eating frequency, within-day energy balance, and adiposity
TRANSCRIPT
Georgia State University Georgia State University
ScholarWorks @ Georgia State University ScholarWorks @ Georgia State University
Nutrition Theses Department of Nutrition
Spring 3-23-2015
Eating Frequency, Within-Day Energy Balance, And Adiposity In Eating Frequency, Within-Day Energy Balance, And Adiposity In
Free-Living Adults Consuming Self-Selected Diets Free-Living Adults Consuming Self-Selected Diets
Ayla C. Shaw
Follow this and additional works at: https://scholarworks.gsu.edu/nutrition_theses
Recommended Citation Recommended Citation Shaw, Ayla C., "Eating Frequency, Within-Day Energy Balance, And Adiposity In Free-Living Adults Consuming Self-Selected Diets." Thesis, Georgia State University, 2015. doi: https://doi.org/10.57709/7041873
This Thesis is brought to you for free and open access by the Department of Nutrition at ScholarWorks @ Georgia State University. It has been accepted for inclusion in Nutrition Theses by an authorized administrator of ScholarWorks @ Georgia State University. For more information, please contact [email protected].
ACCEPTANCE
This thesis, EATING FREQUENCY, WITHIN-DAY ENERGY BALANCE, AND
ADIPOSITY IN FREE-LIVING ADULTS CONSUMING SELF-SELECTED
DIETS, by Ayla C. Shaw was prepared under the direction of the Master’s Thesis
Advisory Committee. It is accepted by the committee members in partial fulfillment
of the requirements for the degree Master of Science in the Byrdine F. Lewis School
of Nursing and Health Professions, Georgia State University. The Master’s Thesis
Advisory Committee, as representatives of the faculty, certify that this thesis has met
all standards of excellence and scholarship as determined by the faculty.
_____________________ ________________________
Megan A. McCrory, PhD Anita M. Nucci, PhD, RD, LD
Committee Chair Committee Member
______________________
Dan Benardot, PhD, RD, LD, FACSM
Committee Member
______________________
Date
AUTHOR’S STATEMENT
In presenting this thesis as a partial fulfillment of the requirements for the advanced
degree from Georgia State University, I agree that the library of Georgia State
University shall make it available for inspection and circulation in accordance with its
regulations governing materials of this type. I agree that permission to quote, to copy
from, or to publish this thesis may be granted by the professor under whose direction
it was written, by the Byrdine F. Lewis School of Nursing and Health Professions
director of graduate studies and research, or by me. Such quoting, copying, or
publishing must be solely for scholarly purposes and will not involve potential
financial gain. It is understood that any copying from or publication of this thesis
which involves potential financial gain will not be allowed without my written
permission.
____________________________
Signature of Author
NOTICE TO BORROWERS
All theses deposited in the Georgia State University library must be used in
accordance with the stipulations prescribed by the author in the preceding statement.
The author of this thesis is:
Ayla C. Shaw
2975 Pine St
Duluth, GA 30326
The director of this thesis is:
Megan A. McCrory, PhD
Associate Professor
Department of Nutrition
Byrdine F. Lewis School of Nursing and Health Professions
Georgia State University
Atlanta, Georgia 30302
VITA
Ayla C. Shaw
ADDRESS: 2975 Pine St
Duluth, GA 30096
EDUCATION: M.S. 2015 Georgia State University
Health Sciences
B.S. 2008 University of Georgia
Statistics
PROFESSIONAL EXPERIENCE:
Graduate Research Assistant 2014-2015
Georgia State University, Atlanta, GA
Catastrophe Risk Analyst 2009-2015
Aspen Insurance, Atlanta, GA
PROFESSIONAL SOCIETIES AND ORGANIZATIONS:
American Dietetic Association 2014-present
Georgia Dietetic Association 2014-present
Oncology Nutrition Dietetic Practice Group 2014-present
AWARDS:
Graduation with Honors, Cum laude 2008
Dean’s List – University of Georgia 2005-2008
HOPE Scholarship Recipient 2004-2008
ABSTRACT
EATING FREQUENCY, WITHIN-DAY ENERGY BALANCE, AND ADIPOSITY
IN FREE-LIVING ADULTS CONSUMING SELF-SELECTED DIETS
By Ayla Shaw
Background: The relationship between eating frequency (EF) and adiposity is
surrounded by controversy. Numerous cross-sectional studies have been performed
on the subject, yet the results are mixed. While some of these studies show an inverse
relationship between EF and adiposity, this is likely due to underreporting of EF and
total energy intake when diets are self-reported. In studies where underreporting was
taken into account, EF is positively associated with both energy intake and adiposity.
Intervention trials have failed to show a significant effect of EF on energy intake or
weight change, but only a small number exist.
Objective: In this study, we examined associations among EF, energy intake, and
adiposity in free living adults consuming self-selected diets. In conducting this
analysis, two common methodological problems in this research area were addressed:
1) the lack of consideration of energy balance fluctuations throughout the day, and 2)
a tendency not to account for implausible reporting of energy intake. We
hypothesized that individuals with higher EF would have higher BMI, percentage
body fat, and energy intake. Additionally individuals with greater fluctuations in
energy balance will have higher BMI, percentage body fat, and EF.
Methods: We performed a secondary analysis of data collected as part of a previous
study in our laboratory on diet and energy regulation (unpublished). One hundred and
twenty-six participants were enrolled (62.4 % female, and 75.2% Caucasian), and one
participant dropped the study due to pregnancy. Mean ±SD age, BMI, and percentage
body fat of the remaining 125 participants were 29.8 ±12.2 years, 24.5±3.9 kg/m2,
and 27.8±9.8% respectively. We analyzed one day of dietary intake collected using a
multiple pass 24 hour recall. Energy intake was calculated by NDS (Nutrition Data
System for Research, version 2011 (n=36) and version 2010 (n=89)). An eating
occasion was defined as any occurrence of energy intake > 0 kcal separated by at
least 1 hour. EF was defined as the number of eating occasions per day. A
specifically designed spreadsheet that generates within-day energy balance was used
to produce estimates of hourly energy balance. We also used total energy expenditure
measured by doubly labeled water and the Huang et al. (2005)1 method to identify
implausible reporters (cutoff for plausibility was reported energy intake (REI) within
±16.8% of TEE) and conducted Pearson’s correlations and regression analysis in
both the total sample and a subsample in which implausible energy intake reporters
were excluded from analysis.
Results: We identified 59.2% of the sample as implausible reporters (n=74; 47 under-
reported and 27 over-reported). Mean ±SD EF and energy intake were 4.7±1.5 and
2356±964 kcal in the total sample and 4.8±1.6 and 2371±689 kcal in the plausible
sample. In the total sample EF was positively correlated to energy intake among
women (r=0.244, p=0.032). No other significant relationships were observed between
EF and either energy intake, BMI, or percentage body fat, in the total or plausible
sample. In the total sample, maximum energy deficit > 400 kcal in a 24 hours period
was significantly and positively correlated with percentage body fat
(r=0.211,p=0.019) and negatively correlated with EF (r=-0.243, p=0.007) when
controlling for sex and age. Separating the sample by sex we observed significant
positive correlation between percentage body fat and maximum energy deficit in men
(r=0.382, p=0.009) but not in women. No significant relationships between
fluctuations in energy balance and percentage body fat were observed in the plausible
sample.
Conclusion: No evidence was found to suggest a relationship between EF and
adiposity. The significant positive relationships observed between maximum energy
deficit and adiposity in the total sample are consistent with previous findings. The
number of implausible reporters identified in our analysis supports that over and
under-reporting is a major issue associated with self-reporting of dietary intake.
EATING FREQUENCY, WITHIN-DAY ENERGY BALANCE, AND ADIPOSITY
IN FREE-LIVING ADULTS CONSUMING SELF-SELECTED DIETS
by
Ayla Shaw
A Thesis
Presented in Partial Fulfillment of Requirements for the Degree of
Master of Science in Health Sciences
The Byrdine F. Lewis School of Nursing and Health Professions
Department of Nutrition
Georgia State University
Atlanta, Georgia
2015
ii
ACKNOWLEDGMENTS
This thesis would never have been possible without the guidance and support
of all my professors, family, and friends. I would like to foremost thank Dr. Megan
McCrory for her patience and mentoring throughout this experience, without whom
this thesis would never have been completed. Thank you for taking me on and
enabling me to accomplish this goal. I would also like to recognize my committee
members, Dr. Dan Benardot and Dr. Anita Nucci. Thank you for your encouragement
and guidance throughout my time at Georgia State University. Additionally, I would
like to express my gratitude to Barbara Hopkins for her unwavering support and belief
in me to accomplish this goal. I have learnt a great deal from you all over the past few
years, and I will be forever grateful. Last, but certainly not least, I would like to thank
my parents who have always encouraged me to follow my dreams. I cannot thank you
enough for all your support throughout my life but especially in my decision to
change careers and pursue this dream. I love you both. I can’t wait to start this next
new and exciting period of my life. I would never have made it here without you all.
iii
TABLE OF CONTENTS
List of Tables………………………………………………………………………….iv
List of Figures………………………………………………………………………….v
Abbreviations and Definitions……………………………………………………….. vi
I. INTRODUCTION…………………………………………………………1
II. LITERATURE REVIEW……………………………………….…………5
Issues with Previous Research……………………………………………. 5
Investigating the Presence of Implausible Reporting ……………………6
EF and Energy Intake ………………………………….………………10
EF and Adiposity …………………………………….…………………..12
Within-Day Energy Balance ……………………………………..............13
III. METHODS ………………………………………………………………16
Data Collection and Protocol ……………………………………………16
Anthropometry and Body Composition …………………………………18
Resting Metabolic Rate …………………………………………………18
EF and Energy Intake ………………………………...………………….18
Physical Activity Level …………………………………………………19
Within-Day Energy Balance ……………………………………………..19
Identification of Plausible and Implausible Energy Intake Reporters……21
Statistical Analysis ………………………………………………………22
IV. RESULTS ..................................................................................................24
Descriptive Statistics ..................................................................................24
Comparison of pTEE and TEEDLW ............................................................25
EF, Energy Intake, and Adiposity ..............................................................27
Within-day Energy Balance and Adiposity...…………………………….28
V. DISCUSSION AND CONCLUSIONS .....................................................36
REFERENCES ............................................................................................................42
APPENDICES .............................................................................................................48
A………………………………………………………………………………48
B………………………………………………………………………………49
C………………………………………………………………………………52
iv
LIST OF TABLES
Table Page
1. Summary Of Cross-Sectional Studies Investigating The Correlation ………...7
Between EF, Energy Intake, and Adiposity
2. Effects of EF on Adiposity and Energy Intake in EF Intervention …………...9
Studies in which Participants are Advised to Follow a Low or High
EF Regimen
3. Participant Characteristics, Mean±SD ………………………………………25
4. Within-day Energy Balance Variables ………………………………………28
5. Summary of Significant Partial Correlations, Controlling for Age…….…….31
and Sex
6. Summary of Significant Partial Correlations by Sex, Controlling …………32
for Age
v
LIST OF FIGURES
Figure Page
1. Percent Energy Intake of EER by Association of EF to Energy Intake ……..11
for Studies in which Implausible Reporters were Included
2. Percent Energy Intake of EER by Association of EF to Energy Intake ……. 11
for Studies in which Implausible Reporters were Excluded
3. Percent Energy Intake of EER by Association of EF to Adiposity ………….11
for Studies in which Implausible Reporters were Included
4. Percent Energy Intake of EER by Association of EF to Adiposity ………….11
for Studies in which Implausible Reporters were Excluded
5. Bland Altman Plot of Mean TEE by Difference in TEEDLW-pTEE …………26
for Total Sample with DLW
6. Bland Altman Plot of Mean TEE by Difference in TEEDLW-pTEE………… 27
for Plausible Sample with DLW
7. Percentage Body Fat by Maximum Energy Deficit (kcal/d) in the ………… 33
Total Sample
8. EF by Maximum Energy Deficit (kcal/d) in the Total Sample ………………34
9. Percentage Body Fat by Maximum Energy Deficit (kcal/d) for Men in …….35
the Total Sample
10. EF by Maximum Energy Deficit (kcal/d) for Women in the Total ………….35
Sample
vi
ABBREVIATIONS AND DEFINITIONS
Adiposity Consisting of, resembling, or relating to fat;
obesity
Anabolic State Metabolic state that occurs when the body has
excess energy to use for cell growth and
development. Defined by an energy balance > 0
ANOVA Analysis of Variance
BEE Basal Energy Expenditure
BMI Body Mass Index, weight(kg) divided by
height(m) squared
Catabolic State Metabolic state that occurs when the body has
inadequate energy and the body must
breakdown tissue to provide energy. Defined by
an energy balance less than 0
CV Coefficient of Variance
DLW Doubly Labeled Water
DRI Dietary Reference Intakes
EER Estimated Energy Requirement
EF Eating Frequency
EI Energy Intake
End-of-Day Energy Balance Difference between total energy intake and total
energy expenditure in a 24 hour period
Energy Balance Calories consumed minus calories expended
Energy Deficit Negative energy balance where expenditure
exceeds intake
Energy Surplus Positive energy balance where intake exceeds
expenditure
Hourly Energy Balance Energy balance divided into one hour intervals.
