(Adam Drenowski) Obesity Treatment and Prevention New Directions
(Adam Drenowski) Obesity Treatment and Prevention New Directions
(Adam Drenowski) Obesity Treatment and Prevention New Directions
Adult
Obesity Treatment
and Prevention:
New Directions
Adam Drewnowski
Barbara J. Rolls
Obesity Treatment and Prevention: New Directions
Nestlé Nutrition Institute
Workshop Series
Vol. 73
Obesity Treatment
and Prevention:
New Directions
Editors
616.3'98--dc23
2012034925
The material contained in this volume was submitted as previously unpublished material, except in the instances in
which credit has been given to the source from which some of the illustrative material was derived.
Great care has been taken to maintain the accuracy of the information contained in the volume. However, neither
Nestec Ltd. nor S. Karger AG can be held responsible for errors or for any consequences arising from the use of the
information contained herein.
© 2012 Nestec Ltd., Vevey (Switzerland) and S. Karger AG, Basel (Switzerland). All rights reserved. This book is
protected by copyright. No part of it may be reproduced, stored in a retrieval system, or transmitted, in any form or
by any means, electronic, mechanical, photocopying, or recording, or otherwise, without the written permission of
the publisher.
VII Preface
XI Foreword
XV Contributors
V
123 The Importance of Systems Thinking to Address Obesity
Finegood, D.T. (Canada)
139 Summary Discussion on New Directions for Prevention
For more information on related publications, please consult the NNI website:
www.nestlenutrition–institute.org
VI Contents
Preface
Obesity continues to be a major problem for global public health. Despite best
efforts by health care providers, public health agencies, and the private sector,
few prevention efforts have been effective. As a result, bodyweights have con-
tinued their upward surge, contributing to ill-health among both children and
adults. Although many treatment programs are available, few can boast of long-
term success, with minimal risk of relapse and weight regain.
One glimmer of hope has been the recent leveling off in obesity rates in the
US and in European countries. Whether temporary or permanent, it provides
a much-needed breathing space. This is the opportunity to regroup, reexamine
past approaches, assess relative success rates, and point to new directions for the
future. Those new directions can be global in their reach. Obesity and diabetes
are no longer limited to wealthy societies. Their rates in developing countries,
while still relatively low, are on the rise.
Clearly, obesity treatment and prevention can benefit from a major paradigm
shift. The standard advice to eat less and exercise more is both simplistic and
often ineffective. The science presented at this workshop suggested a sequence
of steps that could result in new ways to address obesity at the personal as well
as at the population level. First, we need to achieve a clear understanding of
who becomes obese, where, and for what reason. Second, we need to parse out
the likely contributions to overeating by the brain, biology, economics, and the
environment. Then, based on our understanding of disease etiology and its dis-
tribution by geography and by social strata, we need to develop targeted yet
comprehensive strategies for obesity prevention and treatment for both indi-
viduals and groups.
The conventional disease model of obesity, built around the individual
patient, needs to be reconciled with some of the contemporary thinking about
obesity in its social and environmental context. That may require a shift of
emphasis from the individual to the group and a full consideration of the bio-
logical as well as the social and economic determinants of health.
One issue that deserves serious study is the observed social gradient in obe-
sity. Obesity rates, especially for women, are strongly influenced by the social
VII
and economic environments. In developing countries, it is the more affluent
urban women who are gaining weight. In developed countries, higher rates of
obesity are observed in relatively more deprived areas. In both cases, women’s
bodyweights are affected by the changing social conditions more than body-
weights of men. The global obesity epidemic can thus be viewed through the
prism of social disadvantage and women’s health.
Clear thinking about the biological and social determinants of weight and
health is essential to the success of any intervention, whether at the individual
or population level. Few studies have examined the success of prevention or
treatment strategies across different socioeconomic strata. Little is known about
social and economic barriers to the adoption of lower energy density diets and
more active lifestyles. If the reasons why some women gain weight and others
do not are either misunderstood or ignored altogether, then the proposed pre-
vention plans will fail. If the reasons why some people stay with a diet regime
whereas others fail are unclear, then the treatment plans will be ineffective in the
long-term. Barriers to treatment success may be inherent in the brain, the biol-
ogy, or the environment.
Current advances in research – discussed here – have the power to change our
thinking. The traditional disease model holds that people become obese through
a combination of genetic predisposition, faulty metabolism, and acquired bad
habits. The new emphasis must move beyond a focus on personal responsibil-
ity, calorie counting, and the macronutrient composition of the diet. We need a
greater understanding of behaviors determining food intake and physical activ-
ity and how environmental and biological influences combine to determine food
choices, amounts consumed, and activity levels. Appeals to individual motiva-
tion need to be tempered by the recognition that food-seeking behaviors are
bound by biological, economic, social and environmental constraints.
Our thinking about the role of the adipose tissue has also undergone a para-
digm shift. Formerly viewed as a passive fat repository, the adipose tissue is now
regarded as an active neuroendocrine organ functioning in concert with body
physiology. Similarly, the obese person is often thought of as the victim of pas-
sive overeating. Our thinking about obesity could benefit from a broader inte-
grative approach that places more emphasis on how obese people behave and
function within the surrounding food environment. On one hand, food choices
and dietary behaviors are driven by biology and the brain. On the other hand,
access to healthy foods and opportunities for physical activity can be limited by
material resources, transportation, and the built environment. New studies on
spatial epidemiology and behavioral economics provide a contextual framework
for cutting-edge research on biology, human development and behavior.
The chief aim of this workshop was to summarize some of the key issues in
obesity treatment and prevention in order to promote novel and interdisciplin-
ary approaches and to explore cutting-edge ideas that spanned child develop-
ment, nutrition, behavioral sciences, economics, geography and public health.
Preface IX
This page intentionally left blank
Foreword
XI
We highly appreciate the renowned chairpersons Prof. Barbara Rolls and
Prof. Adam Drewnowski who prepared such a high level of scientific program
and brought together distinguished world experts in the area of obesity.
Finally, we would like to acknowledge the great work and organization of
Laura Taylor and her team from Jenny Craig, thanks to which we could enjoy
the scientific discussions in SanDiego, Calif.
XV
Prof. Michael Rosenbaum Denise Deming/USA
Columbia University Medical Center Llona Fordham/USA
Division of Molecular Genetics Chavanne Hanson/USA
Department of Pediatrics Linda Hsieh/USA
Russ Berrie Pavilion Maureen Huhmann/USA
1150 St. Nicholas Avenue, Rm 620 Wendy Johnson-Askew/USA
New York, NY 10032 Bill Kesting/USA
USA Jennifer Lovejoy/USA
E-Mail: mr475@columbia.edu Lauren McDonnell/USA
Juan Ochoa/USA
Attendees Elizabeth Roark/USA
Sachiko T. St. Jeor/USA
Denis Barclay/Switzerland Heidi Storm/USA
Ferdinand Haschke/Switzerland Roland Sturm/USA
Petra Klassen/Switzerland Lisa Talamini/USA
Laura Taylor/Switzerland Molly Wangsgaard/USA
Jamy Ard/USA Hope Warshaw/USA
Molly Bray/USA Christine Zoumas/USA
Anne Dattilo/USA
XVI Contributors
Obesity Treatment: Challenges and Opportunities
Drewnowski A, Rolls BJ (eds): Obesity Treatment and Prevention: New Directions.
Nestlé Nutr Inst Workshop Ser, vol 73, pp 1–20,
Nestec Ltd., Vevey/S. Karger AG., Basel, © 2012
Abstract
The long-term stability of bodyweight despite wide variation in energy intake and expen-
diture suggests that at usual weight energy intake and output are ‘coupled’ to maintain
body energy stores. Our model for some of the molecular mechanics of this regulation of
energy stores is based on the concept of a neurally encoded ‘threshold’ for minimum
body fat, below which compensatory physiology is invoked to restore body fat. The exis-
tence of such a centrally encoded threshold is supported by the similarities in response to
maintenance of a reduced weight between lean and obese individuals, and the tendency
for weight-reduced individuals to regain weight to levels of fat stores similar to those
present prior to initial weight loss. Brain responses to food and the observed changes in
energy expenditure that occur during maintenance of a reduced weight are largely
reversed by the administration of the adipocyte-derived hormone, leptin.
Copyright © 2012 Nestec Ltd., Vevey/S. Karger AG, Basel
Obesity has become the most prevalent and costly nutritional problem in the
United States, and currently accounts for over 15% of total US health care
spending [1]. Modest (10%) weight loss will prevent or ameliorate many of the
major medical/metabolic consequences of obesity [2]. While most patients can
achieve such weight loss by conventional means, the majority cannot maintain
the reduced weight [3] for extended periods of time.
The long-term constancy of bodyweight (the average American adult gains
only about 0.5–1.5 kg per year despite ingesting over 900,000–1,000,000 kcal),
the 80–85% recidivism rate to previous levels of adiposity following otherwise
successful weight loss, the observation that individuals successful at maintaining
weight loss engage in dietary restriction and increased physical activity com-
pared to weight-matched controls, and the demonstration that there is similar
metabolic opposition to sustained weight loss in both lean and obese individu-
als all support the view that energy stores are physiologically regulated around
an individualized centrally perceived ideal (based on genetics, development,
and environment) [4–7]. The long-term persistence of hypometabolism [8] and
hyperphagia [9] in weight-reduced individuals compared to themselves prior to
weight loss or to individuals ‘naturally’ at the same weight, provides ample evi-
dence for the ‘biological’ basis of the difficulties in sustaining weight loss.
The steadily increasing prevalence of obesity suggests that metabolic
‘defenses’ against gain of fat are inherently weaker than those resisting its loss.
One model for the molecular mechanics of this regulation is based on the con-
cept of a centrally encoded ‘threshold’ for minimum body energy stores (fat).
The critical role of fat stores in reproduction and survival during periods of
undernutrition supports the idea of evolutionary ‘emphasis’ on preserving
somatic fat stores [6]. These inferences are consistent with genetic arguments
related to the so-called ‘thrifty genotype’ [10] and suggest that the conventional
‘thermostatic’ model of ‘set-point’ regulation of body fat [11] is not correct.
More likely, the control system is designed to keep body fat above a critical
lower limit or ‘threshold’ against threats of physical or reproductive extinction
during times of undernutrition. As a corollary, relative metabolic and behav-
ioral ‘leniency’ regarding increases in body fat would be anticipated for their
survival advantage.
The genes and developmental processes that affect bodyweight do so by
affecting the molecular and structural components of a CNS system that sense
and react to ambient concentrations of leptin, insulin, metabolic substrates,
and other molecules that reflect the mass and functional status of somatic fat
(energy) stores. This construct operates as a threshold sensor for relevant sig-
nals such as leptin. The threshold for any ligand (e.g. leptin) is determined by
the functional sensitivity (by virtue of differences in rates of expression or struc-
tural/functional integrity) of its central and peripheral molecular components,
e.g. leptin, insulin, ghrelin, melanocortin, and melanocyte-concentrating hor-
mone receptors, and various orexigenic and anorexigenic neuropeptides, and
other molecules such as AMP-kinase, acetyl CoA-carboxylase, carnitine palmi-
toyl transfersase-1, malonyl CoA, FoxO1, PI3 kinase, etc. [6, 12].
In this model a major, but not the only, afferent signal is the adipocyte-
derived hormone leptin. Subthreshold circulating and CSF concentrations of
leptin invoke changes in energy expenditure and intake that result in regain of
bodyweight (fat) [6, 13]. The threshold determines the signal intensity required
to affect gene expression, neural connections, other cellular elements (glia) [6].
Both the level at which this threshold is set and the intensity of responses to
deviations below this threshold are the result of genetic, developmental, and
environmental (e.g. nutritional) factors.
