL, M1, M3, M6) ⫹ 6-mo ⌬body composition (approach 1)†
6-Mo TEE(BL, M3, M6) ⫹ 6-month ⌬body composition (approach 2)
6-Mo TEE(BL, M1, M6) ⫹ 6-mo ⌬body composition (approach 3)
6-Mo TEE(BL, M6) ⫹ 6-mo ⌬body composition (approach 4)
Short-term approaches
2-Wk TEE ⫹ ⌬body composition between time points (approach 5)
2-Wk TEE ⫹ 2-Wk ⌬body composition (approach 6)
2-Wk TEE ⫹ 2-Wk weekly clinic weight regression (approach 7)
2-Wk TEE ⫹ 4-Wk daily home weight regression (approach 8)
1-Wk dietary intake (food diary, self-reported; approach 9)
Energy Intake, kcal/day
Table 3. Energy intake and %CR during 6-mo CR interventions quantified using 9 different approaches
QUANTIFYING ENERGY INTAKE AND %CR DURING WEIGHT LOSS
AJP-Endocrinol Metab • doi:10.1152/ajpendo.00290.2011 • www.ajpendo.org
E445
Fig. 4. Bland-Altman comparisons of %CR computed using the reference
approach that incorporated 4 TEE assessment time points with alternative
approaches based on 2 or 3 TEE time points. TEE was assessed using the DLW
method. %CR was computed using a time-weighted average TEE plus 6-mo
change in body composition. Time points included BL, M1, M3, and M6.
when determining long-term changes in body energy stores
during weight loss.
Although it has been known for many years that metabolic
rate and 24-h energy expenditure decline quickly upon initiation of an energy-restricted diet, previous studies (2, 22)
utilized room calorimeters, more restrictive diets (i.e., ⱖ50%
CR), and relatively small sample sizes. The current study is the
first to demonstrate that TEE in free-living individuals decreases significantly within the first 2– 4 wk of beginning a
moderate (i.e., 25%) CR diet. Similar to the observations of de
Boer et al. (2), in which 24-h energy expenditure decreased
E446
QUANTIFYING ENERGY INTAKE AND %CR DURING WEIGHT LOSS
due to the small magnitude of changes in FM and FFM relative
to measurement error. Importantly, the EI quantification approach based on regression of daily home weights yielded EI
and %CR results that were not statistically different from the
reference approach, whereas approaches that computed EI and
%CR using 2-wk body composition changes or regression of
weekly clinic weights produced results that differed significantly from the reference approach. These results are consistent with the findings of Hall and Chow (6), who demonstrated
that daily body weight measurements can be used to estimate
an individual’s change in energy intake if the assessment
period exceeds 28 days. Although their mathematical model
provides accurate and precise estimates of changes in EI during
weight loss, these authors point out that determining an individual’s EI and adherence to a diet necessitates an accurate
assessment of baseline EI, which can only be achieved using
DLW. In support of this assertion, we found that estimates of
baseline EI computed using the Dietary Reference Intakes
Estimated Energy Requirement (EER) formulae (9) [as proposed in the model of Hall and Chow (6)] differed substantially
from baseline EI determined by DLW, with the degree of error
dependent upon the physical activity coefficient used in each
EER formula (range: ⫺25.6 to ⫹0.6% error for “low active”,
⫺17.4 to ⫹14.2% error for “active”). Individualizing the
physical activity level can minimize this error.
In agreement with previous studies that documented low
accuracy of self-reported EI (13, 15), the food diary data in the
current study revealed underreporting of EI compared with
TEE-determined EI at baseline. Self-reported EI remained
lower than the reference approach when averaged across the
6-mo CR intervention, although the degree of underestimation
was relatively small due to the extensive training and practice
in monitoring food intake during the CR intervention. As
expected, %CR based on food diary data was highly variable
and did not agree very well with the reference approach. It is
noteworthy that objective measurements using DLW and body
composition changes indicate that, on average, adherence to
the CR intervention was excellent, with 25.5% CR achieved
compared with 27.9% prescribed.