Ours is an estimate of hourly energy balance
because we did not have the exact time of
energy expenditure, and used estimates of PAL
to determine energy expenditure for each hour
vii
Lean Mass to Height Ratio Ratio of lean mass per unit of height, calculated
by dividing height (cm) by fat mass (kg) so that
individuals with higher lean mass will have a
higher ratio
NHANES National Health and Nutrition Examination
Survey
NutriTiming Nutrition analysis software used for assessing
within-day energy balance, divided into 24 one
hour segments, as well as macronutrient and
micronutrient intakes
PAL Physical Activity Level
REI Reported Energy Intake
RMR Resting Metabolic Rate
SD Standard Deviation
All Surpluses and Deficits Sum of the absolute value of all energy
surpluses and deficits, > 400 kcal, in a 24 hour
period
Max Surplus and Deficit Sum of the maximum energy surplus, > 400
kcal, and maximum energy deficit, > 400 kcal
TEE Total Energy Expenditure
Within-Day Energy Balance Energy balance that is assessed throughout the
day rather than in 24 hour intervals
1
CHAPTER I
EATING FREQUENCY, WITHIN-DAY ENERGY BALANCE, AND ADIPOSITY
IN FREE-LIVING ADULTS CONSUMING SELF-SELECTED DIETS
Introduction
The obesity epidemic occurring in the United States has inundated our popular
culture with countless weight loss programs and diet fads, and mainstream media
promotes the idea that eating more frequently leads to lower energy intake and better
weight control. This notion could lead one to infer that people who eat more
frequently have a lower BMI and percentage body fat. However, the true relationship
between eating frequency (EF) and adiposity is surrounded by controversy. While
some cross-sectional studies show an inverse relationship between EF and adiposity,
this is likely due to implausible reporting of EF and total intake2. The idea that EF
leads to lower energy intake is based on the suggestion that eating more frequently
leads to better appetite control. However, recent intervention studies provide little
evidence to support this theory. Similarly, as recently reviewed3, when compared to
the traditional 3 meals a day eating pattern, eating > 3 meals a day offers minimal to
no improvements in appetite control and food intake while eating < 3 meals a day
appears to negatively affect appetite control.
There are more mixed results concerning the impact of changing EF on
adiposity. For example, in a recent weight loss intervention for postpartum women
where EF was not manipulated as part of the intervention, EF was examined in
relation to energy intake and body weight4. The analysis revealed a positive
association between EF and energy intake before the intervention, while a decrease in
EF from baseline was associated with lower energy intake during the intervention
period. However, there were no significant associations of EF with weight before or
weight loss during the intervention period. In another trial in which individuals were
2
randomly assigned to eat either three or six times a day, EF had no significant effect
on weight loss5. Mixed findings such as these may suggest the existence of other
unmeasured factors contributing to adiposity. An alternate view of energy balance
proposed by Benardot6 may provide insight into some of these unmeasured
confounding variables.
Studies on the relationship between EF, energy intake and adiposity have
relied on the traditional view of energy balance, defined by the difference between
energy intake and expenditure within a 24 hour period. Under this model, energy
intake exceeding energy expenditure aggregated over a 24 hour period leads to weight
gain, whereas energy intake less than energy expenditure aggregated over a 24 hour
period leads to weight loss. However, ongoing research by Benardot (2013)6 suggests
this view of energy balance may be incomplete. He has taken a different approach by
evaluating within-day fluctuations in energy balance that may be contributing to
adiposity. His theory is based on evidence suggesting that delayed eating and/or large
intakes of energy may result in hyperinsulinemia, which inhibits the down regulation
of ghrelin, the appetite stimulating hormone6–9
. The prolonged presence of ghrelin
may lead to increased appetite, and therefore, either a larger intake of energy at the
next eating occasion, a shortened inter-meal interval, or both, thus increasing EF and
total daily energy intake. If increasing energy intake at any single eating occasion
results in a high positive energy balance (or energy surplus), this would lead to the
synthesis of body fat. Although energy surpluses can lead to increases in percentage
body fat, so can very low energy intakes. Energy intakes low enough to create an
energy deficit within a single eating occasion can result in low blood glucose levels
which stimulate cortisol production and up-regulate gluconeogenesis. According to
Benardot (2013), a likely result would be the breakdown of lean tissue to release
3
alanine for the alanine-glucose cycle in the liver. Large deficits in energy balance
stimulate muscle catabolism and high cortisol levels in athletes10,11
. A second way
large within meal energy deficits may lead to higher body fat may be explained by the
glucostatic theory of energy regulation, introduced by Mayer (1955)12
. According to
this theory, hypoglycemia is itself an appetite stimulus. While controversial, some
data exist to support this theory13
. It follows that multiple energy surpluses and
deficits within a day could have even greater effects on adiposity. In this
“microeconomic view of energy balance” Benardot (2013) has observed that
relatively large surpluses and deficits in energy balance, > 400 kcal, are associated
with a higher percentage body fat6.
In this study we examined associations among EF, energy intake, and
adiposity in free living adults consuming self-selected diets. In conducting this
analysis, we addressed two methodological problems rampant among existing
research in this area: 1) a tendency not to account for implausible reporting of energy
intake, and 2) the lack of consideration of energy balance fluctuations throughout the
day. Through an estimation of within-day energy balance values that were derived
from reported energy intake and estimations of energy expenditure, we aimed to use
both a microeconomic approach and the more traditional macroeconomic approach
for determining energy balance. We also used estimates of total energy expenditure
measured by doubly labeled water and the method of Huang et al1 to identify
implausible reporters and conduct the analysis in both the total sample and a
subsample in which implausible energy intake reports are excluded from analysis.
We hypothesize that once implausible reporters have been excluded from the
analysis that participants who ate more frequently would have higher BMI,
percentage body fat, and energy intake in a 24 hour period. We also expected that
4
participants with larger energy surpluses and deficits to have higher percentage body
fat and EF.
5
CHAPTER II
Review of Literature
According to one analysis of data from NHANES collected between 1971 and
2002, on average Americans had a mean±SD EF of 5.06±0.04 from 1999-2002, and
this has not significantly increased since 197114
. This study also observed significant
increases in reported energy intake per eating episode over the three decades from a
mean±SD of 410±4 kcal to 470±4 kcal14
. However, another study15
examining the
same data reported that there was an increase in EF (+1.1 occasions/d) and total
energy intake (+570 kcal/d) between 1977-78 and 2003-06. Both of these studies
observations are consistent with the increase in obesity levels observed in the United
States over this time period. In NHANES 1971-1975 15% of adults were identified as
obese, whereas 33% were identified as obese in NHANES 1999-200216
. This
percentage has continued to rise, with recent studies in 2011-2012 stating that 34.9%
of Americans are now classified as obese17
. These findings have inspired many
previous studies investigating the relationships among EF, energy intake, and obesity,
yet the relationship remains uncertain.
Issues with Previous Research
Several issues exist with the previous research, which may partially explain
the variability among studies. As mentioned before, self- reporting could lead to bias
in the data collected. Studies on the validity of self-reporting have found that
individuals on the extreme ends of the BMI scale tend to over or under report their
intake18,19
. In cross-sectional studies in which implausible reporting is taken into
account, EF has been positively associated with both energy intake and adiposity20,21
.
Randomized controlled trials using EF as an intervention are few but also
contain bias due to compliance being tracked by self-report. A review of controlled
6
studies on EF found that the existing studies were done over short time periods and
had small sample sizes22
. Many of these interventions also have no significant
difference in reported EF between the low and high frequency intervention groups.
Additionally, there is no standard for defining eating occasions with the result being a
variety of definitions among studies contributing to the diverse outcomes.
Investigating the Presence of Implausible Energy Intake- Reporting
We conducted an extensive review and analysis of data from the literature to
further understand the current information on the subject. References from McCrory
et al.2 tables were used as a basis from which to begin. Literature searches were
conducted using the GSU online library resources including PubMed and Medline
databases to collect any additional or recently published references through April
2015 that were not included in the McCrory et al.2 article. Criteria for the search
included: Cross-sectional studies evaluating the association between EF and energy
intake and/or EF and adiposity in free living adults eating self-selected diets;
longitudinal studies evaluating the relationship between changes in EF and changes in
energy intake over a defined period of time; and, intervention studies where energy
intake was not controlled and EF was the main intervention. Selected articles were
dated from September 1964 to April 2015.
Table 1 contains a summary of cross-sectional studies which have been
divided into two categories; studies which made no mention or did not account for
implausible reporting1,23–35
, and studies that used a method such as that described by
Huang et al.1 or Goldberg et al.
36 to identify and exclude potentially implausible
reporters from the analysis21,37–42
. These methods use an estimate of TEE compared to
reported energy intake in order to identify minimum cut-offs for determining
plausibility. For this literature review we used EER derived from DRI equations to
7
Table 1. Cross-sectional Studies Investigating the Association between EF, Energy Intake, and Adiposity
First Author (Year) Sex N Age,
y
Weight,
kg
Height,
m
BMI,
kg/m2
REI,
kcal
EER,
kcal
% REI
of EER
Association
of EF and
EI
Association of
EF and
adiposity
Implausible reporting not taken into account
Fabry (1964)23
M 379 60-64 n/r n/r n/r n/r n/r n/r n/r -
Metzner (1977)24
M 948 46.4 79.7 n/r n/r 2839 *2884 98% n/r -
F 1080 45.6 66.0 n/r n/r 1786 *2160 83% n/r -
Dreon (1988)25
M 155 44.3 95.0 1.79 29.8 2570 2950 87% NS NS
Edelstein (1992)26
M&F 2034 70.4 72.6 1.70 25.1 1812 2162 84% NS NS
Kant (1995)27
M 2580 44.5 n/r n/r 25.9 2414 *2820 86% + -
F 4567 45.9 n/r n/r 25.3 1526 *2212 69% + -
Ortega (1998)35
M&F 150 73.5 n/r n/r 26.3 1785 n/r n/r n/r - (women)
Wahlqvist (1999)28
M&F 293 70+ n/r n/r n/r n/r n/r n/r NS -
Amosa (2001)34
F 80 18-27 n/r n/r 29.9 n/r n/r n/r n/r NS
Titan (2001)29
M 6890 57.8 n/r n/r 26.4 2209 *2844 78% + -
F 7776 56.4 n/r n/r 25.9 1937 *2242 86% + +
Berteus Forslund
(2002)30
F 177 48.7 86.3 1.65 31.9 2579 2346 110% + +
Ma (2003)31
M 251 48.0 n/r n/r 28.6 2259 *2951 77% n/r -
F 248 48.0 n/r n/r 26.6 1641 *2264 72% n/r -
Berteus Forslund
(2005)32
M 2396 46.6 112.0 1.80 34.7 3135 3205 98% + +
F 2955 44.5 97.3 1.66 35.4 2592 2520 103% + +
Huang (2005)1 M&F 6499 46.5 n/r n/r 25.9 2000 2418 83% n/r NS
Berg (2009)33
M&F 3610 25-74 n/r n/r n/r 2185 n/r n/r n/r NS
Reicks (2014)43
M&F 2702 18-80 n/r n/r 26.8 n/r n/r n/r n/a -
Karatzi(2015)44
M&F 164 46.8 n/r n/r 27.0 1899 *2249 84% + -a
8
Implausible reporting taken into account
Summerbell (1996)37
M 71 13-91 n/r n/r 22.2 2082 *2608 80% NS NS
F 149 13-91 n/r n/r 23.7 1717 *2110 81% NS NS
Drummond (1998)38
M 42 37.4 79.6 1.77 25.3 2638 2764 95% NS -
F 37 34.0 59.3 1.61 22.8 1948 2088 93% + NS
Ruidavets (2002)39
M 330 45-64 n/r n/r 26.5 2395 *2849 84% n/r -
Howarth (2007)40
M&F 2685 49.3 n/r n/r 25.3 2277 2382 96% + +
Hartline-Grafton (2009)41
F 329 47.3 n/r n/r 29.1 1833 *2079 88% + NS
Mills (2011)42
F 1099 49.6 n/r n/r 27.7 2129 *2284 93% + NS
Murakami (2014)21
M 678 42.4 n/r n/r 27.3 2351 *2888 81% + +
F 809 42.4 n/r n/r 26.8 1657 *2271 73% + +
Aljuraiban (2015)45
M&F 2385 48.9 n/r n/r 28.3 2340 *2369 99% - -
EER = estimated energy requirement; REI = reported energy intake; BMI = body mass index; EF = eating frequency; EI = energy intake
*EER estimated using DRI Tables; n/r = values or information not reported or unable to calculate values due to missing information; NS = not significant a= study did not make adjustments for physical activity or age
9
Table 2. Effects of EF on Adiposity and Energy Intake in EF Intervention Studies in which Participants are Advised to Follow a Low or
High EF Regimen
1st Author
(year)
Design N Sex Age,
y
BMI,
kg/m2
Low
EF
Goal,
times/d
High
EF
Goal,
times/d
Reported
Low EF,
times/d
Reported
High EF,
times/d
Difference in
change in
body weight
between
groups
Difference in
Energy Intake
between
groups
Arnold
(1993)46
2-wk
randomized
crossover
19 M&F 32.1 23.1 3 9 3.2 8.3 NS NS
Arnold
(1994)47
4-wk
randomized
crossover
16 M&F 49.9 26.5 3 9 3.2 8.2 NS NS
McGrath
(1994)48
3-wk switch
from normal
23 M 29.6 24.1 3 6 3.3 5.9 n/r n/r
Arnold
(1997)49
4-wk
randomized
crossover
13 M&F 46-
70
29.9 3 9 3.1 7.9 NS n/r
Thomsen
(1997)50
2-wk
randomized
crossover
10 M&F 60 28.1 3 8 n/r n/r NS High EF
reported higher
EI
Berteus
Forslund
(2007)5
1-yr
randomized
parallel
93 M&F 41.2 39.1 3 6 3.6 5.3 NS NS
Bachman
(2012)51
6-mo
randomized
controlled
51 M&F 51 35.5 3 GR* 3 5.8 NS NS
BMI = body mass index; EF = eating frequency; GR = grazing, eating 100 kcal every 2-3 hours; n/r = not reported; NS = not significant
10
estimate TEE. Average EER of the participants has been either provided by the study
or estimated using several methods. If a study provided the sex, average age, height,
and weight of participants, the appropriate DRI equations for a combined non-obese
and obese population were used to calculate EER. If any of these values were
missing, EER was estimated using DRI tables for 30 year old males or females with
assumed population average heights of 1.75m for males and 1.625m for females, and
a low active physical activity level, unless specifically stated otherwise in the article.