2 Rosenbaum · Leibel
Obese individuals are still frequently perceived as willful agents of their
excess adiposity by community [14], health professionals [15], and even by
themselves [14]. The inability of most individuals to sustain weight loss is attrib-
uted to a psychological lack of ‘will power’ (somehow lean individuals who are
unable to sustain even a small degree of weight loss are spared this bias) [15].
In the context of this threshold model, the metabolism and behaviors that make
it so difficult for even highly motivated individuals to sustain weight loss are
predominantly the predictable biological consequences of CNS-mediated pro-
cesses that occur as a result of decreased energy stores rather than indications of
a pathological lack of willpower that is somehow unique to formerly overweight
or obese individuals. The obese and never-obese differ primarily in the level at
which this threshold for body fatness is set rather than in response to deviations
below that threshold. Obesity is thus a chronic disease that continues to mani-
fest itself through persistent metabolic and behavioral opposition to sustaining
a reduced bodyweight even after the physical and comorbid manifestations of
the illness may have been ‘cured’ by weight reduction [5–9].
TEE = Twenty-four-hour energy expenditure; REE = resting energy expenditure; NREE = non-
resting energy expenditure; T3 = triiodothyronine; T4 = thyroxine; TSH = thyroid-stimulating
hormone; rT3 = reverse T3; MHCI = myosin heavy chain I; SERCA2 = sarcoplasmic endoplasmic
reticulum Ca2+-dependent ATPase 2.
4 Rosenbaum · Leibel
energy expenditure (decreased) and intake (increased) [22]. Data regarding
adaptive thermogenesis in response to reduced weight maintenance are pre-
sented in table 1.
The decline in energy expenditure following weight reduction would have
little clinical consequence if energy intake were proportionately reduced.
However, maintenance of a reduced bodyweight, as well as other ‘low-leptin’
states (see below), is also associated with decreased satiation and increased hun-
ger, thus creating the optimal biological circumstance to promote weight gain in
leptin-deficient and weight-reduced humans and rodents [6, 7, 25]. Individuals
maintaining a reduced bodyweight after non-surgical weight loss are inclined
towards greater rather than diminished energy intake [9, 17]. More specifically,
subjects maintaining a reduced weight show delayed satiation, decreased per-
ception of how much food has been eaten, decreased desire to switch to a differ-
ent food, increased hunger and craving for food, and an increased preference for
calorically dense foods [6, 7, 9, 17, 26, 27].
Brain areas integrating peripheral signals with neural tracts regulating energy intake
related to food restraint or decreased activity in brain areas related to food reward
as better candidates for lifestyle intervention. Furthermore, it should be possible
either to train individuals or to provide pharmacotherapy to alter their behavior
and reduce risk for weight regain, and to monitor progress by examining brain
function in response to food. For example, transcranial direct current stimula-
tion of the prefrontal cortex has been shown to reduce food cravings [31].
The autonomic (decreased SNS tone and increased PNS tone), neuroen-
docrine (decreased circulating concentrations of bioactive thyroid hormones,
leptin, and, if weight loss is sufficiently large, gonadotropins), and metabolic
(decreased energy expenditure largely due to increased skeletal muscle work
efficiency) effects of sustained weight loss are mediated centrally and, to a lesser
degree, peripherally.
The central and peripheral mechanisms accounting for changes in auto-
nomic, neuroendocrine, and metabolic systems relevant to energy homeostasis
are summarized in table 3.
6 Rosenbaum · Leibel
Table 3. Centrally and peripherally mediated effects of weight loss on systems relevant to energy expenditure
Centrally mediated
↓T3, T4, TSH ↓leptin→↓POMC (ARC) →↓pro-TRH (PVN) No effect of weight loss on TRH response
but leptin reversed decline in T3 and T4
[13]
↓ SNS tone ↓leptin →↑MCH (lateral hypothalamic area) Leptin reverses decline in SNS tone [13]
(norepinephrine) and ↓ CART→ ↓ SNS tone
↑ PNS tone
Delayed satiation and ↓leptin → ↓POMC and ↑ NPY/AgRP Leptin reverses behavioral and neuronal
increased hunger expression and ↑ sensitivity to satiety changes in intake behavior [17, 22]
signals (e.g. CCK) → ↑ energy intake
Peripherally mediated
↓Leptin Reduced weight maintenance → ↓ fat mass Following weight loss, there is little effect
→ ↓leptin; negative energy balance → on circulating leptin per unit fat mass;
↓leptin expression/fat mass during weight loss circulating leptin per
unit fat mass is decreased relative to
during weight stability [32]
↑ Skeletal muscle ↓T3 and SNS tone →↑ expression of more ↑ MHCI and SERCA2 expression, ↑
chemomechanical efficient isoforms of MHC and SERCA → ↑ utilization of FFA as fuel, are largely
efficiency efficiency and utilization of FFA as fuel by reversed by leptin [13, 16, 19, 20]
muscle repletion
REE ↓T3 and SNS tone lead to reduction (~–15%) in REE [21]
ARC = Arcuate nucleus of the hypothalamus; PVN = paraventricular nucleus of the hypothalamus.
Neuroendocrine Function
The decline in circulating concentrations of triiodothyronine (T3) and thyroxine
(T4), without a compensatory increase in thyroid-stimulating hormone (TSH)
production, following weight loss indicates central effects of weight loss on the
hypothalamic-pituitary-thyroid (HPT) axis. TSH release in response to thyroid-
releasing hormone (TRH) is not diminished following weight loss; therefore,
the predominant effect of weight loss on the HPT axis must be via decreased
production of TRH. As discussed below, this most likely occurs as the result of
decreased hypothalamic (arcuate nucleus) pro-opiomelanocortin (POMC) in low
leptin states coupled with leptin-sensitive decreased activity of prohormone con-
vertases in the arcuate and paraventricular nuclei of the hypothalamus [39]. The
net effect of decreased expression of POMC and the prohormone convertases
would be decreased production of the cleavage product of POMC, α-melanocyte-
stimulating hormone, which stimulates production of hypothalamic pro-TRH
in the paraventricular nucleus of the hypothalamus. The increase in circulating
concentrations of reverse T3, the bioinactive enantiomer of T3, is due to decreased
clearance of T3 as a result of decreased expression of deiodinase 1 and decreased
transport of T4 into the liver and kidney where it would normally be converted
8 Rosenbaum · Leibel
to T3 (resulting in a net increased availability of T4 for conversion to reverse T3).
Leptin is made almost exclusively by adipose tissue, and the decline in leptin
following weight loss is mostly a reflection of decreased adipocyte volume. The
even more pronounced decline in leptin during active weight loss reflects direct
metabolic effects on rates of leptin production per adipocyte [40].
Skeletal Muscle
The increase in skeletal muscle work efficiency is of sufficient magnitude to
account for most of the reduction in energy expenditure following weight loss
[5, 8, 13, 16, 19, 20]. The centrally and peripherally mediated autonomic and
neuroendocrine changes described above directly impact skeletal muscle. As
noted above, the molecular phenotype of skeletal muscle in individuals follow-
ing dietary weight loss is one of increased expression of the more chemome-
chanically efficient myosin heavy chain I (MHCI) and SERCA2 isoforms [5,
16]; these molecular phenotypes are characteristic of slow twitch muscle fibers
whose chemomechanical efficiency is greater than that of fast twitch fibers.
The promoter region of the MHCI isoform contains inhibitory T3 response ele-
ments that result in increased MHCI expression in hypothyroid states [41]. The
decline in SNS tone following weight loss promotes decreased expression of
the more powerful, less efficient, MHCII isoform, decreases thermogenesis in
brown adipose tissue, and contributes to the decreased circulating concentra-
tions of T3 [5, 24]. The slow-twitch muscle isoform (SERCA2) is encoded by a
gene whose transcription is inhibited by T3 (unlike SERCA1 whose expression
is stimulated by T3) [16], and β-adrenergic stimulation decreases the expres-
sion of SERCA2a in cardiac muscle and increases the ratio of SERCA1/2 in
skeletal muscle [42]. Skeletal muscle expresses both the long- and short-forms
of the leptin receptor. Perfusion of isolated mouse hearts with leptin results
in increased fatty acid oxidation and decreased chemomechanical work effi-
ciency, which is consistent with findings of increased muscle work efficiency in
low leptin states [13, 19].
Caloric Intake
Caloric intake, assuming that there is ad libitum availability of nutrients, repre-
sents the central nervous system’s response to the sum total of multiple internal
and external sensory inputs. These inputs include short-term (e.g. glucose) and
longer-term (e.g. leptin) internal ‘biological’ signals regarding levels of energy
stored, food anticipation (based in part on environmental stimuli such as the
time of day), hunger, satiation, wanting of food, and liking of the food that is
available [43]. These signals determine how hard an individual will work for
food, what type of food will be preferred, how much will be eaten, how fast, and
when feeding will cease [44]. During dynamic weight loss, i.e. during negative
energy balance, one could argue that changes in energy intake behaviors are due
to alterations in short-term signals relevant to systemic energy status, and not
10 Rosenbaum · Leibel
Table 4. Effects of leptin repletion following weight loss on energy homeostasis [5–7]
Acknowledgements
This work was supported by NIH grants RO1DK64773, RR00645 and UL1 TR000040.
Disclosure Statement
The authors declare that no financial or other conflict exists in relation to the content of
the chapter.
12 Rosenbaum · Leibel
References
1 Tsai A, Williamson D, Glick H: Direct 14 Puhl R, Mass-Racusin C, Schwartz M,
medical cost of overweight and obesity in the Brownell K: Weight stigmatization and bias
USA: a quantitative systematic review. Obes reduction: perspectives of overweight and
Rev 2011;12:50–61. obese adults. Health Educ Res 2008;23:
2 Aronne L, Isoldi K: Overweight and obesity: 347–358.
key components of cardiometabolic risk. 15 Brownell K, Puhl R, Schwartz M, Rudd L
Clin Cornerstone 2007;8:29–37. (eds): Weight Bias: Nature, Consequences,
3 Phelan S, Wing R: Prevalence of successful and Remedies. New York, Guilford Press,
weight loss. Arch Int Med 2005;165:2430. 2005.
4 Tsai A, Wadden T: Systematic review: an 16 Baldwin K, Joanisse D, Haddad F, et al:
evaluation of major commercial weight loss Effects of weight loss and leptin on skeletal
programs in the United States. Ann Int Med muscle in human subjects. Am J Physiol
2005;142:56–66. Regul Integr Comp Physiol 2011;301:
5 Rosenbaum M, Leibel R: Adaptive ther- R1259–R1266.
mogenesis in humans. Int J Obes 2010;34: 17 Kissileff H, Thornton M, Torres M, et al:
S47–S55. Maintenance of reduced body weight in
6 Leibel R, Rosenbaum M: Metabolic response humans is associated with leptin-reversible
to weight perturbation; in Clément K (ed): declines in satiation. Am J Clin Nutr, in
Novel Insights into Adipose Cell Functions, press.
Research and Perspectives in Endocrine 18 Ravussin E, Lillioja S, Anderson T, et al:
Interactions. Heidelberg, Springer, 2010, Determinants of 24-hour energy expenditure
pp 121–133. in man. Methods and results using a respira-
7 Rosenbaum M, Kissileff H, Mayer L, et al: tory chamber. J Clin Invest 1986;78:
Energy intake in weight-reduced humans. 1568–1578.
Brain Res 2010;1350:95–102. 19 Rosenbaum M, Vandenborne K, Goldsmith
8 Rosenbaum M, Hirsch J, Gallagher D, Leibel R, et al: Effects of experimental weight per-
R: Long-term persistence of adaptive ther- turbation on skeletal muscle work efficiency
mogenesis in subjects who have maintained in human subjects. Am J Physiol 2003;
a reduced body weight. Am J Clin Nutr 2008; 285:R183–R192.