Del Corral et al. (4) presented an alternative method of
assessing dietary adherence in weight loss regimens that is also
based on measurements of TEE and changes in body energy
stores. In their model, adherence is computed as the ratio of
actual daily energy loss to expected energy loss; actual daily
energy loss is calculated from fat mass and fat-free mass by the
same approach as in our study, and expected daily energy loss
is calculated by subtracting prescribed EI from average TEE,
with average TEE based on pre- and postintervention measurements: % adherence ⫽ 100 ⫻ [TEE(average of BL, final) ⫺ EIactual]/
[TEE(average of BL, final) ⫺ EIprescribed].
In contrast, in the current study we computed %CR as the
reduction in EI relative to baseline EI, and adherence can be
expressed as the ratio of actual %CR to prescribed %CR: %adherence ⫽ 100 ⫻ [(1 ⫺ EIactual/EIBL ) ⫻ 100/(%CRprescribed)].
Because TEE declines quickly upon initiation of CR (2, 22)
due to weight loss and metabolic adaptation (19), %CR and
%adherence calculated using our approaches will generally be
higher than those based on the method of Del Corral et al. (4)
Inclusion of exercise during weight loss may attenuate the
decline in TEE, thereby yielding more similar adherence results between the two methods.
AJP-Endocrinol Metab • doi:10.1152/ajpendo.00290.2011 • www.ajpendo.org
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.6 on June 14, 2017
8.9% by days 6 and 7 of a 58% CR diet, we observed a 9.6%
decrease in TEE by weeks 2– 4 of a 25% CR diet. However,
whereas 24-h energy expenditure continued to decline and
reached 85% of the baseline value at week 6 in the study of de
Boer et al. (2), TEE in the current study remained at 90% of
baseline TEE at months 3 and 6, which was likely due to the
more mild level of CR.
The reference approach for computing average 6-mo EI and
%CR in the current study included measurement of TEE at four
time points (baseline and months 1, 3, and 6) and changes in
body composition over the 6-mo period of CR. We compared
this approach with several alternative approaches that included
fewer TEE assessments or different approaches for computing
changes in body energy stores. Because TEE declined rapidly
upon initiation of CR, inclusion of a TEE measurement during
month 1 provided the best estimate of average energy intake
and %CR during the 6-mo CR intervention. As expected,
exclusion of both month 1 and month 3 TEE results yielded
discrepant estimates of EI and %CR compared with the reference approach. In the absence of interim TEE measurements,
average EI and %CR can be estimated by weighting the
baseline and final TEE values to account for the rapid decrease
in TEE that occurs with CR. In the current study, weighting the
baseline and final TEE values as 1⁄6 and 5⁄6 of the average month
6 TEE, respectively, produced EI and %CR values that were
similar to the reference approach. However, this assumption
would not be valid in cases where TEE either declines continuously during the CR intervention or increases due to weight
regain or engagement in higher levels of physical activity.
Furthermore, because individuals differ in the degree to which
TEE declines in response to CR (12, 29), as well as in the time
course of response, measuring TEE at an interim time point
enables the most valid individual estimate of EI.
Estimates of long-term changes in energy stores based on
body composition and body weight yielded group EI and %CR
estimates that were similar and highly correlated within individuals. This suggests that the easier and less costly weightbased approach can substitute adequately for body composition
measurements in some circumstances. However, for certain
populations or with interventions that combine CR and exercise, the actual ratio of changes in fat mass to changes in
fat-free mass may cause the true energy coefficient of weight
change to differ from 7.4 kcal/g. Furthermore, the relative
contributions of fat mass and fat-free mass to weight change
are influenced by sex, degree of adiposity, and type of exercise
intervention. In addition, during periods of weight regain (a
common phenomenon following weight loss), the ratio of
changes in fat mass to fat-free mass may differ from that during
weight loss.