An even distribution between BMI data points was assumed to extrapolate EER
values. If not enough information was provided or too many assumptions had to be
made, EER was recorded as not reported. The proportion of average reported energy
intake (REI) to EER was calculated to determine which studies have the highest
probability of implausible reporting. Scatterplots of the percent REI of EER by study
outcome (positive correlation, negative correlation, or not significant) are presented in
Figures 1 through 4 to visualize the relationship between the likelihood of implausible
reporting and the reported correlation between either adiposity or energy intake. Once
again the studies were divided by whether or not implausible reporters were excluded
from analysis. Table 2 summarizes the intervention studies where EF was the primary
intervention.
EF and Energy Intake
The evidence suggests that if a relationship between EF and energy intake
exists it is positive. However, a cross-sectional study comparing eating habits of
weight loss maintainers (WLM) to a sample of the general population found that
WLM had significantly higher EF than the general population53
. From Figures 1 and
2 one can observe only one study45
reporting a negative relationship and more
positive associations (ten studies)21,27,29,30,32,38,40–42,44
are seen than
11
Figures 1 and 2. Percent Energy Intake of EER by association of EF to Energy Intake for studies in which implausible reporters were included vs excluded
Figures 3 and 4. Percent Energy Intake of EER by association of EF to Adiposity for studies in which implausible reporters were included vs excluded
EF = eating frequency; REI = reported energy intake; EER = estimated energy requirement
0.6
0.7
0.8
0.9
1
%R
EI
of
ER
R
Negative No Significance Positive
1. EF v Energy Intake
Implausible Reporters Included
Mean
0.6
0.7
0.8
0.9
1
%EI
of
ERR
Negative No Significance Positive
2. EF v Energy Intake
Implausible Reporters Excluded
Mean
0.6
0.7
0.8
0.9
1
%EI
of
ERR
Negative No Significance Positive
3. EF v Adiposity
Implausible Reporters Included
Mean
0.6
0.7
0.8
0.9
1
%EI
of
ERR
Negative No Significance Positive
4. EF v Adiposity
Implausible Reporters Excluded
Mean
12
non-significant associations (five studies)25,26,28,37,38
. Also, the mean percent REI of
EER for studies observing no significant relationship and studies observing a positive
relationship, for both those including and excluding implausible reporters, appear to
be approximately equal. Independent t-tests were performed and there were no
significant differences between the means. This suggests that self-reporting bias was
either not present in the included analyses or any reporting bias which may have been
present was inconsequential when considering the relationship between EF and
energy intake. Only two longitudinal studies matching the criteria were found. One
study observed that the EF was positively correlated with energy intake at baseline,
and the reduction in EF during the intervention period was positively associated with
a reduction in energy intake during the same period4. The other study observed that
higher EF was a significant independent predictor of higher energy intake14
. Table 2
indicates that most of the EF intervention trials found no significant difference in
energy intake between low and high EF groups5,46–49
. The one study that did observe
a significant relationship found that individuals in the high EF group reported higher
energy intake 50
. The lack of significant relationships found in these studies may be
due to the methodological concerns with intervention trials previously discussed.
EF and Adiposity
While the literature generally supports that a relationship between EF and
energy intake exists, there is less certainty concerning the relationship between EF
and adiposity. Given a positive relationship between energy intake and EF we would
expect to see a positive relationship between EF and adiposity. Individuals with
higher BMI typically have higher energy requirements and would require more
energy to meet them. If two individuals have the same TEE, then we would expect to
observe a higher BMI with higher EF and energy intake. Figures 3 shows a greater
13
number of studies reporting a negative correlation between EF and adiposity when
implausible reporters are included (seven studies)23,24,27–29,31,35
versus Figure 4 where
they are excluded (three studies)38,39,45
. An examination of Figure 3, suggests that
studies with negative correlations have a lower mean percent REI of EER than studies
with positive correlations. However, referencing Figure 4, where implausible reports
have been excluded, the mean percent REI of EER in studies with positive
correlations is higher. This observation does not support the hypothesis that when
implausible reporters are excluded from the analysis the relationship between EF and
adiposity is positive. However, one-way ANOVA analyses were performed and found
no significant difference in mean percent REI of EER among the studies observing
negative, positive, and non-significant relationships in either figure. In contrast, a
longitudinal study by Kant and Graubard (2006) offers support to the hypothesis after
observing that the inverse association of number of eating episodes with the
likelihood of obesity was not significant after exclusion of low-energy reporters14
.
From Table 2, none of the intervention trials surveyed found a significant effect of EF
on body weight. One potential reason for this may be because there is no effect.
Another reason for the absence of significance may be due to the issues with current
intervention trials previous discussed or a result of confounding factors not considered
in previous studies. Potential confounding factors may include physical activity as
well as peaks and deficits in energy balance as proposed by Benardot6.
Within-Day Energy Balance
Due to the recent development of the theory, few studies exist that evaluate
the impact of within-day fluctuations in energy balance on adiposity. The existing
studies have been all been cross-sectional and performed on relatively small
populations with similar anthropometric and socioeconomic characteristics. A study
14
performed on elite gymnast and runners (n=62) observed a significant positive
relationship between the number of energy deficits (>300 kcal) in a 24 hour period
and percentage body fat54
. This study also observed a significant negative relationship
between the largest daily energy surplus (> 300 kcal) and percentage body fat in
gymnasts. In April 2014, the International Olympic Committee released a consensus
statement acknowledging the importance of avoiding “Relative Energy Deficiency in
Sport” (RED-S) by maintaining proper energy balance during physical activity55
. A
recent unpublished study assessing U.S. female national team figure skaters (n=10)
found a significant association between the hours spent in an energy deficit (>400
kcal) and percentage body fat6. Higher percentage body fat was correlated with a
greater amount of time spent in an energy deficit. This study also observed that the
end-of-day energy balance was not significantly related to percentage body fat.
Another small study (n=12) performed on children and adolescents (ages 8-14)
observed similar findings56
. Once again, the end-of-day energy balance was not
associated with percentage body fat, and a greater time spent in energy deficit was
significantly associated with higher percentage body fat. A study performed on both
male and female college students (n=31, ages 18-30) observed hours spent in an
energy deficit (>400 kcal) was positively associated with percentage body fat, while
hours spent in an optimal energy balance (>-400 kcal and <400 kcal) was negatively
associated with percentage body fat57
. Hours in an anabolic state were negatively
associated with percentage body fat and, conversely, hours in a catabolic state were
positively associated with percentage body fat. Another study performed on college
students (n=17, ages 20-28) also found that a greater number of hours spent in an
energy deficit (>400 kcal) was associated with higher percentage body fat, and a
greater number of hours an optimal energy balance was associated with lower
15
percentage body fat58
. The combined results of these studies support the notion that
the traditional 24 hour view of energy balance is inaccurate in assessing body
composition in athletes, adolescents, and young adults. To this date, no other study
assessing the role of within-day energy balance and adiposity in free living adults
consuming self-selected diets is known.
16
CHAPTER III
Methods
Data Collection and Protocol
The present study was a secondary analysis study using data from a previous
study on diet and energy regulation in adults conducted by Dr. Megan A McCrory,
PhD, funded by the National Institutes of Health (NIH R01 DK07582). This was a
Purdue University Ingestive Behavior Research Center postdoctoral fellowship. The
study took place at the Department of Nutrition Science at Purdue University in West
Lafayette, IN from August 2010 to May 2012. It was a 10 day observational study.
All procedures were performed by trained and experienced research personnel. The
Institutional Review Board at Purdue University approved the study and all
participants provided written informed consent prior to participation.
Participants were recruited by fliers placed in academic buildings and at retail
centers; on the internet via the campus newsletter
(http://www.purdue.edu/newsroom/purduetoday/); on Craigslist (www.craigslist.org);
and by contacting individuals who had participated in previous studies conducted in
the McCrory laboratory at Purdue University or who had participated in Healthy
Purdue (Purdue University’s faculty/staff wellness program), and who had consented
to be contacted about future studies. One hundred and twenty-six men and women
between the ages 18-65 y with a BMI of ≥18.5 and ≤45 kg/m2 were originally
recruited into the study.
Exclusion criteria included: Smoking or tobacco use within the past 6 months,
≥5 lbs weight loss or gain within the past 6 months, pregnancy or lactation within the
past 1 year, consuming >3 alcoholic beverages per day, physical activity ≥12 hours
per week, chronic disease (cardiovascular disease, diabetes, HIV, cancer, and
17
metabolic abnormalities such as thyroid disorders), eating disorders, mental disorders
(such as depression) and use of medications or dietary supplements that would affect
energy regulation, appetite, mood or taste.
Potential participants were prescreened for eligibility by a telephone interview
and detailed health questionnaire. Qualified individuals were scheduled for an in-
person health screening visit which included anthropometric measurements, a
pregnancy test, blood pressure, and a blood sample. Blood was analyzed for a
standard metabolic panel (including glucose, blood lipids, and kidney and liver
enzymes), a thyroid test and a complete blood count. Those individuals who qualified
for the study after the study physician’s approval were invited to participate.
Qualified individuals were scheduled to enroll in the 10-day study within two
months of the initial health screening. Participants arrived at the research center after
an overnight fast ≥10 hours. Anthropometric and body composition measurements
were taken, and after a provided standard snack was consumed, questionnaires and
interviews (including an unscheduled dietary recall) were administered. After the
initial visit, two more unscheduled 24-hour dietary recalls were conducted via
telephone within a ten day period. Physical activity interviews took place at the
research center on day 0, 5 (or 6) and day 10. Anthropometry and body composition
measurements were taken again on day 10.