88:906–912. 20 Goldsmith R, Joanisse D, Gallagher D, et al:
9 Sumithran P, Prendergast L, Delbridge E, et Effects of experimental weight perturbation
al: Long-term persistence of hormonal adap- on skeletal muscle work efficiency, fuel utili-
tations to weight loss. N Eng J Med 2011;365: zation, and biochemistry in human subjects.
1597–1604. Am J Physiol 2010;298:R79–R88.
10 Neel J: Diabetes mellitus: a ‘thrifty’ genotype 21 Leibel R, Rosenbaum M, Hirsch J: Changes
rendered detrimental by ‘progress’? Am J in energy expenditure resulting from altered
Hum Genet 1962;14:353–362. body weight. N Eng J Med 1995;332:
11 Keesey RE: A set point analysis of the regula- 621–628.
tion of body weight; in Stunkard AJ, (ed): 22 Rosenbaum M, Sy M, Pavlovich K, et al:
Obesity. New York, Saunders, 1986, Leptin reverses weight loss-induced changes
pp 144–165. in regional neural activity responses to visual
12 Blouet C, Schwartz G: Hypothalamic nutri- food stimuli. J Clin Invest 2008;118:
ent sensing in the control of energy homeo- 2583–2591.
stasis. Brain Res 2010;209:1–12. 23 Aronne L, Mackintosh R, Rosenbaum M,
13 Rosenbaum M, Goldsmith R, Bloomfield D, et al: Autonomic nervous system activity in
et al: Low dose leptin reverses skeletal mus- weight gain and weight loss. Am J Physiol
cle, autonomic, and neuroendocrine adapta- 1995;38:R222–R225.
tions to maintenance of reduced weight. J
Clin Invest 2005;115:3579–3586.
14 Rosenbaum · Leibel
48 Korner J, Savontaus E, Chua S, et al: Leptin 52 Heymsfield SB, Greenberg AS, Fujioka K,
regulation of Agrp and Npy mRNA in the et al: Recombinant leptin for weight loss
rat hypothalamus. J Neuroendocrinol 2001; in obese and lean adults: a randomized,
13:959–966. controlled, dose-escalation trial. JAMA
49 Legradi G, Emerson C, Ahima R, et al: 1999;292:1568–1575.
Leptin prevents fasting-induced suppres- 53 Ahima R, Prabakaran D, Mantzoros C, et
sion of prothyrotropin-releasing hormone al: Role of leptin in the neuroendocrine
messenger ribonucleic acid in neurons of response to fasting. Nature 1996;382:250–
the hypothalamic paraventricular nucleus. 252.
Endocrinology 1998;138:2569–2576. 54 Schwartz M: Brain pathways controlling
50 Westerterp-Plantenga M, Saris W, Hukshorn food intake and body weight. Exp Biol Med
C, Campfield L: Effects of weekly admin- 2001;226:978–981.
istration of pegylated recombinant human 55 Thaler J, Yi C, Schur E, et al: Obesity is asso-
OB protein on appetite profile and energy ciated with hypothalamic injury in rodents
metabolism in obese men. Am J Clin Nutr and humans. J Clin Invest 2012;122:153–162.
2001;74:426–434. 56 Dietz W: Periods of risk in childhood for
51 Hukshorn C, Westerterp-Plantenga M, Saris the development of adult obesity. J Nutr
W: Pegylated human recombinant leptin 1997;127:1884S–1886S.
(PEG-OB) causes additional weight loss in 57 Power C, Parsons T: Nutritional and other
severely energy-restricted overweight men. influences in childhood as predictors of obe-
Am J Clin Nutr 2003;77:771–776. sity. Proc Nutr Soc 2000;59:267–272.
Discussion
Dr. Drewnowski: Far being from me to ask a provocative question, but one issue that
has been occupying the public and was recently published in the Lancet is the equivalency
between 3,500 calories and one pound of weight gain or loss [1], and this has been the
absolute bible for decades. Every dietitian, every health professional equates 3,500
calories to a pound of weight gain or loss. It does not seem to be exactly true. What is
your position on this?
Dr. Rosenbaum: You are asking, what is the caloric density of weight change? No
generalization is absolutely true, and the caloric density of weight gained or lost depends
upon its composition as fat mass, fat-free mass, and, of course, water. The average
person who loses weight by diet alone loses about two thirds to three quarters of it as fat
and a quarter to a third of it as lean body mass. But people lose weight by different
means in terms of diet composition and exercise. Clearly, if you are losing energy stores
due to the same degree of negative energy balance, the proportion of energy stores lost
as fat mass and fat-free mass will be affected by the type and intensity of the exercise.
The caloric density of fat is roughly 7 times that of muscle, so whether you are losing
energy as fat or muscle is going to make a difference in terms of how much absolute
weight you lose.
Dr. Drewnowski: I am trying to get at the issue of public health policy. There was a
paper published in American Journal of Public Health [2] suggesting that if people in
California went to a fast food restaurant, looked at the menu labeling, and as a result ate
100 calories less, then 100 calories per day per year would equate into millions of pounds
lost and the obesity epidemic in California would go away. This was one spin on the
relation between calories and pounds lost. The other spin was a study published in the
16 Rosenbaum · Leibel
A corollary to your question is whether the changes in energy homeostatic systems
that occur during weight loss are distinct from those that occur during reduced weight
maintenance. We looked at the differences between dynamic weight loss and static
reduced weight maintenance by studying subjects at the very end of weight loss and
comparing them to themselves when maintaining a similar weight. This allows separation
of the effects of dynamic weight change and weight maintenance at similar weights.
There are marked differences between subjects during dynamic weight loss (negative
energy balance) and maintenance at the same weight (eumetabolism). For example,
there is a tremendous decline in resting energy expenditure during dynamic weight loss
and a much more profound decline in thyroid hormones and in leptin during dynamic
weight loss than during static weight maintenance. It should also be noted that leptin is
not the only signal operant in states of weight loss. For example, we have found that the
increase in parasympathetic nervous system tone and decrease in TSH persist in weight-
reduced individuals even after leptin repletion [12]. So, there are clearly signals affecting
energy expenditure that are related to energy balance as opposed to absolute bodyweight,
i.e. those are present during weight loss as distinct from reduced weight maintenance,
and some of these are not leptin dependent, i.e. those that are evident even in leptin-
deficient mice. Your question as to whether genotype would be predictive of phenotypes,
such as the composition of weight loss or the decline in energy expenditure or thyroid
hormones during dynamic weight loss is a good one that we have not examined.
Dr. Bray: Because Dr. Drewnowski brought up epigenetics, it is important to note
that there is more and more evidence that at least methylation is much more labile than
we think it is. Is it possible that when you are giving a liquid diet for such long periods of
time, the methylation patterns can actually change in a way that turns on and off genes
that results in what you have seen which is depressed metabolism?
Dr. Rosenbaum: This is an interesting question, but we have not looked at the effects of
weight loss or a liquid formula diet on the methylation of relevant genes.
Dr. Bray: Related to that is if you gave 800 calories of lettuce or something that was
not your liquid diet, do you think you would see exactly the same response?
Dr. Rosenbaum: The composition of weight lost is clearly going to be somewhat
sensitive to diet composition. On a high carbohydrate or high salt diet one might
anticipate a smaller initial loss of water weight. Sacks et al. [13] looked at the effect of diet
macronutrient composition (low fat versus high fat, average protein versus high protein,
and highest and lowest carbohydrate content) on the change in bodyweight over 2 years.
At 6 months, participants had lost about 7% of their initial bodyweight with no group
differences. Some weight regain had occurred in all groups at 12 months. No between-
group differences were evident at 2 years. The authors concluded that there was no
evidence of any major macronutrient influence on weight loss.
There is also a substantial body of data suggesting that diet composition does not
have a big effect on energy expenditure during maintenance of a reduced weight. Rudy
Leibel showed that wide variations in macronutrient composition of a liquid formula
diet did not affect energy requirements to maintain weight in inpatients [14]. It should
be noted that subsequent to this meeting, Ludwig’s group [JAMA 2012;307:2627–2634]
reported that subjects who had lost weight had greater energy expenditure on a very
low-carbohydrate diet than on an isocaloric low glycemic index or low-fat diet; while
this would suggest that some of the decline in TEE and REE following weight loss may
be attenuated on a very low-carbohydrate diet, the anticipated clinical consequences of
18 Rosenbaum · Leibel
Dr. St. Jeor: I am interested in phenotyping too. Would you include the T3 levels and
compare them against some other norms? I think we are returning to the time when
everybody thought that people were obese because their thyroid was not functioning,
and thyroid medication seemed to be the answer. Do you have any thoughts on how you
might approach that and include that in your assessment and your treatment?
Dr. Rosenbaum: We are looking at this now in the same way we did in our leptin
studies [20, 21]. We are doing thyroid studies of subjects at their usual weight and
following weight loss while they are receiving a placebo or T3 repletion in a crossover
design. Physiological doses of thyroid hormone would not promote weight loss but may
assist in reduced weight maintenance. The basic idea is that the treatment to maintain a
negative energy balance and lose weight is not going to be the same as remaining in a
eumetabolic state to maintain weight loss.
Dr. Drewnowski: Another idea is that the brain is incapable of perceiving calories
presented in liquid form. What’s your take on that?
Dr. Rosenbaum: I think the brain knows when and what you have been eating. The
similarities between our studies and the outpatient studies of Weinsier and of the
National Weight Control Registry would all suggest that the brain perceives ingested
calories regardless of the physical form in which they are presented to the gut. In the
same vein, many feeding studies are done with subjects ingesting milkshakes or other
liquids, and it seems that the intake of these liquids is predictably affected by CCK and
other interventions. It is likely that the central nervous system perception of calories
given intravenously, thereby bypassing most if not all gut peptide signaling, would be
different from that of ingested calories.
Dr. Finegood: Maybe there is a difference between sugar-sweetened beverages and a
mixed meal.
Dr. Rosenbaum: If you are asking whether there are differences in how our brains
react to sugar-sweetened beverages versus a mixed meal, then I would say that there
probably are significant distinctions. Diet composition will acutely affect multiple signals
regarding energy homeostasis including CCK, ghrelin, and other gut peptides, and all of
these will impact on appetite. The immediate biochemical responses to what has been
eaten might be viewed as signals predominantly reflecting short-term energy intake,
while others that reflect energy stores, such as leptin, might be viewed as signals better
reflecting longer-term energy balance.
References
1 Hall KD, Sacks G, Chandramohan D, et 3 Chen L, Appel LJ, Loria C, Lin P-H, et al:
al: Quantification of the effect of energy Reduction in consumption of sugar-sweet-
imbalance on bodyweight. Lancet 2011;378: ened beverages is associated with weight loss:
826–837. the PREMIER trial. Am J Clin Nutr 2009;
2 Kuo T, Jarosz CJ, Simon P, Fielding JE: Menu 89:1299–1306.
labeling as a potential strategy for combat- 4 Rosenbaum M, Leibel RL: Adaptive ther-
ing the obesity epidemic: a health impact mogenesis in humans. Int J Obes 2010;34:
assessment. Am J Public Health 2009;99: S47–S55.
1680–1686.
20 Rosenbaum · Leibel
Obesity Treatment: Challenges and Opportunities
Drewnowski A, Rolls BJ (eds): Obesity Treatment and Prevention: New Directions.
Nestlé Nutr Inst Workshop Ser, vol 73, pp 21–36,
Nestec Ltd., Vevey/S. Karger AG., Basel, © 2012
Abstract
Weight loss interventions involving diet and physical activity typically result in 8–10%
weight loss within 6–12 months after initiating treatment. Physical activity is a key com-
ponent of these interventions for a variety of reasons. Weight loss achieved with physical
activity averages approximately 1–3 kg, and the effects of physical activity on weight loss
appear to be additive to what is observed with dietary restriction alone. Moreover, physi-
cal activity is an important behavior for prevention of weight regain and maintenance of
significant weight loss resulting from dietary restriction, and physical activity contributes
to weight loss in patients who have undergone bariatric surgery. However, there is signifi-
cant interindividual variability in the weight loss resulting from physical activity, with both
biological and behavioral factors contributing to this variability. Thus, additional research
is needed to understand the role of physical activity in energy balance and body weight
regulation, along with an understanding of the optimal intervention strategies to pro-
mote physical activity participation in overweight and obese individuals.