We also explored the influence of using serial short-term
measurements of body weight and body composition to quantify changes in energy stores. The advantage of this approach
is that it enables one to compute EI at a single time point during
a weight loss intervention; the disadvantages include the potentially large errors in estimating daily change in body energy
stores and the assumption that changes in energy stores during
the 2-wk TEE assessment period accurately reflect changes that
occur between TEE assessments. As expected, regression of
daily body weights proved to be a better indicator of short-term
changes in body energy stores than body composition measurements at the beginning and end of each 2-wk TEE period
QUANTIFYING ENERGY INTAKE AND %CR DURING WEIGHT LOSS
ACKNOWLEDGMENTS
We acknowledge other CALERIE investigators and team members who
contributed to these studies: Pennington Biomedical Research Center: Andrew
Deutsch, Paula J. Geiselman, Frank Greenway, Leonie Heilbronn, Enette
Larson-Meyer, Michael Lefevre, Marlene Most-Windhauser, and Steven
Smith; Tufts University: Cheryl H. Gilhooly, Paul J. Fuss, and Gerard E.
Dallal; Washington University School of Medicine: John O. Holloszy, Hassan
Arif, Ali Ehsani, Luigi Fontana, Samuel Klein, Christopher Koenig, Kathleen
A. Obert, Kenneth B. Schechtman, Morgan Schram, Karen Steger-May,
Richard I. Stein, Dennis T. Villareal, Laura Weber, and Edward P. Weiss;
Duke Clinical Research Institute: Connie Bales. We would also like to
acknowledge helpful comments on DLW methods from William Wong, Baylor
College of Medicine.
GRANTS
This research was supported by National Institute on Aging Cooperative
Agreements U01-AG-020487, U01-AG-020478, U01-AG-020480, and U01AG-022132 and National Institutes of Health Grants MO1-RR-00036, P60DK-020579-30, and P30-DK-056341. L. M. Redman was supported by a Neil
Hamilton-Fairley Training Fellowship awarded by the National Health and
Medical Research Council of Australia (ID 349553). This trial was registered
at clinicaltrials.gov as NCT00099138, NCT00099151, and NCT00099099.
DISCLOSURES
None of the authors has a personal or financial conflict of interest.
AUTHOR CONTRIBUTIONS
S.B. Racette, S.K.D., E.C.H., S.B. Roberts, E.R., and J.P.D. did the
conception and design of the research; S.B. Racette, S.K.D., J.P.D., and L.R.
performed the experiments; S.B. Racette, M.B., J.P.D., and L.R. analyzed the
data; S.B. Racette, S.K.D., M.B., E.C.H., S.B. Roberts, E.R., C.P., J.P.D.,
W.E.K., J.R., and L.R. interpreted the results of the experiments; S.B. Racette
prepared the figures; S.B. Racette, S.K.D., E.C.H., and L.R. drafted the
manuscript; S.B. Racette, S.K.D., M.B., E.C.H., S.B. Roberts, E.R., C.P.,
J.P.D., W.E.K., J.R., and L.R. edited and revised the manuscript; all authors
approved the final version of the manuscript.
REFERENCES
1. Das SK, Gilhooly CH, Golden JK, Pittas AG, Fuss PJ, Cheatham RA,
Tyler S, Tsay M, McCrory MA, Lichtenstein AH, Dallal GE, Dutta C,
Bhapkar MV, Delany JP, Saltzman E, Roberts SB. Long-term effects
of 2 energy-restricted diets differing in glycemic load on dietary adherence, body composition, and metabolism in CALERIE: a 1-y randomized
controlled trial. Am J Clin Nutr 85: 1023–1030, 2007.
2. de Boer JO, van Es AJ, Roovers LC, van Raaij JM, Hautvast JG.
Adaptation of energy metabolism of overweight women to low-energy
intake, studied with whole-body calorimeters. Am J Clin Nutr 44: 585–
595, 1986.
3. de Jonge L, DeLany JP, Nguyen T, Howard J, Hadley EC, Redman
LM, Ravussin E. Validation study of energy expenditure and intake
during calorie restriction using doubly labeled water and changes in body
composition. Am J Clin Nutr 85: 73–79, 2007.
4. Del Corral P, Chandler-Laney PC, Casazza K, Gower BA, Hunter
GR. Effect of dietary adherence with or without exercise on weight loss:
a mechanistic approach to a global problem. J Clin Endocrinol Metab 94:
1602–1607, 2009.
5. Hall KD. Predicting metabolic adaptation, body weight change, and
energy intake in humans. Am J Physiol Endocrinol Metab 298: E449 –
E466, 2010.
6. Hall KD, Chow CC. Estimating changes in free-living energy intake and
its confidence interval. Am J Clin Nutr 94: 66 –74, 2011.