During the initial visit, individuals completed a questionnaire and self-
reported their ethnicity as black or African, white, Alaskan native or American
Indian, Asian, native Hawaiian or Pacific Islander, or other by questionnaire. For the
purpose of this analysis, the participants were then classified as either white (non-
Hispanic) or non-white due to the majority of the sample being white.
18
Anthropometry and Body Composition
Body mass and body composition were assessed by using air displacement
plethysmography (BOD POD Gold Standard Body Composition Measurement
System (Cosmed USA, Concord, CA)). The procedure and validation of this method
are described elsewhere59,60
. Body weight was measured to the nearest 0.01 kg on the
scale accompanying the BOD POD. A wall-mounted stadiometer was used to
measure height to the nearest 0.1 cm. Body mass index (BMI) was calculated as mass
(kg) divided by height (m) squared.
Resting Metabolic Rate
RMR was calculated using the DRI predicted basal energy expenditure (BEE)
equation61
, using sex, weight, height and age.
EF and energy intake
Dietary intake was assessed by multiple-pass 24 hour dietary recalls62
. A food
portion visual (2D Food Portion Visual; Nutrition Counseling Enterprises,
Framingham, MA) and measuring cups/spoons were used to estimate portion sizes
consumed. When conceivable participants were asked to clarify any details and to
give additional information deemed necessary. Energy and nutrient intake were
calculated by NDS (Nutrition Data System for Research, version 2011 (n=36) and
version 2010 (n=89)). For this study, we used the first multiple-pass 24 hour dietary
recall for each participant collected at the initial visit to the research center. Eating
occasions were defined as any occurrence of food or beverage intake > 0 calories. If
two or more eating occasions were reported within 59 minutes of each other, they
were combined. An example of the multiple pass 24 hour dietary recall form used is
included in Appendix A.
19
Physical Activity Level (PAL)
During the first visit, a Stanford Seven-Day Physical Activity Recall63
was
performed. The seven-day physical activity recall is a reliable measure of group total
daily energy expenditure and physical activity energy expenditure when results are
compared to TEE measured by doubly labeled water (DLW)64
. A copy of the
questionnaire is included in Appendix B. Participants were asked to recall the
approximate times of day they fell asleep, woke up, and performed physical activity
or exercise for at least 10 minutes over the past week. For the purposes of this study,
we only used the one day of activity recall that corresponded to the equivalent date
from the initial multiple-pass 24 hour dietary recall. The exact times of physical
activity were not provided from the recall, and were instead categorized as morning,
afternoon, or evening. We calculated PALs for each time period in order to predict
hourly energy expenditure. The morning time period was defined from when the
participant woke up until noon. The afternoon time period was defined as noon to
5pm, and the evening was defined as 5pm until bedtime. The number of hours of
moderate, hard, and very hard activity were determined for each time period, and the
rest of the time was categorized as light activity. MET values were assumed for sleep,
light, moderate, hard and very hard activities (1, 1.5, 3, 6 and 10 respectively). For
each time period, the number of hours at each activity level was multiplied by the
appropriate MET value, summed, and then divided by the total number of hours in the
time period to determine PAL’s for morning, afternoon, and evening. This same
procedure was used to determine the PAL for the full 24 hours.
With-in Day Energy Balance
A specifically designed spreadsheet was used to estimate within-day energy
balance based on the method of NutriTiming software (NutriTiming LLC, Atlanta
20
GA). NutriTiming estimates hourly energy balance through a system that analyzes
energy consumed and expended in hourly units. A copy of the Excel input file used is
included in Appendix C. Energy intakes were entered into the spreadsheet using the
reported times of eating occasions from the initial multiple-pass 24 hour dietary
recalls. Energy intake was entered into the hour interval based on the start time of
each eating occasion. Energy intakes from any eating occurrences with start times in
the same hour were summed. The time of the last eating occasion the previous day
was not provided, so we used the time of the last meal on the multiple-pass 24 hour
dietary recall to estimate the starting energy balance at the beginning of the 24 hour
period. BEE from DRI equations was divided by 24 to determine the hourly REE,
which is the minimal energy expenditure for every hour in the spreadsheet. The times
each participant woke up and went to sleep were collected from the seven day PAR.
Since we did not have the exact times of physical activity, the PALs determined for
morning, afternoon, and evening were entered for each hour in the pre-determined
time periods. These were multiplied by the REE to estimate energy expenditure for
each hour in the day. After energy intake and estimated energy expenditure were
entered, an energy balance for each hour of the day was estimated and used to
calculate the values for energy surpluses and deficits throughout the twenty four hour
period.
Variables of interest were calculated from the estimated hourly energy balance
data. The maximum surplus was defined as the largest energy surplus, > 400 kcal, in
the 24 hour period. The maximum deficit was defined as the absolute value of the
lowest energy deficit, less than -400 kcal, in the 24 hour period. The maximum
energy imbalance was defined as the greater of the maximum surplus and maximum
deficit. The sum of the maximum surplus and maximum deficit (max surplus and
21
deficit), as well as the sum of the absolute values of all surpluses and deficits (all
surpluses and deficits), > 400kcal, in a 24 hour period were also calculated for use in
the analyses. We also calculated variables for the number of hours spent in an energy
surplus > 400 kcal, number of hours in an energy deficit less than -400 kcal, number
of hours in an anabolic state, and number of hours in a catabolic state.
Identification of Plausible and Implausible Energy Intake Reporters
For measuring TEE, a 10-day DLW study was conducted. After participants
had fasted overnight and after the collection of baseline urine specimens, a mixed
2H2
18O dose containing 0.15 g H2
18O /kg body weight and 0.07 g of
2H2O /kg body
weight was given orally on the morning of study day 1. Participants voided at 1.5 and
3 h after dosing. Post-dose urine samples were collected at 4.5 and 6 h, day 5 (or day
6) and day 10. A small, standardized snack was given after the 4.5h dose. Previous
research has found this to have no effect on the accuracy of the method65
. Sample
aliquots were frozen at -80 deg C until shipped overnight to the Marine Biological
Laboratory, Woods Hole, MA for analysis. Abundances of 2H2 and 18O in urine
samples and in dilutions of the isotope doses were measured in duplicate using
isotope-ratio mass spectrometry (GV Instruments Isoprime Continuous Flow, city,
state, country). TEE was calculated with the use of standard equations66
with a food
quotient value of 0.86 (normal mixed western diet). Please note that large errors in
respiratory quotient have a small effect on the error of calculations of TEE67
.
Physiologically plausible reports of 3d mean REI from multiple-pass 24h
dietary recalls were identified using previously established methods1,68
. However, the
procedures were modified so that instead of using a predicted energy requirement,
TEE measured by DLW was used. Seven participants did have accurate
measurements of TEE from DLW. For these participants energy requirement
22
estimated using the appropriate DRI equations for EER based on each participant’s
sex and BMI was used for TEE. Then, the ±1SD cutoff range for the comparison of
REI with TEE was calculated by using the formula √[(CV*REI/d)2 + (CV*mTEE)
2 ]
where the within-subject coefficient of variation (CV) for REI was calculated as
25.4%, the number of days of intake (d) was three, and the within-subject CV for
measured TEE (mTEE) was given a value of 8.2% from the literature as
recommended by Black69
. Thus, participants having a REI within ±16.8% of TEE
were considered physiological plausible energy intake reporters1.
Statistical analysis
Statistical analysis was performed using SPSS (version 22.0 SPSS. Inc.:
Chicago, IL). Histograms of continuous variables were computed to check for outliers
and any unusual distributions. Descriptive statistics were calculated (including but not
limited to: mean, 95% CI for mean, median, variance, SD, minimum, maximum,
interquartile range) for the total sample and broken out by sex and by plausible, over,
and under reporters. Independent t-tests were performed to look for significant
difference between men and women and implausible and plausible reporters. One-
way ANOVA’s with Tukey multiple comparison tests were performed to test for
differences between plausible, over, and under-reporters. Chi-squared analysis were
performed to test if differences in the proportion of values observed over 0 were
statistically significant between sexes and among plausible, over, and under reporters.
Bivariate and partial correlations were performed to test for associations between
variables. These were performed in the total sample and for men, women, plausible,
over, and under-reporters. Then scatterplots of the most significantly correlated
variables were observed for any obvious relationships or data points driving the
outcomes. Next linear regression analyses were performed to test for variables
23
explaining a significant proportion of the variance in percentage body fat, BMI,
energy intake, and EF. Regression analyses were also used to test for mediating
relationships between variables. Again, regression analyses were performed in the
total sample and for men, women, plausible, over, and under-reporters
Additionally, several analyses were performed to test the validity of the
methods used to estimate the PALs for morning, afternoon, and evening for both the
total and plausible sample. Predicted TEE (pTEE) for the full 24 hours was calculated
by multiplying BEE by the 24 hour PAL and compared to TEE measured from DLW
(TEEDLW). The seven participants who did not have TEE values from DLW were
excluded. Paired t-tests were performed to compare the difference in pTEE and
TEEDLW. Paired t-test were also performed to compare the difference in pTEE and
TEEDLW for plausible, under and over-reporters. These analyses were repeated for the
samples including the seven participants without TEEDLW. Linear regression analyses
were performed to determine if pTEE was significant in predicting TEEDLW. Finally,
Bland Altman plots of the mean difference in TEE assess by the two methods
(TEEDLW minus pTEE) versus the mean TEE assessed by the two methods for the
total sample and plausible sample were examined. For all analyses, a p-value of <0.05
was accepted as significant.
24
CHAPTER IV
Results
Descriptive Statistics
A summary of participant characteristics is provided in Table 3. In both the
total and plausible samples, men had significantly lower mean percentage body fat
and significantly higher mean percentage fat free mass and energy intake than
women. Men also had much more variability in their lean mass to height ratios, and
this ratio was significantly higher than that in women in the total sample. When the
characteristics between plausible and implausible reporters were compared, we
observed no significant differences in age, BMI, TEE, percentage body fat, PAL, EF,
REI, or REI as a percentage of TEE (REI/TEE x 100%). However, when the
implausible reporters were divided into over and under-reporters, we observed several
differences. Although the difference was not statistically significant, under-reporters
had the highest BMI and percentage body fat, while over-reporters had the lowest.
Under-reporters did have a significantly higher TEE than plausible and over-reporters.
The difference in TEE between plausible and over-reporters was not significant. As
expected, compared to plausible reporters, under-reporters had significantly lower
REI and REI/TEE, and over-reporters had significantly higher REI and REI/TEE x
100%.
Age was significantly and positively correlated with EF in the total (r=0.361,
p<0.001) and plausible samples (r=0.419, p=0.024). In the total sample, older
individuals had significantly higher percentage body fat (r=0.207, p=0.020).
Individuals with higher REI had significantly lower percentage body fat (r=-0.262,
p=0.003), and higher EF was significantly associated with lower TEE (r=-0.202,
p=0.024) in the total sample.