Copyright © 2012 Nestec Ltd., Vevey/S. Karger AG, Basel
Introduction
22 Jakicic
activity appears to improve weight loss by 2–3 kg above the magnitude achieved
with the dietary intervention alone. For example, Goodpaster et al. [13] studied
adults classified with class II or III obesity and compared a diet only interven-
tion prescribed at 1,200–2,100 kcal/day with the diet intervention combined
with physical activity over a period of 6 months. Weight loss resulting from the
diet alone intervention was 8.2 kg compared to 10.9 kg in the combined diet and
physical activity intervention, a difference of 2.7 kg. Wing et al. [10] reported
that the combination of diet plus physical activity improved weight loss by 1.2
kg compared to the magnitude of weight loss observed with a diet intervention
alone (9.1 vs. 10.3 kg). These findings are consistent with the conclusions based
on a review conducted by Curioni and Lourenco [14].
While the magnitude of weight loss achieved with physical activity over a
period of 6 months may be considered to be modest, physical activity may play
an important role in the maintenance of weight loss. As indicated above, it is
common for initial weight loss to be followed by weight regain of the magni-
tude of 33–50% over a period of 1–3 years [4]. However, there is evidence that
physical activity may enhance long-term weight loss, or minimize or prevent
weight regain. For example, Jakicic et al. [15] reported that individuals who
achieved a weight loss of 14.2 kg (16.8% of initial bodyweight) at 24 months
had increased their physical activity to approximately 1,500 kcal/week above
baseline levels, whereas individuals achieving less weight loss were engaging
in physical activity that was on average <700 kcal/week above baseline levels.
In another study that involved the combination of diet plus physical activity,
Jakicic et al. [16] reported that overweight and obese women who engaged in
physical activity of 291 min/week from 0 to 6 months, 282 min/week from 7 to
12 months, and 281 min/week from 13 to 18 months reduced weight by 13.1
kg by the end of the 18-month intervention. This weight loss at 18 months was
significantly greater than the 8.2 and 3.5 kg observed in individuals averaging
approximately 210 or 121 min/week of physical activity across the 18-month
intervention, respectively. Unick et al. [17] have also reported that physical
activity is an important predictor of the ability to lose ≥10% of initial body-
weight within 6 months and to sustain this weight loss at 24 months. These
findings support the importance of physical activity for improving long-term
weight loss. However, as indicated within the 2009 ACSM Position Stand [12],
support for physical activity to improve long-term weight loss comes mainly
from secondary post-hoc analyses or observation studies. This may suggest the
need for additional randomized studies with sufficient sample sizes to better
understand the role of physical activity for improving long-term weight loss
and prevention of weight regain.
Physical activity has also been shown to be important for patients who have
undergone bariatric surgery to induce weight loss. It has been shown that phys-
ical activity contributes to improvements in weight loss at 6–24 months fol-
lowing bariatric surgery [18–20]. Evans et al. [20] reported greater weight loss
in bariatric surgery patients who participated in ≥150 min per week of physi-
cal activity compared to those patients participating in <150 min per week.
Thus, it is important for patients who undergo bariatric surgery to participate
in sufficient amounts of physical activity to maximize the weight loss achieved
with bariatric surgery. However, as summarized by O’Brien [21], research gaps
remain in our understanding of the role of physical activity for patients who
have undergone bariatric surgery, and this warrants additional research.
24 Jakicic
12
10
8
6
Weight change (kg)
4
2
0
–2
–4
–6
–8
–10
–12
–14
26 Jakicic
and obese adults, and the development and implementation of interventions to
overcome these reported barriers.
Behavioral
Health-related
weight loss
outcomes
intervention
Bodyweight
Diet and
adiposity
Fig. 3. Theoretical pathway for physical activity to influence bodyweight and health-
related outcomes.
In addition to the potential benefits of physical activity on weight loss, there are
additional health benefits that may be realized in overweight and obese individ-
uals through participation in physical activity. One of the key benefits of physi-
cal activity is an improvement in cardiorespiratory fitness. Recent reviews of the
literature concluded that higher levels of physical activity or cardiorespiratory
fitness are associated with improvements in health-related outcomes [35–37].
The association between higher levels of physical activity or cardiorespiratory
fitness and reduced health risk is present in overweight and obese adults even
in the absence of weight loss. Thus, overweight and obese adults can realize
significant health benefits from participation in physical activity, and the added
weight loss may further enhance these benefits, as illustrated in figure 3.
Conclusions
Physical activity can contribute to both short-term and long-term weight loss in
overweight and obese adults. These benefits of physical activity are also present
in individuals who have undergone bariatric surgery. In addition, physical activ-
ity and the concurrent improvement in fitness are associated with improvements
in a variety of health-related outcomes. However, there is significant interindi-
vidual variability in the weight loss that is achieved with physical activity, with
28 Jakicic
both behavioral and biological factors contributing to these findings, which
warrants further investigation. Regardless, physical activity should be included
as a component of a comprehensive intervention for weight loss for overweight
and obese adults.
Disclosure Statement
Dr. Jakicic reported serving on the scientific advisory board for Alere Wellbeing, has
received an honorarium for a scientific presentation from Jenny Craig and from the
Nestle Nutrition Institute, and has served as the Principal Investigator on research grants
awarded to the University of Pittsburgh from the Beverage Institute for Health and
Wellness and BodyMedia, Inc.
References
1 Flegal KM, Carroll MD, Ogden CL, Curtin LR: 8 Jakicic JM, Otto AD, Semler L, et al: Effect of
Prevalence and trends in obesity among US physical activity on 18-month weight change in
adults, 1999–2008. JAMA 2010;303:235–241. overweight adults. Obesity 2011;19:100–109.
2 National Institutes of Health National 9 Hagan RD, Upton SJ, Wong L, Whittam J:
Heart Lung and Blood Institute. Clinical The effects of aerobic conditioning and/or
Guidelines on the Identification, Evaluation, calorie restriction in overweight men and
and Treatment of Overweight and Obesity women. Med Sci Sports Exerc 1986;18:87–94.
in Adults – The Evidence Report. Obes Res 10 Wing RR, Venditti EM, Jakicic JM, et al:
1998;6(suppl 2):71S–76S, 96S. Lifestyle intervention in overweight indi-
3 Wing RR: Behavioral Weight Control; in viduals with a family history of diabetes.
Wadden TA, Stunkard AJ (eds): Handbook of Diabetes Care 1998;21:350–359.
Obesity Treatment. New York, The Guilford 11 Physical Activity Guidelines Advisory
Press, 2002, pp 301–316. Committee Report 2008. US Department of
4 Perri MG, Corsica JA: Improving the mainte- Health and Human Services, 2008; http://
nance of weight lost in behavioral treatment www.health.gov/paguidelines/committeere-
of obesity; in Wadden T, Stunkard AJ (eds): port.aspx.), accessed January 19, 2009.
Handbook of Obesity Treatment. New York, 12 Donnelly JE, Jakicic J, Blair SN, et al: ACSM
The Guilford Press, 2002, pp 357–379. position stand on appropriate intervention
5 Diabetes Prevention Program Research strategies for weight loss and prevention
Group. Reduction in the incidence of type 2 of weight regain for adults. Med Sci Sports
diabetes with lifestyle intervention or met- Exerc 2009;42:459–471.
formin. N Engl J Med 2002;346:393–403. 13 Goodpaster BH, DeLany JP, Otto AD, et al:
6 Look AHEAD Research Group: Reduction Effects of diet and physical activity interven-
in weight and cardiovascular disease risk tions on weight loss and cardiometabolic risk
factors in individuals with type 2 diabetes: factors in severely obese adults: a random-
one-year results of the Look AHEAD trial. ized trial. JAMA 2010;304:1795–1802.
Diabetes Care 2007;30:1374–1383. 14 Curioni CC, Lourenco PM: Long-term
7 Look AHEAD Research Group. Long-term weight loss after diet and exercise: systematic
effects of a lifestyle intervention on weight review. Int J Obes 2005;29:1168–1174.
and cardiovascular risk factors with type 15 Jakicic JM, Marcus BH, Lang W, Janney C:
2 diabetes: four year results of the Look Effect of exercise on 24-month weight loss
AHEAD Trial. Arch Int Med 2010;170: in overweight women. Arch Int Med 2008;
1566–1575. 168:1550–1559.
30 Jakicic
Discussion
Dr. Bray: You made a statement that dietary modification and exercise are effective
for weight loss, and the combination of these behaviors is more effective than either one
alone. If people persist in these behaviors, they will maintain their current weight or
their weight loss. I am interested in your comments about whether our focus in
interventions should be on adherence to these behaviors and understanding why
adherence diminishes over time despite initial success at weight loss. This may require a
better understanding of both metabolic and/or psychological factors that may be
influencing adherence.
Dr. Jakicic: It is clear that the combination of an energy-restricted diet and physical
activity is the most effective non-surgical and non-pharmacological intervention for
inducing and maintaining weight loss. Thus, the focus of interventions should be on
promoting adoption and maintenance of these key behaviors. Results of 18- to
24-month studies conducted in our laboratory have supported that both components
of energy balance result in improved weight loss [1, 2]. Behavior change is complex,
with many interactions that need to be considered to understand the most effective
strategies for improving adoption and adherence to these key behaviors related to
energy balance and weight control. For example, when considering behavioral
constructs, data from our studies have consistently shown that self-efficacy is predictive
of adoption and maintenance of exercise behavior in overweight and obese adults.
What is interesting is that self-efficacy for physical activity and structured exercise
prior to initiating an intervention are not predictive of behavior change. Rather, it is
the change in self-efficacy once the intervention is undertaken that is predictive of
future physical activity behavior. This suggests that we need to focus our interventions
on exposing participants to physical activity experiences that allow them to realize that
they can do a physical activity and that this activity can be beneficial to them. Too
many times clinicians may simply prescribe activity to patients without addressing the
issues that may influence self-efficacy, which ultimately does not result in meaningful
or sustained behavior change. In addition, we need to consider the barriers that exist
related to modifying eating and activity behaviors. For example, simply instructing a
patient to exercise without also addressing barriers will be less effective than an
intervention that is specifically designed to identify and address the barriers to exercise.
Moreover, eating and exercise are not independent behaviors. For some individuals,
exercise has been shown to stimulate a decrease in energy intake, and for others
exercise has been shown to stimulate an increase in energy intake. Whether this is a
result of cognitive control or other metabolic or physiological control mechanisms is
not well understood. Regardless of the mechanism, the interaction of these behaviors
may be important in understanding factors that influence adoption, maintenance, and
effectiveness of these behaviors for weight control, and this needs further
investigation.
Dr. Bray: Another observation from our studies is that adherence is associated with
higher intensity exercise rather than with the moderate-intensity exercise in many of the
public health recommendations. We may need to consider this as well when examining
factors that influence adherence.
Dr. Jakicic: This may be related to how exercise intensity is defined. For example,
typical public health recommendations for physical activity intensity are based on data
32 Jakicic
physiological adaptations that influence weight and health outcomes. To date, most of
what we know about this has been from observational or cross-sectional studies, with
few studies using well-designed paradigms to better understand these important research
questions.