7. Heilbronn LK, de Jonge L, Frisard MI, DeLany JP, Larson-Meyer
DE, Rood J, Nguyen T, Martin CK, Volaufova J, Most MM, Greenway FL, Smith SR, Deutsch WA, Williamson DA, Ravussin E. Effect
of 6-month calorie restriction on biomarkers of longevity, metabolic
adaptation, and oxidative stress in overweight individuals: a randomized
controlled trial. JAMA 295: 1539 –1548, 2006.
8. Holloszy JO, Fontana L. Caloric restriction in humans. Exp Gerontol 42:
709 –712, 2007.
9. Institute of Medicine, Food, and Nutrition Board. Dietary Reference
Intakes for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol,
Protein, and Amino Acids. Washington, DC: The National Academies,
2005.
10. Johnson RK, Soultanakis RP, Matthews DE. Literacy and body fatness
are associated with underreporting of energy intake in US low-income
women using the multiple-pass 24-hour recall: a doubly labeled water
study. J Am Diet Assoc 98: 1136 –1140, 1998.
11. Larsson CL, Westerterp KR, Johansson GK. Validity of reported
energy expenditure and energy and protein intakes in Swedish adolescent
vegans and omnivores. Am J Clin Nutr 75: 268 –274, 2002.
12. Leibel RL, Rosenbaum M, Hirsch J. Changes in energy expenditure
resulting from altered body weight. N Engl J Med 332: 621–628, 1995.
13. Martin LJ, Su W, Jones PJ, Lockwood GA, Tritchler DL, Boyd NF.
Comparison of energy intakes determined by food records and doubly
labeled water in women participating in a dietary-intervention trial. Am J
Clin Nutr 63: 483–490, 1996.
AJP-Endocrinol Metab • doi:10.1152/ajpendo.00290.2011 • www.ajpendo.org
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.6 on June 14, 2017
Recently, comprehensive physiological models have been
developed to predict changes in body weight and body composition in response to CR. These models have also been used
to estimate the changes in EI that account for an observed
degree and time course of weight loss for individual subjects
and for groups (5, 25, 26). These approaches have great
potential value for gauging individual adherence during a CR
intervention, where the complexities of EI estimates from TEE
and changes in body composition in real time are limited.
Furthermore, the individual estimates of EI in these models
require validation in larger numbers of subjects. Therefore,
quantification of EI from assessments of TEE and changes in
body energy stores is the preferred approach for estimating
%CR across a variety of populations. TEE-based energy intake
data from a wider range of CR and weight-loss trials could be
used to refine and extend physiological models to improve their
ability to estimate EI from weight changes.
A limitation of the current study is the absence of a comparative measurement of EI that is 100% accurate and precise.
Although the DLW method is the gold standard for quantifying
TEE and energy intake in free-living individuals (9), the
relatively small analytical and physiological errors result in a
combined error of ⬃5% (28), which may be magnified when
EI is computed during periods of weight loss or weight regain.
The duplicate baseline DLW periods and the stringent procedures for isotope dosing, sample acquisition, isotope analyses,
and body weight measurement in the current study were
designed to minimize errors and optimize the accuracy and
precision of this method.
In summary, numerous models may be used to quantify EI
and %CR during CR interventions. We presented eight approaches that incorporate TEE measured by DLW and changes
in body energy stores computed six different ways. The optimal approach for quantifying average EI and %CR during a
6-mo CR intervention relies on TEE measurements at baseline
and during the first few weeks of CR to capture the rapid
reduction in daily energy expenditure; long-term changes in
body energy stores can be computed from body composition or
body weight. The most reliable approach for determining
free-living energy intake (and %CR) at a single time point
during active weight loss relies on a TEE measurement combined with an estimate of daily change in body energy stores
computed from either regression of daily weights over a 4-wk
period or a change in body composition during the preceding
months. These long- and short-term approaches can be used to
determine adherence to weight loss interventions and to enhance the interpretation of results when actual %CR differs
from the level prescribed.