25
Table 3. Participant Characteristics, Mean±SD
Total Sample
Plausible
reporters
Over-
reporters
Under-
reporters
n 125 51 27 47
Female, % 62 65 52 66
Age, y 29.8±12 30.1±13 31.3±13 28.6±11
BMI, kg/m2 24.5±3.8 24.6±4.4 23.2±3.1 25.1±3.5
Body Fat, %
mass 27.8±9.8 28.5±10.4 24.0±9.8 29.2±8.8
Fat Free
Mass, kg 51.7±11.1 51.3±12.1 52.3±9.8 51.7±10.9
Lean Mass to
Height Ratio 11.3±12 12.3±17 12.8±6 9.3±4
TEE, kcal/d 2,425±537 2,380±504a 2,107±481
a 2,656±504
b
PAL, 24hr 1.47±0.17 1.46±0.19 1.49±0.17 1.47±0.15
EF 4.69±1.47 4.76±1.57 4.96±1.32 4.45±1.44
REI, kcal/d 2,356±964 2,371±689a 3,237±1,239
b 1,833±619
c
REI/TEE,% 100±42 101±27a 151±39
b 70±23
c
Total Sample Plausible Sample
Males Females Males Females
n 47 78 18 33
Age, y 30.2±12.6 29.6±12.0 35.0±14 27.5±11
BMI, kg/m2 24.9±3.5 24.2±4.0 25.5±3.7 24.1±4.7
Body Fat, %
mass 20.0±8.8
a 32.5±7.1
b 19.9±9.9
a 33.2±7.3
b
Fat Free
Mass, kg 63.6±7.0
a 44.5±5.4
b 65.4±6.5 43.6±5.8
Lean Mass to
Height Ratio 16.1±18
a 8.4±3
b 19.6±27 8.3±3
TEE, kcal/d 2835±505a 2178±384
b 2850±391
a 2123±351
b
PAL, 24hr 1.49±0.18 1.46±0.16 1.50±0.23 1.44±0.16
EF 4.47±1.5 4.82±1.5 4.94±1.7 4.67±1.5
REI, kcal/d 2866±1048a 2048±765
b 2746±523
a 2166±689
b
REI/TEE,% 105±44 97±41 97±17 103±32 BMI=Body Mass Index; TEE=Total Energy Expenditure; PAL=Physical Activity Level;
EF=Eating Frequency; REI=Reported Energy Intake; Means with different superscript letters
are significantly different from one another
Comparison of pTEE and TEEDLW
Excluding the seven participants missing TEEDLW, a significant difference
was observed in the total sample (n=118, p=0.006, mean±SD of pTEE-TEEDLW
=120±470 kcal) but not in the plausible sample (n=47, p=0.236) between pTEE and
TEEDLW. In the total sample, the mean difference divided out over the 24 hour period
equates to an average difference of 5 kcal per hour. Under-reporters had significantly
26
lower pTEE than TEEDLW (n=45, p<.0001, Mean±SD of pTEE-TEEDLW = -365±344
kcal). Over-reporters had significantly higher pTEE than TEEDLW (n=26, p=.004,
Mean±SD of pTEE-TEEDLW =241±384). The results of the analyses including the
seven participants missing TEEDLW produced similar results.
Figure 5. Bland Altman Plot of Mean TEE by Difference in TEEDLW-pTEE for
Total Sample with DLW
The results of the regression analysis showed for both the total sample
(TEEDLW= 882 + 0.668pTEE) and the plausible sample (TEEDLW= 1121 +
0.544pTEE) the slope was significant (p=values<0.001). The 95% confidence
intervals for the total (0.489;0.847) and plausible (0.285;0.804) samples showed both
slopes were significantly different from 0. Figures 5 and 6 show the Bland Altman
plots for the total and plausible samples. The plot of the plausible sample does not
display any concerning relationships. However, from the plot of the total sample, we
observed that pTEE is underestimated at higher values of TEE and overestimated at
lower values.
27
Figure 6. Bland Altman Plot of Mean TEE by Difference in TEEDLW-pTEE for
Plausible Sample with DLW
EF, Energy Intake, and Adiposity
No significant correlations between EF and REI, BMI, percentage body fat, or
lean mass to height ratio were observed in either the total or plausible sample, even
when controlling for age and sex. However, we did observe that the relationship
between EF and BMI was negative (r=-0.027, p=0.766) in the total sample and was
positive (r=0.153, p=0.282) in the plausible sample, but in neither case were these
relationships statistically significant. From the regression analysis, for both the total
and plausible samples, EF was not significant in accounting for variances in BMI,
percentage body fat, or REI. These relationships remained insignificant when
controlling for age and sex.
Running the correlation and regression models separately for sex, we observed
significant relationships between EF and energy intake in the total sample among
women. Controlling for age, EF was positively correlated with energy intake
(r=0.244, p=0.032). The regression analysis also observed a positive relationship
28
between EF and energy intake (β=0.26) when controlling for age. EF accounted for a
small but significant amount of the variance in energy intake (r2=5.95%, p=0.032). No
significant relationships were observed in men or in the plausible sample.
Within-day Energy Balance and Adiposity
Table 4. Within-day Energy Balance Variables
Total Sample Plausible
reporters
Over-
reporters
Under-
reporters
Max Surplus
kcal/d,
mean±SD
666±791 642±586a
1,425±1,108b
257±334c
% of individuals
with values >0 81 88
a 100
b 62
c
Max Deficit
kcal/d,
mean±SD
152±317 97±205a 19±65
a 289±436
b
% of individuals
with values >0 40 35
a 15
b 60
c
Max Energy
Imbalance
kcal/d,
mean±SD
791±749 707±546a 1,425±1,108
b 519±415
a
% of individuals
with values >0 98 100 100 96
Max Surplus
and Deficit
kcal/d,
mean±SD
819±751 739±556a
1,444±1,107b
546±423a
% of individuals
with values >0 98 100 100 96
All Surpluses
and Deficits
kcal/d,
mean±SD
6,815±9,242 5,785±7,090a
15,123±13,727b
3,161±3,713a
% of individuals
with values >0 98 100 100 96
Hours in
Surplus
>400kcal,
mean±SD
8.4±7.4 8.5±7.1a
15.3±6.0b
4.3±5.2c
% of individuals
with values >0 81 88
a 100
b 62
c
Hours in Deficit
>400kcal,
mean±SD
2.6±4.7 1.8±3.7a
0.3±0.8a
4.6±6.0b
% of individuals
with values >0 40 35
a 15
b 60
c
29
Hours in
Anabolic State,
mean±SD
15.4±7.2 15.9±6.3a
20.9±3.4b
11.8±7.7c
% of individuals
with values >0 97 98 100 94
Hours in
Catabolic State,
mean±SD
8.6±7.2 8.1±6.3a
3.1±3.4b
12.2±7.7c
% of individuals
with values >0 83
84
a 63
b 94
c
Total Plausible
Males Females Males Females
Max Surplus
kcal/d,
mean±SD
881±975a
537±628b
680±576 622±598
% of individuals
with values >0 83 79 89 88
Max Deficit
kcal/d,
mean±SD
233±411a
104±234b
162±292 61±130
% of individuals
with values >0 55
a 31
b 50 27
Max Energy
Imbalance
kcal/d,
mean±SD
1,062±895a
628±593b
782±519 666±564
% of individuals
with values >0 100 97 100 100
Max Surplus
and Deficit
kcal/d,
mean±SD
1,114±838a
641±598b
842±52 683±57
% of individuals
with values >0 100 97 100 100
All Surpluses
and Deficits
kcal/d,
mean±SD
9,637±11,340a
5,115±7,277b
6,175±6,347 5,572±7,550
% of individuals
with values >0 100 97 100 100
Hours in
Surplus
>400kcal,
mean±SD
9.7±7.9 7.6±7.0 9.5±8.0 7.9±6.6
% of individuals
with values >0 83 80 89 88
30
Hours in Deficit
>400kcal,
mean±SD
3.5±5.4 2.0±4.1 2.8±4.6 1.3±3.0
% of individuals
with values >0 55
a 31
b 50 27
Hours in
Anabolic State,
mean±SD
14.8±7.8 15.8±6.9 15.6±7.3 16.1±5.8
% of individuals
with values >0 94 62 94 100
Hours in
Catabolic State,
mean±SD
9.2±7.8 8.2±7.3 8.4±7.3 7.9±5.8
% of individuals
with values >0 81 85 83 85
Means and percentages with different superscript letters are significantly different from one
another.
Table 4 provides the mean and SD of the energy balance variables, as well as
the percentage of participants that had observed values > 0. Under-reporters had
significantly lower mean maximum surplus and higher mean maximum deficit than
plausible and over-reporters. Over-reporters had significantly higher maximum
surplus, maximum energy imbalance, max surplus and deficit, and all surpluses and
deficits than under and plausible reporters. Under-reporters spent significantly fewer
and over-reporters spent significantly more hours in a catabolic state and hours in an
energy deficit > 400 kcal than plausible reporters, whereas, under-reporters spent
significantly more and over-reporters spent significantly fewer hours in an anabolic
state and hours in an energy surplus > 400 kcal than plausible reporters. A
significantly higher proportion of over-reporters and a significantly lower proportion
of under-reporters had values > 0 for maximum surpluses and hours in a surplus >
400 kcal. The opposite was observed for maximum deficit, hours in a deficit >
400kcal, and hours in a catabolic state.
Men had significantly higher maximum surplus, maximum deficit, maximum
energy imbalance, sum max surplus and deficit, and sum all surpluses and deficits in
31
the total sample compared to women. However, the hours spent in energy deficit >
400 kcal, energy surplus > 400kcal, anabolic state, and catabolic state were not
significantly different. No significant differences in the within-day energy balance
variables between men and women were observed in the plausible sample. In the total
sample a significantly higher proportion of men observed values > 0 for maximum
deficit and hours in a deficit > 400 kcal.
Table 5. Summary of Significant Partial Correlations, Controlling for Age and
Sex
Variable 1 Variable 2
Partial
correlation
coefficient (r)
p-value
Total Sample
EF Max Deficit, kcal/d -0.243 0.007
EF Hours in Surplus >400 kcal 0.198 0.028
EF Hours in Deficit >400 kcal -0.305 0.001
EF Hours in Anabolic State 0.341 <0.001
EF Hours in Catabolic State -0.341 <0.001
Body Fat Mass, % Max Deficit, kcal 0.211 0.019
Body Fat Mass, % Hours in Deficit >400 kcal 0.198 0.029
Body Fat Mass, % Hours in Anabolic State -0.178 0.048
Body Fat Mass, % Hours in Catabolic State 0.178 0.048
Plausible Sample
Lean mass to
Height Ratio Hours in Deficit >400kcal 0.456 0.001
EF=Eating Frequency
Table 5 provides a summary of the significant partial correlation coefficients
(r) and corresponding p-values for analyses performed in the total and plausible
samples controlling for sex and age. In the total sample, lower EF was significantly
correlated with higher maximum energy deficit, more hours in energy deficit > 400
kcal and more hours in a catabolic state. Higher EF was significantly correlated with
more hours in an energy surplus > 400 kcal and hours in an anabolic state. Higher
maximum energy deficit and more hours in an energy deficit > 400 kcal were
significantly correlated with higher percentage body fat. Higher percentage body fat
was also significantly correlated to more hours in an energy surplus > 400 kcal and
32
hours in an anabolic state. Note that the correlations for percentage fat free mass
produced the same significant results and values with opposite signs than percentage
body fat. All of the fore mentioned correlations became insignificant then implausible
reporters were excluded. In the plausible sample we observed a significant correlation
between more hours in an energy deficit > 400 kcal and higher lean mass to height
ratio.
Table 6. Summary of Significant Partial Correlations by Sex, Controlling for
Age
Variable 1 Variable 2 Partial correlation
coefficient (r) p-value
Men
Total Sample
EF Hours in Anabolic State 0.326 0.027
EF Hours in Catabolic State -0.326 0.027
Body Fat Mass,% Max Deficit, kcal 0.382 0.009
Body Fat Mass, % Hours in Deficit >400kcal 0.342 0.020
Body Fat Mass, % Hours in Surplus >400kcal -0.292 0.049
Body Fat Mass, % Hours in Anabolic State 0.326 0.027
Body Fat Mass, % Hours in Catabolic State -0.326 0.027
Plausible Sample
Lean Mass to
Height Ratio Hours in Deficit >400kcal 0.649 0.005
Women
Total Sample
EF Max Deficit, kcal -0.326 0.004
EF Hours in Deficit >400kcal -0.342 0.002
EF Hours in Surplus >400kcal 0.236 0.039
EF Hours in Anabolic State 0.354 0.002
EF Hours in Catabolic State -0.354 0.002 EF=Eating Frequency
Additional partial correlation analyses were performed, separating the
population into males and females and controlling for age, in both the total and
plausible samples. A summary of these results is included in Table 6. A greater
number of significant correlations were observed in the male population, similar to
that observed in the total and plausible samples including both men and women. The
only significant correlations observed in the women were in the total sample between
EF and several of the energy balance variables. No significant correlations between
33
markers of adiposity or energy balance variables were observed in the population of
women.
We further investigated the relationships between maximum energy deficit,
EF, and body fat percentage using regression analysis. Figures 7 and 8 show
scatterplots of the maximum energy deficit by percentage body fat and EF for the total
sample.
Figure 7. Percentage Body Fat by Maximum Energy Deficit (kcal/d) in the Total
Sample
Considering the total sample, when controlling for sex and age, we observed a
significant proportion of the variance in percentage body fat was explained by
maximum energy deficit (r2=4.5%, p=0.019), with larger energy deficits associated
with higher percentage body fat. Larger maximum energy deficits were significantly
related to lower EF (r2=5.9%, p=0.003). The relationships observed in the total
sample were no longer significant when implausible reporters were excluded from the
analysis. Regression analyses to test if EF or REI was mediating the relationship
34
between maximum deficits and either percentage body fat or fat free mass did not
observe any mediating relationships.