Dr. Ard: Observations from our studies have suggested that self-efficacy may decline
over a period of an 18-month intervention despite exposure to well-designed and
implemented behavioral interventions. This may be a result of initial perceptions related
to ease of behavior change or the health benefits of change in these behaviors not
matching what actually occurs once exposed to the intervention. Thus, the perception is
reset based on these initial experiences which influence self-efficacy. For example, many
individuals undertake exercise to reduce bodyweight, and after 2 months see very little
change in bodyweight, which results in a change in their perspective of the importance
of exercise which ultimately influences adherence to this behavior. Moreover, regarding
weight maintenance, it appears that there is a need for individuals to continue to have to
increase their dose of exercise over time to continue to maintain their weight loss, which
may influence long-term adherence, and this may be a result of continued muscle
efficiency resulting from exercise participation. It may be important to consider factors
around the optimal timing and dosing of exercise in the course of weight reduction and
in the course of weight maintenance, so that this continued increase in exercise dose
may not be necessary to maintain weight loss.
Dr. Jakicic: Your observation that self-efficacy for physical activity is difficult to
sustain long-term is consistent with data from our own studies. We have shown that in
response to a behavioral weight loss intervention, self-efficacy for physical activity
increases over the initial 6 months of the intervention, with a gradual decline thereafter.
However, when we examine the data in greater detail, participants who maintain higher
levels of physical activity at 12 months continue to also sustain higher levels of self-efficacy
for physical activity. This suggests that maintaining self-efficacy needs to remain a focal
point of behavioral interventions. However, it is important to better understand factors
that contribute to the ability of some participants to maintain higher levels of self-efficacy
versus others being unable to maintain self-efficacy for physical activity, which appears to
be associated with the inability to sustain physical activity behavior. Whether this is a
result of a physiological drive due to improved muscle efficiency needs further examination
and understanding. This decline in self-efficacy over time may also be a result of the
modest weight loss of 0.5–3.0 kg that occurs with regular exercise, which may be less
weight loss than most individuals expect based on the effort required to engage in
sufficient levels of physical activity. Thus, further work is needed to understand how to
improve the effectiveness of our interventions for sustaining physical activity.
Dr. Haschke: With regard to the measurement of body composition, all of the devices
rely on certain assumptions. One of those assumptions is that the percentage of water in
fat-free mass is constant. However, during a dynamic process like weight loss, water
content in fat-free mass is changing, and therefore our assumptions may not be correct,
and this may influence the accuracy of the measurement. My question is whether other
traditional measures such as subcutaneous fat measurement using skinfold measurement
or just abdominal circumference would be sufficient from a clinical perspective because
this may provide valuable information relative to health risk. It may not be body
composition per se, but rather changes in fat deposition at specific anatomical sites that
is important to measure and to understand.
34 Jakicic
amount of supervised exercise led to weight loss for men but prevention of weight gain
for women. However, an interesting finding in this study was that for a fixed duration
and intensity of exercise, the men expended approximately 600 calories per session and
the women expended about 400 calories per session, and this probably contributed to
the observed difference in weight change in response to the exercise. In studies in which
the energy expenditure of the exercise was held constant for both men and women,
similar reductions in bodyweight were observed. To achieve this similar energy
expenditure, women would have to exercise longer or at a higher intensity compared to
men, which may also have implications from an adherence perspective. Finding the
balance between energy expenditure and energy intake that results in optimal adherence
may be necessary from a clinical perspective to make these programs feasible and
effective for both men and women.
Dr. Rosenbaum: The evidence presented illustrates the challenges for behavioral
programs to elicit and sustain substantial weight loss. This may suggest that the
emphasis should be on interventions to behaviors to improve health rather than
behaviors to lose weight. If true that a high volume of exercise is required to lose and
maintain significant weight loss, this may pose significant barriers for many individuals
due to the time commitment that is required. So, perhaps we should de-emphasize the
influence of behavior change on weight loss because for many this may be modest, and
instead place a greater emphasis on good health that may include resting heart rate,
blood pressure, lipids and lipid particle size, and other health outcomes. Maybe this will
influence behavior change to a greater magnitude than placing the emphasis on
bodyweight.
Dr. Jakicic: I agree that ultimately the focus should be on health-related outcomes.
However, I do not believe that there is evidence from clinical trials showing that a focus
on these health outcomes improves adherence compared to a focus on weight loss.
Substantial work remains to be done to better understand how to best influence behavior
change and whether the focus of interventions should be on weight or some other health
outcome. We also have to consider the dose of key behaviors patients can adopt and
maintain to influence desired health-related outcomes.
Dr. Goran: You alluded to the health effects of exercise that appear to be independent
of the health effects of weight loss. Is there dose at which the health effects of exercise are
observed?
Dr. Jakicic: There is an important influence of exercise on many health outcomes,
and the health improvements appear to be present, independent of changes in
bodyweight. In general, the typically recommended minimal dose of 150 min of
moderate-intensity physical activity per week appears to be sufficient. However, public
health recommendations for physical activity also state that additional health benefits
may be achieved with more physical activity. Whether significant health benefits can be
achieved with a lower dose or intensity of physical activity is less clear from the available
evidence.
Dr. Bray: The majority of research supporting physical activity and the dose of
physical activity appears to be based on research that examined cardiovascular forms of
activity. Are similar data available for resistance training?
Dr. Jakicic: Within the past few years, the American College of Sports Medicine and
American Heart Association have more clearly defined the resistance training that may
be necessary to elicit a variety of health benefits. A major benefit of resistance training is
References
1 Jakicic JM, Wing RR, Winters-Hart C: 3 King AC, Haskell WL, Young DR, et al:
Relationship of physical activity to eating Long-term effects of varying intensities and
behaviors and weight loss in women. Med Sci formats of physical activity on participa-
Sports Exerc 2002;34:1653–1659. tion rates, fitness, and lipoproteins in men
2 Unick JL, Jakicic JM, Marcus BH: and women aged 50–65 years. Circulation
Contribution of behavior intervention com- 1995;91:2596–2604.
ponents to 24-month weight loss. Med Sci 4 Donnelly JE, Hill JO, Jacobsen DJ, et al:
Sports Exerc 2010;42:745–753. Effects of a 16-month randomized controlled
exercise trial on body weight and composi-
tion in young, overweight men and women.
Arch Intern Med 2003;163:1343–1350.
36 Jakicic
Obesity Treatment: Challenges and Opportunities
Drewnowski A, Rolls BJ (eds): Obesity Treatment and Prevention: New Directions.
Nestlé Nutr Inst Workshop Ser, vol 73, pp 37–48,
Nestec Ltd., Vevey/S. Karger AG., Basel, © 2012
Abstract
In an ‘obesogenic’ environment, getting people to eat appropriate amounts is challeng-
ing. Several food-based strategies have the potential to promote satiety and moderate
energy intake. Components of foods such as macronutrients and functional ingredients
can affect satiety; however, for weight management a more comprehensive approach is
needed that emphasizes behavioral strategies to improve the overall diet. Research shows
that large portions of energy-dense foods facilitate overconsumption and that reductions
in portion size and energy density are associated with reduced energy intake. While this
suggests that people should eat smaller portions, recent data show that if people lower
the energy density of their diet, they can continue to eat their usual amount of food while
limiting calories. Furthermore, serving larger portions of low-energy-dense foods can be
used strategically to encourage their consumption and reduce dietary energy density,
and this has been shown to be associated with decreased energy intake while maintain-
ing satiety. This new understanding of how portion size can be used positively to manage
energy intake has the potential to help people achieve sustainable improvements in their
energy intake and bodyweight. Science-based strategies that increase the availability of
affordable nutrient-rich, lower energy-dense foods are urgently needed.
Copyright © 2012 Nestec Ltd., Vevey/S. Karger AG, Basel
Biological Approach
38 Rolls
Functional Ingredients
Another major thrust of satiety research is the characterization of functional
ingredients that affect hunger and satiety mechanisms. While much of the focus
in this area has been on various types of fiber [6] or protein [7], the scope for
new products is broad and will continue to grow as the biological bases of hun-
ger and satiety become better understood. However, the utility of such products
for weight management remains to be established [8]. Demonstrating that an
ingredient or a functional food affects hunger, satiety, or energy intake in the
short-term does not indicate that these effects are robust or persistent enough to
resist physiological regulatory systems for the maintenance of energy balance.
Nor does it show that in the context of an obesogenic environment bodyweight
will be affected [9].
A critical question that is raised by studies of food components that enhance
satiety is whether any one of these can have an impact on its own. Is it realistic to
expect the addition of a few grams of fiber or protein to not only control hunger
but also to decrease daily energy intake and reduce bodyweight? Perhaps effects
will only be seen if such food components are consumed in combination or as
part of a more comprehensive approach to both dietary and behavioral strate-
gies that facilitate weight management.
While the emphasis in studies of eating behavior has often been on the mech-
anisms controlling satiety and energy intake, recent data indicate that people
readily ignore biological satiety signals when surrounded by large portions of
palatable, energy-dense foods. Such overconsumption begins early in life [10]
and often becomes a habitual pattern that continues to be reinforced over the
lifespan. Dietary interventions that rely on people making changes in estab-
lished eating patterns to achieve negative energy balance are often unsustainable
in an obesogenic environment.
Changing eating behavior represents one of the greatest challenges in obe-
sity prevention and treatment. Ultimately, parental and childhood interventions
along with government and food industry initiatives are needed. These initia-
tives will benefit from sound research that identifies robust behavioral responses
to food that could help to moderate intake. Such responses include the effects of
portion size and energy density on energy intake.
Portion Size
The goal of much of the research on portion size has been to establish a relation-
ship between large portions of energy-dense foods and overconsumption. It is
clear that portion size has a powerful, sustained effect on the amount of food
300
200
100
0
180 270 360
Vegetable portion size (g)
Fig. 1. Portion size can be used positively to increase vegetable intake and decrease
energy intake at a meal. As the amount of vegetables served at a meal was increased and
the amount of other meal components (meat and grain) was decreased, vegetable intake
of 48 adults increased by half a serving (29%) and energy intake decreased by 40 calories
(7%). Reprinted from Rolls et al. [13].
consumed. This effect has been demonstrated for different types of foods and
beverages in both laboratory and natural environments, and in overweight and
normal-weight men and women [11]. Of particular importance is that bouts of
overeating associated with large portions are not followed by a compensatory
reduction in intake. For example, when the portion size of all foods served over
11 days was increased, there was a persistent and significant effect of portion
size on energy intake [12]. Thus, characteristics of the eating environment such
as the ready availability of large portions of energy-dense foods can override the
regulation of energy balance over prolonged periods.
Since the effects of portion size are robust and sustained, it is possible that
they could be used to increase intake of nutritious, low-energy-dense foods such
as vegetables. Recent dietary advice relies on this premise. The 2010 Dietary
Guidelines for Americans [5] urge people not only to eat less overall and to
avoid oversized portions, but also to increase the proportion of vegetables and
fruits served at a meal. There is some evidence that this could influence intake.
In one study, increasing the proportion of a low-energy-dense vegetable served
on a plate by substituting it for the meat and grain significantly increased veg-
etable intake and reduced energy intake at the meal (see fig. 1) [13]. Other stud-
ies show that in both children and adults increasing the portion of vegetables
served at the start of a meal increases vegetable intake [14] and can decrease
energy intake at a meal. Such findings support the suggestion that variations
in portion size can be used beneficially to influence the types and amounts of
40 Rolls
foods consumed at a meal, but there are no data on the effects of such strategic
manipulations of portion size on sustained changes in intake or bodyweight.