E447
E448
QUANTIFYING ENERGY INTAKE AND %CR DURING WEIGHT LOSS
21. Roberts SB. Use of the doubly labeled water method for measurement of
energy expenditure, total body water, water intake, and metabolizable
energy intake in humans and small animals. Can J Physiol Pharmacol 67:
1190 –1198, 1989.
22. Rumpler WV, Seale JL, Miles CW, Bodwell CE. Energy-intake restriction and diet-composition effects on energy expenditure in men. Am J Clin
Nutr 53: 430 –436, 1991.
23. Schoeller DA. The energy balance equation: looking back and looking
forward are two very different views. Nutr Rev 67: 249 –254, 2009.
24. Schoeller DA. Measurement of energy expenditure in free-living humans
by using doubly labeled water. J Nutr 118: 1278 –1289, 1988.
25. Thomas DM, Martin CK, Heymsfield S, Redman LM, Schoeller DA,
Levine JA. A simple model predicting individual weight change in
humans. J Biol Dyn 5, 1–21, 2011.
26. Thomas DM, Schoeller DA, Redman LA, Martin CK, Levine JA,
Heymsfield SB. A computational model to determine energy intake during
weight loss. Am J Clin Nutr 92: 1326 –1331, 2010.
27. Trabulsi J, Schoeller DA. Evaluation of dietary assessment instruments
against doubly labeled water, a biomarker of habitual energy intake. Am J
Physiol Endocrinol Metab 281: E891–E899, 2001.
28. Trabulsi J, Troiano RP, Subar AF, Sharbaugh C, Kipnis V, Schatzkin
A, Schoeller DA. Precision of the doubly labeled water method in a
large-scale application: evaluation of a streamlined-dosing protocol in the
Observing Protein and Energy Nutrition (OPEN) study. Eur J Clin Nutr
57: 1370 –1377, 2003.
29. Velthuis-te Wierik EJ, Westerterp KR, van den Berg H. Impact of a
moderately energy-restricted diet on energy metabolism and body composition in non-obese men. Int J Obes Relat Metab Disord 19: 318 –324,
1995.
AJP-Endocrinol Metab • doi:10.1152/ajpendo.00290.2011 • www.ajpendo.org
Downloaded from http://ajpendo.physiology.org/ by 10.220.33.6 on June 14, 2017
14. McCrory MA, Gomez TD, Bernauer EM, Molé PA. Evaluation of a
new air displacement plethysmograph for measuring human body composition. Med Sci Sports Exerc 27: 1686 –1691, 1995.
15. Neuhouser ML, Tinker L, Shaw PA, Schoeller D, Bingham SA, Horn
LV, Beresford SA, Caan B, Thomson C, Satterfield S, Kuller L, Heiss
G, Smit E, Sarto G, Ockene J, Stefanick ML, Assaf A, Runswick S,
Prentice RL. Use of recovery biomarkers to calibrate nutrient consumption self-reports in the Women’s Health Initiative. Am J Epidemiol 167:
1247–1259, 2008.
16. Racette SB, Schoeller DA, Luke AH, Shay K, Hnilicka J, Kushner RF.
Relative dilution spaces of 2H- and 18O-labeled water in humans. Am J
Physiol Endocrinol Metab 267: E585–E590, 1994.
17. Racette SB, Weiss EP, Villareal DT, Arif H, Steger-May K, Schechtman KB, Fontana L, Klein S, Holloszy JO. One year of caloric
restriction in humans: feasibility and effects on body composition and
abdominal adipose tissue. J Gerontol A Biol Sci Med Sci 61: 943–950,
2006.
18. Redman LM, Heilbronn LK, Martin CK, Alfonso A, Smith SR,
Ravussin E. Effect of calorie restriction with or without exercise on body
composition and fat distribution. J Clin Endocrinol Metab 92: 865–872,
2007.
19. Redman LM, Heilbronn LK, Martin CK, de Jonge L, Williamson DA,
Delany JP, Ravussin E; Pennington CALERIE Team. Metabolic and
behavioral compensations in response to caloric restriction: implications
for the maintenance of weight loss. PLoS One 4: e4377, 2009.
20. Redman LM, Ravussin E. Caloric restriction in humans: impact on
physiological, psychological, and behavioral outcomes. Antioxid Redox
Signal 14: 275–287, 2011.