Figure 8. EF by Maximum Energy Deficit (kcal/d) in the Total Sample
Evaluating the regression models separately for men and women, in the total
sample we observed significant positive relationships between maximum energy
deficit and percentage body fat (r2=14.6%, p=0.009) controlling in the men but not in
the women. The women had a significant negative relationship between maximum
energy deficit and EF in the total sample (r2=10.6%, p=.004). No significant
relationships were observed in the plausible sample. Figure 9 contains a scatterplot of
maximum energy deficit by percentage body fat in men for the total sample. Figure 10
contains a scatterplot of maximum energy deficit by EF in women for the total
sample.
35
Figure 9. Percentage Body Fat by Maximum Energy Deficit (kcal/d) for Men in
the Total Sample
Figure 10. EF by Maximum Energy Deficit (kcal/d) for Women in the Total
Sample
36
CHAPTER V
Discussion and Conclusion
In this study we examined the relationships between EF, adiposity, energy
intake, and within-day fluctuations in energy balance using rigorous methodology to
validate self-reported dietary intake and physical activity level. For the most part, our
results were not congruent with our original hypotheses. In particular, we did not
observe any significant relationships of EF with adiposity even after implausible
reporters were excluded. Although we did observe a positive relationship between EF
and REI in women in the total sample, this relationship was not present when
implausible reporters were excluded. As expected, we observed significant positive
relationships in estimated maximum energy deficit, hours in energy deficit > 400 kcal
and hours in a catabolic state with percentage body fat. However, these relationships
were not significant among women or in the plausible sample. The positive
relationship between lean mass to height ratio and estimated hours in an energy
deficit > 400 kcal, which is the opposite of what we would expect, was the only
significant relationship observed in the plausible sample. The results of our study
suggest that neither EF nor estimated within-day energy balance are important
determinants of energy intake or adiposity in normally active young to middle-aged
adults.
Similar to several25,26,33,34,37,41,42
, but not all21,23,24,27–30,32,38–40
previous cross-
sectional studies we observed no significant relationship between EF and adiposity.
The majority of the studies which did not show a significant relationship between EF
and adiposity were done on female populations34,38,41,42
or had a larger proportion of
women included in the analyses26
like our study, which was 62% women in the total
sample, and 65% in the plausible sample. For the most part, these studies also had
37
similar age ranges. The lack of a relationship between EF and adiposity in our study
is also consistent with results observed in intervention trials investigating the effects
of EF on adiposity5,46–51
. The lack of significance observed in these studies could be
because a relationship doesn’t exist, but could also be a result of how an eating
occasion is defined. An example is two studies by Kant and Graubard (2006)14
and
Duffey and Popkin (2011)15
which both analyzed the same NHANES data but defined
an eating occasion differently. Kant and Graubard (2006) determined EF from the
discrete number of clock times when foods or beverages were consumed on the 24-h
dietary recall, and reported that EF did not significantly change between 1971-1975
and 1999-2002. Duffey and Popkin (2011) combined eating occasions within 15
minutes of each other into one, and they reported that EF significantly increased
between 1971-1975 and 1999-2002. However, both studies reported that higher EF
was associated with higher energy intake. Pearcey and de Castro (2002) analyzed the
same data using 5 definition for eating occasions and found no significant difference
in the results70
. The significant relationship observed between EF and energy intake
in the total sample of women in our study is congruent with results from another
study30
performed on 177 Swedish woman where implausible reporting was not
present and with two studies41,42
on women when implausible reporters were
excluded, although we did not observe a significant relationship in our plausible
sample.
The positive relationships we observed in the maximum energy deficit and
hours in various energy balance states with adiposity in the total sample are consistent
with previous studies6,54,56
. However, the lack of significant relationships in the
plausible sample is inconsistent with previous results. One explanation for this
inconsistency is the differences between the populations studied. Previous studies
38
have been performed on athletes6,54
(mean ages: 15.5-26.6 years) and adolecents56
(age range: 8-14 years), whereas our study was on an older population with a wider
age range (mean age: 29.8 years, range: 18-64 years) and more variability in
characteristics among subjects. Additionally, previous studies examining within-day
energy fluctuations have relied on self-reporting of dietary intake and physical
activity and not made considerations for implausible energy intake reporting. Also,
we did not have the exact time of physical activity performed and were only able to
make estimates of hourly energy balance based on PALs calculated for time intervals
spanning several hours. Alternatively, the lack of significance in the plausible sample
may be due to the exclusion of individuals who are reporting their intake but may be
in a state of adaptive thermogenesis71
. Adaptive thermogenesis refers to the changes
in resting and non-resting energy expenditure associated with over and under
consumption of energy. The various mechanisms associated with adaptive
thermogenesis may play a role in determining body composition, and identifying and
including these individuals in the analysis could produce significant results. However,
we excluded individuals who had changes in body weight > 5 lbs. over the previous
six months to reduce the likelihood of adaptive thermogenesis occurring.
The significant relationships observed between variables in the total sample
but not in the plausible sample illustrate the impact of implausible energy intake
reporting on observed relationships between dietary variables and adiposity. Under-
reporting of dietary intake can affect measures of EF, for example if an individual
completely omits one or more eating occasion, then EF would be underestimated.
However, we did not observe a significant difference in EF between plausible, over,
and under-reporters in our study. When one considers the difference in the within-day
energy balance variables among over, under, and plausible reporters these trends
39
make sense when one thinks about how over or under-reporting in a single eating
occasion would affect the within-day energy balance observed For example, under-
reporting in an eating occasion means a lower reported energy intake for that period
of time, and thus a less positive or more negative observed energy balance for the
hour interval and the remainder of the 24 hour period compared to the actual energy
balance. Over-reporting means a higher reported energy intake and thus a more
positive or less negative observed energy balance for the hour interval and remainder
of the 24 hour period compared to the actual energy balance. These differences could
also be explained by under or over-eating. Previous studies have shown that
underreporting of food intake can be explained by undereating72,73
. Under-reporting
versus under-eating can be determined by comparing intake to decreases in body
mass, which indicates consumption less than energy requirements, such as the method
used by Goris et al. (2001)73
. It follows that similar methods could be used to identify
over-reporters from over-eaters. New methods are currently being developed that will
objectively measure EF and dietary intake in order to eliminate the issue of inaccurate
reporting associated with self-reports of dietary intake. The automatic ingestion
monitor (AIM), a novel wearable sensor system developed by Fontana et al. (2014),
has shown promising results being able to detect food intake with an average
accuracy of 89.8%74
. Another system, Image-Diet Day75
, uses a camera-equipped
mobile phone to automatically captures images of food intake in order to aid
participants with computer-assisted multiple-pass 24 hour dietary recalls. The images
were reported to be helpful in recalling dietary intake by 93% of the participants75
.
Our study had several strengths. First, we had a relatively large sample size.
We also used multiple pass 24 hour dietary recalls to assess dietary intake, which
results in energy intakes closer to TEE than other methods, such as food diaries or
40
food frequency questionnaires62
. Another strength of this study is the consideration of
implausible reporting and the use of DLW to measure TEE. However, there were
several limitations to our study. We only used one day of dietary and physical activity
recall which may not be an accurate representation of the normal habits of
participants. The PALs calculated to determine energy expenditure for calculation of
hourly energy balance were based on one day of physical activity recall whereas
DLW was performed over a ten day period. The variance in the one day of recall
could be another possible explanation for the significant difference between
measurements of TEE. Additionally, we did not have the exact time of physical
activity performed, which affected the estimates of hourly energy balance so that we
have no way of discerning if the significant correlations observed between adiposity
and the number of hours in various energy balance states are valid. Therefore, we
may not have an accurate measure of hourly energy balance and this could be a
possible explanation for the positive relationship observed between lean mass to
height ratio and hours in an energy deficit > 400 kcal in the plausible sample. Another
limitation of our study is the significant difference in pTEE and TEEDLW observed in
the total sample. However, the fact that this difference was not observed in the
plausible sample could be explained by the under or over-reporting of physical
activity. It is possible that if someone is inaccurately self-reporting their dietary
intake, then they may also inaccurately report their physical activity. In our sample,
under-reporters had significantly lower pTEE than TEEDLW and over-reporters had
significantly higher pTEE than TEEDLW. However, the average difference between
TEE calculated by the seven day physical activity recall and TEE measured by DLW
was calculated to be only 5 kcal in each one hour period of the day, a difference
41
which would not likely have had a critical impact on the within-day energy balance
calculations.
In conclusion, this study is in agreement with other cross-sectional37
as well as
intervention5,51
studies in community-dwelling individuals consuming self-selected
diets showing no relationship of EF on adiposity. Additionally, within-day
fluctuations in energy balance had little impact on the results, but this could be due to
the difference in sampled population from previous studies6,54,56
, the estimation of
hourly energy balance measurement, or the inability to identify individuals who were
in adaptive thermogenesis in our study. If this study were repeated, we would suggest
including more days of dietary and physical activity recall as well as assessing the
energy balance in larger time intervals (for sleep, morning, afternoon, and evening)
corresponding to the intervals used in the PAR. In addition, we would suggest
recording the exact times of physical activity in order to have a better estimation of
hourly energy balance. The use of an objective monitoring device to record EF would
reduce the implausible reporting. Moving forward more studies accounting for
implausible dietary intake reports need to be conducted to assess the true relationship
between within-day fluctuation of energy balance and adiposity in free-living adults
consuming self-selected diets. Additionally, a standard definition for defining an
eating occasion needs to be developed. In order to truly understand the relationships
among EF, energy intake, and adiposity, more intervention trials, where EF is the
main intervention, need to be performed.
42
REFERENCES
1. Huang TT-K, Roberts SB, Howarth NC, McCrory MA. Effect of screening out
implausible energy intake reports on relationships between diet and BMI. Obes
Res. 2005;13(7):1205-1217. doi:10.1038/oby.2005.143.
2. McCrory MA, Howarth NC, Roberts SB, Huang TT-K. Eating Frequency and
Energy Regulation in Free-Living Adults Consuming Self-Selected Diets. J
Nutr. 2011;141(1):148-153. doi:10.3945/jn.109.114991.
3. Leidy HJ, Campbell WW. The Effect of Eating Frequency on Appetite Control
and Food Intake: Brief Synopsis of Controlled Feeding Studies. J Nutr.
2011;141(1):154-157. doi:10.3945/jn.109.114389.
4. Huseinovic E, Winkvist A, Bertz F, Bertéus Forslund H, Brekke HK. Eating
frequency, energy intake and body weight during a successful weight loss trial in
overweight and obese postpartum women. Eur J Clin Nutr. 2014;68(1):71-76.
doi:10.1038/ejcn.2013.200.
5. Bertéus Forslund H, Klingström S, Hagberg H, Löndahl M, Torgerson JS,
Lindroos AK. Should snacks be recommended in obesity treatment? a 1-year
randomized clinical trial. Eur J Clin Nutr. 2007;62(11):1308-1317.
doi:10.1038/sj.ejcn.1602860.
6. Benardot D. Energy Thermodynamics Revisited: Energy Intake Strategies for
Optimizing Athlete Body Composition and Performance. J Exerc Sci Health.
2013;11(2):1-13.
7. Saad MF, Bernaba B, Hwu C-M, et al. Insulin Regulates Plasma Ghrelin
Concentration. J Clin Endocrinol Metab. 2002;87(8):3997-4000.
doi:10.1210/jcem.87.8.8879.
8. Knerr I, Groschl M, Rascher W, Rauh M. Endocrine Effects of Food Intake:
Insulin, Ghrelin, and Leptin Responses to a Single Bolus of Essential Amino
Acids in Humans. Ann Nutr Metab. 2003;47(6):312-318.
doi:10.1159/000072405.
9. Blom WA, Stafleu A, Graaf C de, Kok FJ, Schaafsma G, Hendriks HF. Ghrelin
response to carbohydrate-enriched breakfast is related to insulin. Am J Clin Nutr.
2005;81(2):367-375.
10. Iwao S, Mori K, Sato Y. Effects of meal frequency on body composition during
weight control in boxers. Scand J Med Sci Sports. 1996;6(5):265-272.
11. Nieman DC, Henson DA, Smith LL, et al. Cytokine changes after a marathon
race. J Appl Physiol. 2001;91(1):109-114.
12. Mayer J. Regulation of energy intake and the body weight: the glucostatic theory
and the lipostatic hypothesis. Ann N Y Acad Sci. 1955;63(1):15-43.