Another portion control strategy that individuals can utilize is to structure
their food environment so that exposure to large portions of energy-dense foods
is limited during several eating occasions in a day. This can be achieved with
pre-portioned foods (PPFs) such as entrées, snack foods, or liquid meals that
are packaged individually in portions appropriate for a meal or snack. Existing
evidence suggests that the consumption of liquid PPFs helps to limit energy
intake and promotes weight loss [15]. There is also limited evidence on the effi-
cacy of solid PPFs for weight management. Providing participants with most
or all of their foods as pre-portioned items facilitated weight loss [16]. It is also
likely that solid PPFs contributed to weight loss and weight loss maintenance
over 2 years in a recent trial that compared a structured weight loss program
including free prepared meals to usual care [17]. Although PPFs show promise
as a useful tool for weight management, little is known about how characteris-
tics such as their energy content and energy density affect efficacy [18]. Nor is
it known if their use leads to better understanding of appropriate portions and
whether their consumption will be continued so that they facilitate weight loss
maintenance.
More research is needed to determine the impact of specific portion control
strategies on body weight regulation. In particular, there is an urgent need for
evidence-based strategies to help consumers limit the overconsumption associ-
ated with large portions of energy-dense foods. In the current environment, it
is difficult for many individuals to eat appropriate amounts of food. Getting
intake back in synchrony with energy needs will be challenging since consum-
ers equate large portions with good value and they have a distorted idea of how
much food is appropriate. If people were to heed the frequently offered advice
simply to ‘eat less’, and were to reduce the portion size of all the foods consumed,
they would probably feel deprived and would not sustain this eating pattern. A
promising approach that would allow people to eat satisfying portions would be
to reduce the energy density of the diet or at least of selected foods.
Energy Density
Dietary energy density has emerged in recent years as one of the most consistent
influences on satiety and energy intake. While the energy density of food is often
related to its fat content, the water content can have an even greater effect. The
combination of water with the macronutrients in foods determines the energy
density, and variations in water can be used to separate the effects of energy density
from those of the macronutrients. A number of systematic studies show that when
the macronutrient content of foods was varied, but the energy density was kept
constant, the effects of fat, carbohydrate, and protein on satiety were similar [19].
On the other hand, the energy density of foods is a robust and significant determi-
nant of satiety and energy intake regardless of macronutrient composition [19].
42 Rolls
with a lower energy density than obese individuals [26]. Furthermore, increases
in dietary energy density were associated with greater weight gain in a prospec-
tive study of 50,000 middle-aged women over 8 years [27].
While data suggest that reducing dietary energy density can facilitate weight
management, more long-term studies are needed to understand how to imple-
ment this approach and facilitate the maintenance of low-energy-dense eating
habits. If people were to adopt lower energy-dense eating patterns, they would
be able to eat satisfying amounts of foods appropriate to meet both energy and
nutrient needs. However, long-term compliance with any diet that requires
deliberate and sustained changes in established eating habits is difficult. A key
question is whether the food environment can be modified to help people lower
the energy density of their diets and to eat appropriate amounts in order to pre-
vent the development of obesity and facilitate weight management.
Environmental Approach
The types of foods that are readily available, affordable, and fit with consumers’
preferences and lifestyle can influence energy intake. This suggests the possi-
bility that the eating environment could be strategically designed to encourage
consumers to choose more foods that are appropriate in portion size and lower
in energy density. A range of strategies has been proposed including educa-
tion and nutritional information such as menu labeling, increased promotion
and availability of low-energy-dense foods such as vegetables and fruits, more
opportunities to choose smaller portions of energy-dense foods, and pricing or
tax incentives to encourage selection of appropriate portions of nutrient-dense,
low-energy-dense foods [28, 29].
At present, there are few data to support the effectiveness of such environmen-
tal approaches for weight management, and they have economic ramifications for
the food and restaurant industries that could present barriers to implementation.
Some of these barriers are evident in surveys asking food providers such as chefs
about the items they serve. In one survey, chefs reported the portions served
were primarily influenced by the presentation of foods, food cost, and customer
expectations [30]. Although most chefs thought that the amount of food they
serve influences how much patrons consume, their opinions were mixed about
whether it is their responsibility or the customer’s to eat an appropriate amount
when served a large portion. In another survey, nearly all chefs thought that calo-
ries in menu items could be reduced by 10–25% without customers noticing,
but they noted low consumer demand as the primary barrier to making such
changes [31]. Clearly, innovative marketing strategies that increase the appeal of
healthier options are needed – for both consumers and food providers.
The opinions of chefs agree with lab-based studies showing that both the
energy content and portion size of foods can be decreased significantly without
2,000
1,500
1,000
500
0
100 85 75
Entrée energy density (%)
Fig. 2. Reducing the energy density of entrées served at main meals decreased daily
energy intake. When puréed vegetables were incorporated to reduce the energy density
of the standard entrées (100%) to 85 and 75%, daily energy intake of 41 adults decreased
by 202 (6%) and 357 kcal (11%), respectively. Daily vegetable intake increased by 129 g
(50%) in the 85% condition and by 217 g (80%) in the 75% condition. Reprinted from Blatt
et al. [32].
people noticing and that these changes will help to moderate energy intake [20].
Energy density can be reduced in a variety of ways such as the addition of veg-
etables or fruits to recipes or by lowering the fat or sugar content. Herbs and
spices can be used to enhance flavors and mask changes in foods. A particularly
effective strategy to reduce energy density covertly is to add puréed vegetables.
When the energy density of the main dishes served over a day was reduced by
the addition of puréed vegetables, both adults and preschool children consumed
significantly fewer calories. Large amounts of vegetables were added without
affecting the palatability of the foods, even in people who showed a low pref-
erence for vegetables (see fig. 2) [32, 33]. Such innovative strategies to lower
dietary energy density need to be applied to the development of a range of foods
that are palatable, affordable, and readily available.
Changing the food environment in order to have sustained effects on energy
intake will be challenging. However, the effects of portion size and energy den-
sity on energy intake are robust and should be utilized to develop dietary strate-
gies for weight management and to promote a healthier eating environment.
Acknowledgements
44 Rolls
Disclosure Statement
Dr. Rolls has a licensing agreement with Jenny Craig, Inc. for the use of the Volumetrics
trademark.
References
1 Blundell J: Making claims: functional foods 11 Kral TVE, Rolls BJ: Portion size and the obe-
for managing appetite and weight. Nat Rev sity epidemic; in Cawley J (ed): The Oxford
Endocrinol 2010;6:53–56. Handbook of the Social Science of Obesity
2 Rolls BJ: Plenary lecture 1: dietary strategies – The Causes and Correlates of Diet, Physical
for the prevention and treatment of obesity. Activity and Obesity. Oxford, Oxford
Proc Nutr Soc 2010;69:70–79. University Press, 2011, pp 367–384.
3 Foreyt J, Salas-Salvado J, Caballero B, et al: 12 Rolls BJ, Roe LS, Meengs JS: The effect of large
Weight-reducing diets: are there any differ- portion sizes on energy intake is sustained for
ences? Nutr Rev 2009;67:S99–S101. 11 days. Obesity 2007;15:1535–1543.
4 Sacks FM, Bray GA, Carey VJ, et al: 13 Rolls BJ, Roe LS, Meengs JS: Portion size can
Comparison of weight-loss diets with differ- be used strategically to increase vegetable
ent compositions of fat, protein, and carbo- consumption in adults. Am J Clin Nutr 2010;
hydrates. N Engl J Med 2009;360:859–873. 91:913–922.
5 US Department of Agriculture and US 14 Spill MK, Birch LL, Roe LS, Rolls BJ: Eating
Department of Health and Human Services. vegetables first: the use of portion size to
Dietary Guidelines for Americans, ed 7. increase vegetable intake in preschool chil-
Washington, US Government Printing dren. Am J Clin Nutr 2010;91:1237–1243.
Office, 2010. 15 Heymsfield SB: Meal replacements and energy
6 Wanders AJ, van den Borne JJ, de Graaf C, balance. Physiol Behav 2010;100:90–94.
et al: Effects of dietary fibre on subjective 16 Hannum SM, Carson LA, Evans EM, et al:
appetite, energy intake and body weight: a Use of packaged entrees as part of a weight-
systematic review of randomized controlled loss diet in overweight men: an 8-week ran-
trials. Obes Rev 2011;12:724–739. domized clinical trial. Diabetes Obes Metab
7 Gilbert JA, Bendsen NT, Tremblay A, et al: 2006;8:146–155.
Effect of proteins from different sources on 17 Rock CL, Flatt SW, Sherwood NE, et al:
body composition. Nutr Metab Cardiovasc Effect of a free prepared meal and incentiv-
Dis 2011;21(Suppl 2):B16–B31. ized weight loss program on weight loss
8 de Graaf C: Trustworthy satiety claims are and weight loss maintenance in obese and
good for science and society. Response to overweight women: a randomized controlled
DA Booth and A Nouwen [Satiety. No way to trial. JAMA 2010;304:1803–1810.
slim. Appetite 55 (2010) 718–721]. Appetite, 18 Blatt AD, Williams RA, Roe LS, Rolls BJ:
DOI:10.1016/j. appet.2011.05.312. Effects of energy content and energy density
9 Blundell J, de Graaf C, Hulshof T, et al: of pre-portioned entrées on energy intake.
Appetite control: methodological aspects of Obesity 2012, DOI:10.1038/oby.2011.391.
the evaluation of foods. Obes Rev 2010;11: 19 Rolls BJ: The relationship between dietary
251–270. energy density and energy intake. Physiol
10 Rolls BJ, Leahy KE: Reductions in dietary Behav 2009;97:609–615.
energy density to moderate children’s energy 20 Rolls BJ, Roe LS, Meengs JS: Reductions in
intake; in Dube L, Bechara A, Dagher A, et portion size and energy density of foods are
al. (eds): Obesity Prevention: The Role of additive and lead to sustained decreases in
Society and Brain on Individual Behavior. energy intake. Am J Clin Nutr 2006;83:11–17.
Elsevier, Burlington, 2010, pp 543–554.
Discussion
Dr. Ard: Dr. Rolls, you know I am on your side on this. Do you think we shot ourselves
in the foot with the diet crazes of the 1980s by going so low fat and being so restrictive
that it’s going to be really hard to reverse that messaging to get people to understand that
they actually need to eat more in order to be successful with managing their weight?
Dr. Rolls: Unless we can figure out ways to make the low energy density, reasonable
portion approach more accessible and affordable to become almost the default, I think
we are going to have a very hard time changing the way people are eating now. For a
while, we have had few popular fad diets, but now a couple of best-selling diet books are
back to suggesting that changing the proportions of macronutrients or cutting out whole
food groups will make weight loss easy. We need to figure out how to make healthy
weight management easier and appealing.
Dr. Finegood: Why is it that food companies such as those that sell cereal advertise
the heavily sugar-sweetened, less nutritious cereals but don’t advertise healthier ones?
There are data to suggest that in a blinded test kids don’t necessarily prefer cereal higher
in sugar. Is it that the profit margins are much higher or is it because they don’t think the
lower sugar product will sell well?
Dr. Rolls: I cannot speak for the industry, but one possibility is that they think that
they are more likely to get kids to eat cereal in the first place and that kids who eat any
46 Rolls
kind of cereal, even if it’s sugar sweetened, have a better diet quality than those eating no
breakfast or some other less nutritious option.
Dr. Finegood: But we can’t disentangle that from the advertising efforts.
Dr. Rolls: We know that we have to be marketing the healthy foods more effectively.
That is why the recent TV ads suggesting that parents have to be secretive about vegetables
in foods are so interesting – kids see the message that vegetables are so bad they shouldn’t
be told about them. Product developers know how to put more vegetables in foods without
affecting taste, but marketers have not figured out how to get consumers to buy them. By
the way, we have found that ‘stealth’ vegetables in foods are an effective way to reduce energy
density and to get kids to eat their vegetables. We recently published a study showing that
we could add large amounts of vegetables to a variety of foods without affecting the
palatability. Our problem with the stealth vegetables was with the baked goods – the
children liked them better with more vegetables, probably because they were moister. By
adding pureed vegetables to main dishes across a day, we got the preschoolers meeting their
daily vegetable requirements [1]. If people like vegetables anyway, you don’t have to go
stealth, but why not try some more innovative ways to get vegetables into the day?
Dr. Barclay: The food industry is looking for ways to become increasingly part of the
solution rather than only being considered part of the problem; however, we seem to be
getting some mixed messages from the science. There is a recent paper from Duffey and
Popkin [2] looking at the relative importance of portion size, eating frequency, and
energy density. They found that energy density explained less of the increase in energy
intake over recent decades in US adults compared to portion size and eating frequency.
Dr. Rolls: In that analysis of reported intakes in a population-based data set, a novel
algorithm was used to compare the relative importance of those factors. We plan to
revisit this issue using more conventional methods of analysis. In our controlled studies,
energy density has more robust effects than portion size, but portion size also has
significant effects on energy intake. It is clear that the pattern of eating and properties of
foods such as energy density and portion size can affect intake, and the food industry
should be applying this knowledge to product development.
Dr. Goran: I completely buy the energy density argument, but I was wondering if
there is more research on the factors that drive choice at the moment of consumption.
What are the contextual factors or the intrinsic biological factors that we know about or
need to know about that drive individual choice at the moment of consumption?
Dr. Rolls: There is a large body of information on the determination of an individual’s
food choices that we don’t have the time to cover. This brings us back to the issue of
behavioral phenotypes and personalized approaches. Choice would be something that I
would want to put into any weight management plan so up front we can figure out what
kinds of foods people are going to be able to eat in a sustainable fashion. I dream of a
computer program where people’s food preferences are screened before they go into a
program. You would then use that information to direct them towards a plan that will
accommodate those choices. In the end, maintenance is the issue and that depends on
liking the food.
Dr. Drewnowski: Designing food patterns based on linear programming in people’s
existing choices based on 7 days of records is being done by my colleagues in Marseille,
France. So, this is something that’s absolutely on its way.
Dr. Rosenbaum: I was wondering if people can learn to respond visually to portion
size. If all foods were packaged so that people could see how much food they would get
References
1 Spill MK, Birch LL, Roe LS, Rolls BJ: Hiding 2 Duffey KJ, Popkin BM: Energy density,
vegetables to reduce energy density: an effec- portion size, and eating occasions: con-
tive strategy to increase children’s vegetable tributions to increased energy intake in
intake and reduce energy intake. Am J Clin the United States, 1977–2006. PLoS Med
Nutr 2011;94:735–741. 2011;8:e1001050.
48 Rolls
Obesity Treatment: Challenges and Opportunities
Drewnowski A, Rolls BJ (eds): Obesity Treatment and Prevention: New Directions.
Nestlé Nutr Inst Workshop Ser, vol 73, pp 49–60,
Nestec Ltd., Vevey/S. Karger AG., Basel, © 2012
Abstract
The prevalence of obesity has increased throughout the last three decades due to genetic,
metabolic, behavioral, and environmental factors [1]. Obesity in turn increases risk for a
number of metabolic diseases including type 2 diabetes, cardiovascular disease, fatty liver
disease and some forms of cancer [1]. Despite the well-known link between obesity and
increased morbidity, the mechanism of this remains elusive. Thus, the question ‘why does
increased body fat cause increased metabolic comorbidities’ remains unanswered. By
understanding the underlying basis of obesity-associated metabolic diseases, different
therapies could be designed to target relevant pathways. Although we lack a full under-
standing of the underlying mechanisms that result in disease, several putative explana-
tions exist for why fat affects metabolic health. One such theory is based on the anatomic
location of fat deposition and ectopic fat accumulation [2]. Specifically, current literature
suggests that visceral, liver and skeletal fat accumulation affects organ function and con-
tributes to the development of insulin resistance, fatty liver, and the metabolic syndrome
[3]. However, even in individuals matched for body fat and fat distribution, significant dif-
ferences can exist in metabolic outcomes, and the phenomenon of metabolically healthy
obese has been well described [4]. More recent data suggest the alternative hypothesis
relating excess adipose tissue to disease risk based on the metabolic function and mor-
phological properties of adipose tissue. In this scenario, excess adipose tissue is hypothe-
sized to contribute to a state of chronic inflammation which promotes development of
insulin resistance as well as other metabolic complications by stimulating nuclear factor-κB
and Jun N-terminal kinase pathways in adipocytes and the liver [5]. In this paper, we will
review the hypothesis linking excess adipose tissue to increased disease risk through adi-
pose tissue inflammation. Copyright © 2012 Nestec Ltd., Vevey/S. Karger AG, Basel
Adipose Tissue as an Endocrine Organ
It was once believed that adipocytes were only involved in the storage of trig-
lycerides, but recent studies have demonstrated that they also act as endocrine
organs. Hotamisligil et al. [6] and Karasik’s group first showed that proinflam-
matory cytokine tumor necrosis factor (TNF)-α was produced by adipocytes,
induced insulin resistance, and increased with expanding fat volume. The con-
cept of adipose tissue as a site for the production of cytokines and other sub-
stances has expanded to include leptin, interleukin (IL)-6, resistin, monocyte
chemoattractant protein-1 (MCP-1), plasminogen activator inhibitor-1 (PAI-
1), angiotensinogen, visfatin, retinol-binding protein-4, and serum amyloid
A. Adiponectin is also produced by adipocytes, but its release decreases with
increasing adiposity. Leptin and adiponectin are adipokines that are produced
only by adipocytes; while TNF- α, IL-6, MCP-1, visfatin, and PAI-1 are also
expressed at high levels in activated macrophages. TNF-α, IL-6, resistin, and
other proinflammatory cytokines participate in the induction and maintenance
of the acute inflammatory response associated with obesity. Additionally, MCP-1
and other chemokines recruit macrophages to adipose tissue. These cytokines
and chemokines activate intracellular pathways that promote the development
of insulin resistance, type 2 diabetes and other metabolic complications associ-
ated with obesity [5]. Therefore, adipose tissue and macrophages within adipose
tissue have been shown to play important roles in the regulation of metabolic
pathways through the excretion of adipokines and cytokines.
Despite the observed link between body fat, non-alcoholic fatty liver disease,
and type 2 diabetes, some individuals exhibit ‘metabolically benign obesity’ and
are protected from the metabolic consequences of excess adiposity, possibly due
to differences in adipocyte tissue metabolism and macrophage infiltration. In a
study aimed at identifying insulin-resistant individuals, 17% of the overweight
and obese participants were found to be insulin sensitive [7]. Additionally, a
review by Karelis et al. [4] determined that approximately 20% of the general
population can be categorized as obese but metabolically healthy. In contrast to
this, 18% of the population were found to have a normal bodyweight but suf-
fered from severe metabolic abnormalities. In other studies among obese adults,
the degree of adipose tissue inflammation was closely associated with increased
metabolic risk for type 2 diabetes, cardiovascular disease and fatty liver disease,
whereas obese adults without adipose tissue inflammation have metabolic risk
factors in the healthy range [8, 9]. Despite the fact that most human research
has focused on the links between fat distribution and metabolic risk, these find-
ings suggest a role for adipose tissue inflammation. In a previous study among
50 Goran · Alderete
obese young minority adults, we found that approximately 40% of subjects had
subcutaneous abdominal adipose tissue with crown-like structures, indicating
inflammation, whereas approximately 60% of subjects had no signs of adipose
tissue inflammation. Despite the two groups being identical for overall obe-
sity and subcutaneous abdominal adipose tissue volume, those with inflamed
adipose tissue had approximately 30% greater visceral adipose tissue and 41%
greater liver fat; 53% greater fasting insulin and 23% lower β-cell function, and
22% higher TNF-α. Additionally, adipose tissue from those with inflamed adi-
pose tissue had upregulated nuclear factor-κB (NF-κB) expression activity and
downregulation of insulin signaling [9]. Another recent study by Bremer et al.
[8] demonstrated that, among overweight and obese adults, those with meta-
bolic syndrome have significantly higher levels of infiltrating macrophages/
crown-like structures in their adipose tissue compared to those without the
metabolic syndrome. Given these observations, the disparities in metabolic
diseases among obese individuals may be explained by the degree of chronic
low-grade inflammation of adipose tissue. Therefore, targeting adipose tissue
inflammation has become an important new strategy in treating the metabolic
conditions typically associated with obesity.
One of the few effective anti-inflammatory treatments for these metabolic dis-
eases is weight loss [10]. Studies in diet-induced obese mice have shown that
reductions in adiposity result in decreases in macrophage infiltration of adi-
pose tissue as well as gene expression of pro-inflammatory markers [11, 12].
Specifically, Kosteli et al. [11] found that the murine immune response to
weight loss was dynamic. Caloric restriction of high-fat diet-fed mice resulted
in an initial increase in adipose tissue macrophages; however, the number of
adipose tissue macrophages decreased following an extended period of weight
loss. Vieira et al. [12] examined the effects of diet and exercise on inflamma-
tion among high-fat diet-induced obese mice. This study examined the effects
of weight loss via a low-fat diet, exercise training, or a combination of low-fat
diet and exercise on inflammation. All methods of weight loss resulted in a sig-
nificant attenuation of high-fat diet-induced increases in systemic and adipose
tissue inflammation. Additionally, all three interventions improved insulin sen-
sitivity, reduced adiposity, MCP-1, and TNF-α gene expression. Among obese
humans, caloric restriction to achieve weight loss also decreases markers of
inflammation. For example, a 28-day severe calorie-restrictive diet (800 kcal/
day) among obese females was found to reduce bodyweight by an average of
13 pounds as well as significantly alter markers of inflammation. Specifically,
weight loss resulted in a decreased expression of proinflammatory markers (e.g.
IL-12a, matrix metalloproteinase-9) in white adipose tissue and increased the
Diets high in sugar and fat have been shown to increase systemic mark-
ers of inflammation [14, 15]. Given the link between diet and inflammation,
it is not surprising that dietary alterations have been used to examine obesity
and obesity-associated inflammation among obese humans [12]. Specifically,
healthy adult males given fructose (40 or 80 g/day) or glucose (40 or 80 g/day)
sugar-sweetened beverages for 2 weeks demonstrated significant increases in
high-sensitivity CRP [14]. In addition to sugar intake, trans fat consumption
is related to markers of inflammation. A cross-sectional analysis of 730 women
from the Nurse’s Health Study found that CRP levels were 73% higher and IL-6
levels were 17% higher among those in the highest quintile of trans fat intake
compared to those in the lowest [15]. Among humans, a diet high in fiber (30
g/day), either through diet or fiber supplementation, has been shown to signifi-
cantly reduce CRP levels [16]. Baseline examination of data from 406 partici-
pants from the Finnish Diabetes Prevention Study found that increases in fiber
predicted decreases in CRP and IL-6. They also found that changes in fat and
carbohydrate intake were either weakly or not related to reductions in CRP and
IL-6 [17].
Studies using obese mice have found differing effects of diet composition
on inflammation. For example, Wang et al. [18] used mice with high-fat diet-
induced obesity to examine the effects of weight loss achieved by switching from
high-fat diet to: (1) an ad libitum low-fat normal diet or (2) restricting the high-
fat diet intake to match bodyweight of mice with low-fat normal-diet-induced
weight loss. Weight loss by either of the two methods resulted in decreased
fat mass and liver steatosis; however, effects were greatest among the low-fat
normal-diet-induced weight loss than the high-fat diet restriction-induced
weight loss. Interestingly, weight loss with the low-fat normal diet, but not the
restricted high-fat diet, normalized blood CDC11c+ monocytes and attenu-
ated hepatic inflammation. In contrast, the calorie-restricted high-fat diet sig-
nificantly reduced chemokine levels and CDC11c+ cells in adipose tissue when
compared to low-fat diet-induced weight loss and obese controls. Although
52 Goran · Alderete
these studies demonstrate that changes in diet can affect systemic inflammation
associated with obesity, the direct mechanism and critical dietary components
are not known. Additionally, the dietary changes necessary to elicit weight loss
and decreases in inflammation are likely to be too drastic to sustain over time.