13. Chaput J-P, Tremblay A. The glucostatic theory of appetite control and the risk
of obesity and diabetes. Int J Obes. 2008;33(1):46-53. doi:10.1038/ijo.2008.221.
43
14. Kant AK, Graubard BI. Secular trends in patterns of self-reported food
consumption of adult Americans: NHANES 1971-1975 to NHANES 1999–
2002. Am J Clin Nutr. 2006;84(5):1215-1223.
15. Duffey KJ, Popkin BM. Energy Density, Portion Size, and Eating Occasions:
Contributions to Increased Energy Intake in the United States, 1977–2006. PLoS
Med. 2011;8(6). doi:10.1371/journal.pmed.1001050.
16. Kant AK, Graubard BI. Secular trends in the association of socio-economic
position with self-reported dietary attributes and biomarkers in the US
population: National Health and Nutrition Examination Survey (NHANES)
1971–1975 to NHANES 1999–2002. Public Health Nutr. 2007;10(02):158-167.
doi:10.1017/S1368980007246749.
17. Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of childhood and adult
obesity in the United States, 2011-2012. JAMA. 2014;311(8):806-814.
doi:10.1001/jama.2014.732.
18. Schoeller DA. Limitations in the assessment of dietary energy intake by self-
report. Metabolism. 1995;44(2 Suppl 2):18-22.
19. Livingstone MBE, Black AE. Markers of the validity of reported energy intake.
J Nutr. 2003;133(3):895S - 920S.
20. Howarth NC, Huang TT-K, Roberts SB, Lin B-H, McCrory MA. Eating patterns
and dietary composition in relation to BMI in younger and older adults. Int J
Obes. 2006;31(4):675-684. doi:10.1038/sj.ijo.0803456.
21. Murakami K, Livingstone MBE. Eating frequency in relation to body mass
index and waist circumference in British adults. Int J Obes 2005.
2014;38(9):1200-1206. doi:10.1038/ijo.2014.1.
22. Mattson MP. The need for controlled studies of the effects of meal frequency on
health. The Lancet. 2005;365(9475):1978-1980. doi:10.1016/S0140-
6736(05)66667-6.
23. Fabry P, Hejl Z, Fodor J, Braun T, Zvolankova K. The Frequency Of Meals. Its
Relation To Overweight, Hypercholesterolaemia, And Decreased Glucose-
Tolerance. Lancet. 1964;2(7360):614-615.
24. Metzner HL, Lamphiear DE, Wheeler NC, Larkin FA. The relationship between
frequency of eating and adiposity in adult men and women in the Tecumseh
Community Health Study. Am J Clin Nutr. 1977;30(5):712-715.
25. Dreon DM, Frey-Hewitt B, Ellsworth N, Williams PT, Terry RB, Wood PD.
Dietary fat:carbohydrate ratio and obesity in middle-aged men. Am J Clin Nutr.
1988;47(6):995-1000.
26. Edelstein SL, Barrett-Connor EL, Wingard DL, Cohn BA. Increased meal
frequency associated with decreased cholesterol concentrations; Rancho
Bernardo, CA, 1984-1987. Am J Clin Nutr. 1992;55(3):664-669.
44
27. Kant AK, Schatzkin A, Graubard BI, Ballard-Barbash R. Frequency of eating
occasions and weight change in the NHANES I Epidemiologic Follow-up Study.
Int J Obes Relat Metab Disord J Int Assoc Study Obes. 1995;19(7):468-474.
28. Wahlqvist Ml, Kouris-Blazos A, Wattanapenpaiboon N. The Significance of
Eating Patterns: An Elderly Greek Case Study. Appetite. 1999;32(1):23-32.
doi:10.1006/appe.1998.0192.
29. Titan SMO, Bingham S, Welch A, et al. Frequency of eating and concentrations
of serum cholesterol in the Norfolk population of the European prospective
investigation into cancer (EPIC-Norfolk): cross sectional study. BMJ.
2001;323(7324):1286. doi:10.1136/bmj.323.7324.1286.
30. Bertéus Forslund H, Lindroos AK, Sjöström L, Lissner L. Meal patterns and
obesity in Swedish women-a simple instrument describing usual meal types,
frequency and temporal distribution. Eur J Clin Nutr. 2002;56(8):740-747.
doi:10.1038/sj.ejcn.1601387.
31. Ma Y, Bertone ER, Stanek EJ, et al. Association between Eating Patterns and
Obesity in a Free-living US Adult Population. Am J Epidemiol. 2003;158(1):85-
92. doi:10.1093/aje/kwg117.
32. Bertéus Forslund H, Torgerson JS, Sjöström L, Lindroos AK. Snacking
frequency in relation to energy intake and food choices in obese men and women
compared to a reference population. Int J Obes. 2005;29(6):711-719.
doi:10.1038/sj.ijo.0802950.
33. Berg C, Lappas G, Wolk A, et al. Eating patterns and portion size associated
with obesity in a Swedish population. Appetite. 2009;52(1):21-26.
doi:10.1016/j.appet.2008.07.008.
34. Amosa T, Rush E, Plank L. Frequency of eating occasions reported by young
New Zealand Polynesian and European women. Pac Health Dialog.
2001;8(1):59-65.
35. Ortega RM, Requejo AM, López-Sobaler AM, et al. Difference in the breakfast
habits of overweight/obese and normal weight schoolchildren. Int J Vitam Nutr
Res Int Z Für Vitam- Ernährungsforschung J Int Vitaminol Nutr.
1998;68(2):125-132.
36. Goldberg GR, Black AE, Jebb SA, et al. Critical evaluation of energy intake data
using fundamental principles of energy physiology: 1. Derivation of cut-off
limits to identify under-recording. Eur J Clin Nutr. 1991;45(12):569-581.
37. Summerbell CD, Moody RC, Shanks J, Stock MJ, Geissler C. Relationship
between feeding pattern and body mass index in 220 free-living people in four
age groups. Eur J Clin Nutr. 1996;50(8):513-519.
38. Drummond SE, Crombie NE, Cursiter MC, Kirk TR. Evidence that eating
frequency is inversely related to body weight status in male, but not female, non-
obese adults reporting valid dietary intakes. Int J Obes. 1998;22(2):105-112.
doi:10.1038/sj.ijo.0800552.
45
39. Ruidavets JB, Bongard V, Bataille V, Gourdy P, Ferrières J. Eating frequency
and body fatness in middle-aged men. Int J Obes Relat Metab Disord J Int Assoc
Study Obes. 2002;26(11):1476-1483. doi:10.1038/sj.ijo.0802143.
40. Howarth NC, Huang TT-K, Roberts SB, Lin B-H, McCrory MA. Eating patterns
and dietary composition in relation to BMI in younger and older adults. Int J
Obes 2005. 2007;31(4):675-684. doi:10.1038/sj.ijo.0803456.
41. Hartline-Grafton HL, Rose D, Johnson CC, Rice JC, Webber LS. The influence
of weekday eating patterns on energy intake and BMI among female elementary
school personnel. Obes Silver Spring Md. 2010;18(4):736-742.
doi:10.1038/oby.2009.249.
42. Mills JP, Perry CD, Reicks M. Eating frequency is associated with energy intake
but not obesity in midlife women. Obes Silver Spring Md. 2011;19(3):552-559.
doi:10.1038/oby.2010.265.
43. Reicks M, Degeneffe D, Rendahl A, et al. Associations between eating occasion
characteristics and age, gender, presence of children and BMI among U.S.
adults. J Am Coll Nutr. 2014;33(4):315-327.
doi:10.1080/07315724.2014.887485.
44. Karatzi K, Yannakoulia M, Psaltopoulou T, et al. Meal patterns in healthy
adults: Inverse association of eating frequency with subclinical atherosclerosis
indexes. Clin Nutr Edinb Scotl. 2015;34(2):302-308.
doi:10.1016/j.clnu.2014.04.022.
45. Aljuraiban GS, Chan Q, Oude Griep LM, et al. The Impact of Eating Frequency
and Time of Intake on Nutrient Quality and Body Mass Index: The INTERMAP
Study, a Population-Based Study. J Acad Nutr Diet. 2015;115(4):528-536.e1.
doi:10.1016/j.jand.2014.11.017.
46. Arnold LM, Ball MJ, Duncan AW, Mann J. Effect of isoenergetic intake of three
or nine meals on plasma lipoproteins and glucose metabolism. Am J Clin Nutr.
1993;57(3):446-451.
47. Arnold L, Ball M, Mann J. Metabolic effects of alterations in meal frequency in
hypercholesterolaemic individuals. Atherosclerosis. 1994;108(2):167-174.
48. McGrath SA, Gibney MJ. The effects of altered frequency of eating on plasma
lipids in free-living healthy males on normal self-selected diets. Eur J Clin Nutr.
1994;48(6):402-407.
49. Arnold L, Mann JI, Ball MJ. Metabolic effects of alterations in meal frequency
in type 2 diabetes. Diabetes Care. 1997;20(11):1651-1654.
50. Thomsen C, Christiansen C, Rasmussen OW, Hermansen K. Comparison of the
effects of two weeks’ intervention with different meal frequencies on glucose
metabolism, insulin sensitivity and lipid levels in non-insulin-dependent diabetic
patients. Ann Nutr Metab. 1997;41(3):173-180.
46
51. Bachman JL, Raynor HA. Effects of Manipulating Eating Frequency During a
Behavioral Weight Loss Intervention: A Pilot Randomized Controlled Trial.
Obesity. 2012;20(5):985-992. doi:10.1038/oby.2011.360.
52. Alencar MK, Beam JR, McCormick JJ, et al. Increased meal frequency
attenuates fat-free mass losses and some markers of health status with a portion-
controlled weight loss diet. Nutr Res N Y N. 2015.
doi:10.1016/j.nutres.2015.03.003.
53. Feller S, Müller A, Mayr A, Engeli S, Hilbert A, de Zwaan M. What
distinguishes weight loss maintainers of the German Weight Control Registry
from the general population? Obes Silver Spring Md. 2015.
doi:10.1002/oby.21054.
54. Deutz RC, Benardot D, Martin DE, Cody MM. Relationship between energy
deficits and body composition in elite female gymnasts and runners. Med Sci
Sports Exerc. 2000;32(3):659-668.
55. Mountjoy M, Sundgot-Borgen J, Burke L, et al. Authors’ 2015 additions to the
IOC consensus statement: Relative Energy Deficiency in Sport (RED-S). Br J
Sports Med. 2015;49(7):417-420. doi:10.1136/bjsports-2014-094371.
56. Delfausse L. Within-Day Energy Balance in Children and Adolescents. Unpubl
Masters Thesis Ga State Univ. 2012.
57. Delk A. The Relationships Between Real Time Energy Balance, Hunger, and
Body Composition. Atlanta, Ga. : Georgia State University, 2014.; 2014.
58. Leet R. Within-Day Energy Balance, Body Mass Index, and Body Composition
in College Students. Nutr Theses. 2014.
http://scholarworks.gsu.edu/nutrition_theses/59.
59. McCrory MA, Gomez TD, Bernauer EM, Molé PA. Evaluation of a new air
displacement plethysmograph for measuring human body composition. Med Sci
Sports Exerc. 1995;27(12):1686-1691.
60. Dempster P, Aitkens S. A new air displacement method for the determination of
human body composition. Med Sci Sports Exerc. 1995;27(12):1692-1697.
61. Miller DK, Morley JE, Rubenstein LZ, Pietruszka FM, Strome LS. Formal
geriatric assessment instruments and the care of older general medical
outpatients. J Am Geriatr Soc. 1990;38(6):645-651.
62. Posner BM, Borman CL, Morgan JL, Borden WS, Ohls JC. The validity of a
telephone-administered 24-hour dietary recall methodology. Am J Clin Nutr.
1982;36(3):546-553.
63. Sallis JF, Haskell WL, Wood PD, et al. Physical activity assessment
methodology in the Five-City Project. Am J Epidemiol. 1985;121(1):91-106.
47
64. Washburn RA, Jacobsen DJ, Sonko BJ, Hill JO, Donnelly JE. The validity of the
Stanford Seven-Day Physical Activity Recall in young adults. Med Sci Sports
Exerc. 2003;35(8):1374-1380. doi:10.1249/01.MSS.0000079081.08476.EA.
65. Calazel CM, Young VR, Evans WJ, Roberts SB. Effect of fasting and feeding on
measurement of carbon dioxide production using doubly labeled water. J Appl
Physiol Bethesda Md 1985. 1993;74(4):1824-1829.
66. Butte NF, Wong WW, Adolph AL, Puyau MR, Vohra FA, Zakeri IF. Validation
of Cross-Sectional Time Series and Multivariate Adaptive Regression Splines
Models for the Prediction of Energy Expenditure in Children and Adolescents
Using Doubly Labeled Water. J Nutr. 2010;140(8):1516-1523.
doi:10.3945/jn.109.120162.
67. Surrao J, Sawaya AL, Dallal GE, Tsay R, Roberts SB. Use of food quotients in
human doubly labeled water studies: comparable results obtained with 4 widely
used food intake methods. J Am Diet Assoc. 1998;98(9):1015-1020.
doi:10.1016/S0002-8223(98)00232-6.
68. Huang TT-K, Howarth NC, Lin B-H, Roberts SB, McCrory MA. Energy intake
and meal portions: associations with BMI percentile in U.S. children. Obes Res.
2004;12(11):1875-1885. doi:10.1038/oby.2004.233.
69. Black AE. The sensitivity and specificity of the Goldberg cut-off for EI:BMR
for identifying diet reports of poor validity. Eur J Clin Nutr. 2000;54(5):395-
404.
70. Pearcey SM, Castro JM de. Food intake and meal patterns of weight-stable and
weight-gaining persons. Am J Clin Nutr. 2002;76(1):107-112.
71. Dulloo AG, Seydoux J, Jacquet J. Adaptive thermogenesis and uncoupling
proteins: a reappraisal of their roles in fat metabolism and energy balance.
Physiol Behav. 2004;83(4):587-602. doi:10.1016/j.physbeh.2004.07.028.
72. Goris AHC, Westerterp KR. Underreporting of Habitual Food Intake Is
Explained by Undereating in Highly Motivated Lean Women. J Nutr.
1999;129(4):878-882.
73. Goris AHC, Meijer EP, Westerterp KR. Repeated measurement of habitual food
intake increases under-reporting and induces selective under-reporting. Br J
Nutr. 2001;85(05):629-634. doi:10.1079/BJN2001322.
74. Fontana JM, Farooq M, Sazonov E. Automatic ingestion monitor: a novel
wearable device for monitoring of ingestive behavior. IEEE Trans Biomed Eng.
2014;61(6):1772-1779. doi:10.1109/TBME.2014.2306773.
75. Arab L, Estrin D, Kim DH, Burke J, Goldman J. Feasibility testing of an
automated image-capture method to aid dietary recall. Eur J Clin Nutr.
2011;65(10):1156-1162. doi:10.1038/ejcn.2011.75.
48
APPENDICES
APPENDIX A
Example of Multiple Pass 24 hour Dietary Recall Form
Study ID#_BM1 Multiple Pass 24-hour Dietary Recall Form For Staff Use Only
Date____8/20/2010_____
Interviewer initials___ OHM___
verified_____
Time to bed night before Time to bed on 24 hour recall day
1:a 1:45a
Q2 Start/
Stop time
11:45a
12:15p
4:50p
5:30p
7p
10p
*B = Breakfast, L = Lunch, D = Dinner/Supper, S = Snack, BV = Beverage (not part of a meal) BR=Brunch
†H = Home, F = Friend’s food, FF = Fast food restaurant (name specific restaurant), R = other restaurant, O = Other (please specify)
‡ e.g. Watching TV, eating at table/nothing else except eating, at party, studying, etc
be sure to ask what they were doing physically (seated, standing etc) and what they were doing with their attention (talking, nothing, TV etc)
Eating Occasion Where Obtained Eaten at Home?
B H Y
L F N
D FF
S R
BV O
BR
Q8
a E
ate
n a
t h
om
e?
Q8
b W
ha
t w
ere y
ou
do
ing
wh
ile e
ati
ng
?Q1 Quicklist
Q3
Ea
tin
g O
cca
s. Food/drink
additions
Q4 Descriptions and amounts
Q5
Ho
w M
uch
ea
ten
?
Q7
Wh
ere
Ob
tain
ed
?
o n2 pieces of watermelon l seeded, B4 4 thick
Day Recalled: Su M T W Th F Sa
hot dog d Bun (regular, B4, 14 Thickness), grilled hotdog kosher (Thickness 6,
B4), lettuce (1/8c), tomato(2 slices- B1, thickness 2), mushrooms
(cooked, 1/8c), ketchup (1tbl), mustard regular yellow(2 tsps) (FIVE
S Wrigley, spearmint, 1/8 tsp (10 mints = 100kcal)
checked______
entered______
0.5 liter (tap) all o n
Time up on 24 hour recall day
10:30a
Q8
c #
Oth
er P
eo
ple
Prese
nt?
one chicken wrap l Wrap (spinach B4), Lettuce 1/8c (iceburg), tomato 1/8c diced, cucumber
1/8c diced, cheese (moneray jack regular) 1tbl, bacon (pork) 1 tbl
crumble, 6 chicken strips grilled (4thick length B1)
0.80 O N 1talking
sitting,
looking
out the
window
(people
watching
)
1
smal ceasar salad l 1.25 c romaine and iceburg lettuce, 5 croutons (size of a quarter, 3
thickness) , ceasar dressing (creamy, 1 tbl)
all o n 1
all
1
water l 1
0.75
1
coke (can) d 12oz can regular, ice (1/3c) all ff n
talking,
watching
the
o n2% milk l 1 cup
all ff n
0
bottle of water BV 32oz all h n 0
10 mints all h n standing,
walking
49
APPENDIX B
Seven-Day Physical Activity Recall (PAR) Questionnaire
Study ID ______________ Staff use only: Checked _____ Entered _____ Verified _____
Interviewer
Today's Date Day (circle one): Mon Tues Wed Thurs Fri Sat Sun
1. Were you employed in the last seven days? 0. No (Skip to Q#3) 1. Yes
2. If yes, which days? Mon Tues Wed Thurs Fri Sat Sun
3. What two days do you consider your weekend days? Mon Tues Wed Thurs Fri Sat Sun
DAYS
1 2 3 4 5 6 7
Day: (Mon…)
(yesterday)
Date:
(mm/dd/yy)
Sleep:
In Bed / Up
Total time
Work:
Start / End
Work Breaks
Total Time Worked
M O R N I N G
Moderate
Hard
Very Hard
A F T E R N O O N
Moderate
Hard
Very Hard
E V E N I N G
Moderate
Hard
Very Hard
50
PAR Questionnaire – Page 2 4. Compared to your physical activity over the past three months, was last week's physical activity more, less or
about the same? 1. More 2. Less 3. About the same
KEY Rounding: 10-22 min =0.25 hr Place asterisk (*) to the right of a work-related activity 23-37 min =0.50 hr & the time spent doing it. 38-52 min =0.75 hr 53-1:07 hr:min =1.0 hr 1:08-1:22 hr:min =1.25 hr
INTERVIEWER: Please answer questions below and note any comments on interview. 5. Were there any problems with the 7-Day PAR interview? 0. No 1. Yes (If yes, please explain.)
6. Do you think this was a valid 7-Day PAR interview? 0. No (Please explain)
1. Yes
7. Please list below any activities reported by the subject that you don't know how to classify.
8. Please provide any other comments you may have in the space below.
51
SEVEN-DAY PHYSICAL ACTIVITY RECALL (PAR) LIST OF ACTIVITIES
Intensity Job Home Sport or Recreation **************************************************************************************************************** * Typing * Ironing, sewing * Leisurely walking LIGHT * Standing * Light auto repair * Softball * Driving * Indoor Painting * Bowling * Playing a musical instrument **************************************************************************************************************** * Lifting or carrying * Sweeping, Mopping, * Brisk walking (on level ground) light objects Vacuuming * Shooting baskets (up to 5 lbs) * Clipping hedges * Throwing frisbee MODERATE * Raking * Cycling leisurely * Mowing lawn with on level ground * Painting out- power mower * Swimming laps (easy effort) side of house * Cleaning windows * Weightlifting * Pushing stroller with child **************************************************************************************************************** * Construction work * Scrubbing floors * Brisk walking (uphill) * Lifting or * Shoveling dirt, coal, etc. * Backpacking (on level ground) carrying objects * Mowing lawn with * Brisk cycling on level ground HARD (5-15 lbs) non-power mower without losing breath * Climbing * Carrying child * Tennis (doubles) ladder or stairs (5-15 lbs) * Downhill skiing * Swimming laps (moderate effort) **************************************************************************************************************** * Carrying heavy * Digging ditches * Jogging loads such as * Chopping or * Basketball (in game) bricks or lumber splitting wood * Soccer (in game) * Gardening with * Backpacking (uphill) VERY * Carrying heavy tools * Cycling (uphill or racing) HARD moderate loads * Tennis (singles) up stairs * Cross-country skiing (16-40 lbs) * Swimming laps (hard effort) * Aerobic dancing * Circuit training (using a series of Nautilus machines without stopping)
52
APPENDIX C
NutritimingTM
Input Form
En
ter
Va
lue
s O
nly
in
Lig
ht
Blu
e a
nd
Lig
ht
Gre
y A
rea
sF
01
F0
2F
O3
F0
4F
05
F0
6F
07
F0
8F
09
Kn
ow
nK
ca
l In
En
erg
y
RE
E D
ata
En
try
US
Da
ily
Hrs
RE
E/
HrA
cti
v. K
ca
l O
ut
50
10
01
50
20
02
50
30
05
00
10
00
20
00
Ca
lorie
sB
ala
nce
M=
1;
F=
2:
6-7am
0.0
01
0.0
0350
Ag
e(y
r):
7-8am
0.0
01
0.0
0350
We
igh
t(kg
):
0.0
8-9am
0.0
01
0.0
0350
He
igh
t(cm
):
0.0
9-10am
0.0
01
0.0
0350
RE
E:
10-11am
0.0
01
0.0
0350
Ho
url
y R
EE
:11-12pm
0.0
01
0.0
0350
% C
ha
ng
e:
1.0
012-1pm
0.0
01
0.0
0350
BM
I:#
DIV
/0!
1-2pm
0.0
01
0.0
0350
No
te:
RE
E u
se
s M
iffl
in S
t. J
eo
r2-3pm
0.0
01
0.0
0350
La
st
Ea
t P
rev
. D
ay
:3-4pm
0.0
01
0.0
0350
Est.
Liv
er
Ca
lori
es:
350
4-5pm
0.0
01
0.0
0350
5-6pm
0.0
01
0.0
0350
Acti
vit
yF
acto
r G
uid
e6-7pm
0.0
01
0.0
0350
Resting=1.0
7-8pm
0.0
01
0.0
0350
Typic
al Day=1.5
8-9pm
0.0
01
0.0
0350
Very
Lig
ht=
2.0
9-10pm
0.0
01
0.0
0350
Lig
ht=
3.0
10-11pm
0.0
01
0.0
0350
Modera
te=4.0
11-12am
0.0
01
0.0
0350
Vig
oro
us=5.0
12-1am
0.0
01
0.0
0350
Heavy=6.0
1-2am
0.0
01
0.0
0350
Exhaustive=7.0
2-3am
0.0
01
0.0
0350
3-4am
0.0
01
0.0
0350
Nu
triT
imin
g Q
uic
k-C
he
ck®4-5am
0.0
01
0.0
0350
© 2
01
4,
Da
n B
en
ard
ot
5-6am
0.0
01
0.0
0350
DA
Y'S
EN
DKC
AL O
UT=
0IN
(TO
TA
L K
CA
L &
EB
)=
03
50
.0
F0
1=
Very
Sm
all S
nack (
up
to 5
0 C
alo
ries)
FO
2=
Sm
all S
nack (
~1
00
Calo
ries)
F0
3=
Med
ium
Sn
ack (
~1
50
Calo
ries)
F0
4=
Reg
ula
r S
nack (
~2
00
Calo
ries)
F0
5=
Larg
e S
nack (
~2
50
Calo
ries)
F0
6=
Sm
all M
eal (~
30
0 C
alo
ries)
F0
7=
Med
ium
Meal (~
50
0 C
alo
ries)
F0
8=
Larg
e M
eal (~
10
00
Calo
ries)
F0
9=
Extr
a L
arg
e M
eal (~
20
00
Calo
ries)
0
50
10
0
15
0
20
0
25
0
30
0
35
0
40
0
12
34
56
78
91
01
11
21
31
41
51
61
71
81
92
02
12
22
32
4
Energy Balance (Calories)
Ho
ur
of
the
Day
Wit
hin
-Day
En
erg
y B
alan
ce