Anti-Inflammatory Treatments
54 Goran · Alderete
control improved in all treatment groups. It is interesting to note that mild
hypoglycemic events occurred in 22% of the 3 g/day, 30% of the 3.5 g/day, and
22% in the 4 g/day groups [29].
Overall, these trials noted improvements in insulin sensitivity, fasting glucose,
CRP, and NF-κB activity with a 2- and 4-week high-dose salsalate treatment of
4.0 and 4.5 g/day [27, 28, 30]. These data support the hypothesis that utilizing
a non-steroidal anti-inflammatory drug, such as salsalate, to target adipose tis-
sue inflammation may provide a therapeutic route for treating obesity-related
diseases. Coupled with the above-mentioned findings, and due to the fact that
non-acetylated salicylates do not prolong bleeding times, salsalate may offer a
relatively safe and effective treatment for the low-grade inflammation associ-
ated with obesity. However, it is important to note the limitations of salsalate
treatment. Specifically, the moderately high dose needed to elicit improvements
in inflammation has potential side effects that warrant concern. Participants in
these trials experienced ringing in the ears, alterations in liver function tests, as
well as hypoglycemia. These adverse effects highlight the need to understand
the mechanism in which salsalate targets inflammation and improves metabolic
indices.
Previous studies in obese individuals show that salsalate decreases plasma
markers of inflammation and improves glucose control under conditions of
weight stability. However, the mechanism of these effects is not known, and
no prior clinical study has examined whether the improvement of metabolic
complications is due to reduction in adipose tissue inflammation. Currently,
our group is initiating a double-blind, randomized control trial, among obese
Hispanic young adults, aimed at determining the effects of 4 weeks of salsalate
treatment (4 g/day) on the number of macrophages in adipose tissue. This study
will be the first to examine the effects of salsalate therapy, without weight loss,
on subcutaneous adipose tissue inflammation. In particular, we will examine
the notion that the improvement of metabolic risk after salsalate intervention
occurs due to its effect on suppressing adipose tissue inflammation, and that
without an improvement in adipose tissue inflammation there will be limited
improvement in other metabolic risk factors. Findings from this study have the
potential to elucidate the mechanism in which salsalate improves glucose con-
trol and decreases inflammation. Once the mechanism(s) is identified, safer and
more effective therapies could be designed to target inflammation, and thereby
treat the metabolic complications, associated with obesity.
Conclusions
Given the known link between chronic low-grade inflammation and metabolic
health, it is becoming increasingly important to understand the biological pro-
cesses that contribute to inflammation in adipose tissue. We have reviewed the
Disclosure Statement
The authors declare that no financial or other conflict of interest exists in relation to the
content of the chapter.
References
1 Wang Y, McPherson K, Marsh T, et al: Health 5 Shoelson S, Herrero L, Naaz A: Obesity,
and economic burden of the projected obe- inflammation, and insulin resistance.
sity trends in the USA and the UK. Lancet Gastroenterology 2007;132:1–12.
2011;378:815–825. 6 Hotamisligil G, Shargill N, Spiegelman B:
2 Mittendorfer B: Origins of metabolic com- Adipose expression of tumor necrosis factor-
plications in obesity: adipose tissue and free alpha: direct role in obesity-linked insulin
fatty acid trafficking. Curr Opin Clin Nutr resistance. Science 1993;259:87–91.
Metab Care 2011;14:535–541. 7 McLaughlin T, Abbasi F, Cheal K, et al: Use
3 Rasouli N, Molavi B, Elbein S, Kern PA: of metabolic markers to identify overweight
Ectopic fat accumulation and metabolic syn- individuals who are insulin resistant. Ann
drome. Diabetes Obes Metab 2007;9:1–10. Intern Med 2003;139:802–809.
4 Karelis A, St-Pierre D, Conus F, et al: 8 Bremer A, Devaraj S, Afify A, Jialal I:
Metabolic and body composition factors in Adipose tissue dysregulation in patients
subgroups of obesity: what do we know? J with metabolic syndrome. J Clin Endocrinol
Clin Endocrinol Metab 2004;89:2569–2575. Metab 2011;96:E1782–E1788.
56 Goran · Alderete
9 Le K, Mahurakar S, Alderete T, et al: 19 Geelen A, Brouwer I, Schouten E, et al:
Subcutaneous adipose tissue macrophage Intake of n-3 fatty acids from fish does not
infiltration is associated with hepatic and lower serum concentrations of C-reactive
visceral fat deposition, hyperinsulinemia, protein in healthy subjects. Eur J Clin Nutr
and stimulation of NF-κB stress pathway. 2004;58:1440–1442.
Diabetes 2011;60:2802–2809. 20 Hundal R, Petersen K, Mayerson A, et al:
10 Aron-Wisnewsky J, Tordjman J, Poitou C: Mechanism by which high-dose aspirin
Human adipose tissue macrophages: m1 and improves glucose metabolism in type 2 dia-
m2 cell surface markers in subcutaneous and betes. J Clin Invest 2002;109:1321–1326.
omental depots and after weight loss. J Clin 21 Hellmann J, Tang Y, Kosuri M, et al: Resolvin
Endocrinol Metab 2009;94:4619–4623. D1 decreases adipose tissue macrophage
11 Kosteli A, Sugaru E, Haemmerle G, et al: accumulation and improves insulin sensitiv-
Weight loss and lipolysis promote a dynamic ity in obese-diabetic mice. FASEB J 2011;
immune response in murine adipose tissue. J 25:2399–2407.
Clin Invest 2010;120:3466–3479. 22 Todoric J, Löffler M, Huber J, et al: Adipose
12 Vieira V, Valentine R, Wilund K: Effects of tissue inflammation induced by high-fat diet
exercise and low-fat diet on adipose tissue in obese diabetic mice is prevented by n-3
inflammation and metabolic complications polyunsaturated fatty acids. Diabetologia
in obese mice. Am J Physiol Endocrinol 2006;49:2109–2119.
Metab 2009;296:E1164–E1171. 23 Shoelson S, Lee J, Goldfine A: Inflammation
13 Clément K, Viguerie N, Poitou C, et al: and insulin resistance. J Clin Invest 2006;116:
Weight loss regulates inflammation-related 1793–1801.
genes in white adipose tissue of obese sub- 24 Ogston N, Karastergiou K, Hosseinzadeh-
jects. FASEB J 2004;18:1657–1669. Attar M, et al: Low-dose acetylsalicylic acid
14 Aeberli I, Gerber P, Hochuli M: Low to mod- inhibits the secretion of interleukin-6 from
erate sugar-sweetened beverage consumption white adipose tissue. Int J Obes (Lond) 2008;
impairs glucose and lipid metabolism and 32:1807–1815.
promotes inflammation in healthy young 25 Abe M, Matsuda M, Kobayashi H: Effects of
men: a randomized controlled trial. Am J statins on adipose tissue inflammation: their
Clin Nutr 2011;94:479–485. inhibitory effect on MyD88-independent
15 Lopez-Garcia E, Schulze M, Meigs J, et al: IRF3/IFN-beta pathway in macrophages.
Consumption of trans fatty acids is related Arterioscler Thromb Vasc Biol 2008;28:
to plasma biomarkers of inflammation 871–877.
and endothelial dysfunction. J Nutr 2005; 26 Kopp E, Ghosh S: Inhibition of NF-kappa
135:562–566. B by sodium salicylate and aspirin. Science
16 King D, Egan B, Woolson R, et al: Effect of a 1994;265:956–959.
high-fiber diet vs a fiber-supplemented diet 27 Fleischman A, Shoelson S, Bernier R, et al:
on C-reactive protein level. Arch Intern Med Salsalate improves glycemia and inflam-
2007;167:502. matory parameters in obese young adults.
17 Herder C, Peltonen M, Koenig W, et al, Diabetes Care 2008;31:289–294.
Finnish Diabetes Prevention Study Group: 28 Koska J, Ortega E, Bunt J, et al: The effect of
Anti-inflammatory effect of lifestyle changes salsalate on insulin action and glucose toler-
in the Finnish Diabetes Prevention Study. ance in obese non-diabetic patients: results
Diabetologia 2009;52:433–442. of a randomised double-blind placebo-con-
18 Wang Q, Perrard X, Perrard J, et al: trolled study. Diabetologia 2009;52:385–393.
Differential effect of weight loss with low-fat 29 Goldfine A, Fonseca V, Jablonski K, et al:
diet or high-fat diet restriction on inflamma- The effects of salsalate on glycemic control in
tion in the liver and adipose tissue of mice patients with type 2 diabetes: a randomized
with diet-induced obesity. Atherosclerosis trial. Ann Intern Med 2010;152:346–357.
2011;219:100–108. 30 Goldfine A, Silver R, Aldhahi W, et al: Use of
salsalate to target inflammation in the treat-
ment of insulin resistance and type 2 diabe-
tes. Clin Transl Sci 2008;1:36–43.
Dr. Lovejoy: I question the accuracy of the term ‘healthy obesity’. If we take a
diabetes-centric view, it might be possible, as you have shown here, to talk about
individuals who are metabolically healthy versus those who aren’t. But when we look
broadly, a big part of the public health burden of obesity are things like osteoarthritis
and sleep apnea which require weight loss to reverse. So, there are other consequences of
obesity in addition to metabolic ones. Furthermore, recent studies suggest that even
‘metabolically healthy obesity’ may be a temporary condition, as these obese individuals
develop metabolic abnormalities over time at a higher rate than non-obese [1, 2]. The
risk of using the term is the backlash we saw a few years ago when there was a lot of
media attention around the notion of ‘healthy obesity’ that led many people to
erroneously conclude that obesity is not a problem.
Dr. Goran: Good points, thank you for highlighting that. Most of what I have talked
about relates to the metabolic dysfunction of obesity. The more correct term to use
would be ‘metabolically healthy obesity’. The other aspects that you mention probably
fall outside of the metabolic consequences of obesity.
Dr. Oppert: I have two quick questions. First, could it be that inflammation at the
beginning is something beneficial or a normal reaction to enlarge the adipocytes?
Second, could it be that there is a point of no return? For example, when there is
inflammation and then fibrosis in the liver? Perhaps strategies to reduce inflammation
would not be so beneficial at the beginning, and perhaps could be difficult to put in
place when it’s fibrosis.
Dr. Goran: I think that’s a potentially very interesting point, and we don’t know a lot
about the time course of when the process is established relative to the accumulation of
body fat. It’s unclear whether this is an intrinsic property of adipose tissue or that it’s a
normal reaction to adipocytes growth. I think those are questions that remain.
Dr. Rosenbaum: At the beginning, you are showing that adipocytes attract
macrophages or macrophages come to adipocytes. I was wondering if you go back and
look at the healthy versus the unhealthy obese, do you think the major difference is in
whatever the adipocytes put out there to attract the macrophages, or in the macrophages
themselves and in their aggressiveness in pursuing the adipocytes. Where is the original
difference between those groups?
Dr. Goran: I don’t know who is calling whom, I think that’s a good question as to
who is attracting whom. I don’t think we really know the answer to that.
Dr. Finegood: Is it not about the death of the cells themselves?
Dr. Goran: It is – but then what causes the death of the cells?
Dr. Finegood: Yes, but the point is that the macrophages are called to wall off the
triglyceride that is sitting around presumably because the cells had died. What causes
the difference in the death of the cells I think is important.
Dr. Ochoa: A recent paper showed that M2 responses prevented the insulin resistance
and the hyperglycemia in mice from M1 responses. Do you know if when we do
interventions, we are promoting M2 responses instead of just quieting down the whole
system? Has anyone studied that?
Dr. Goran: I showed the surgical weight loss study where the M1 and M2 responses
were different.
58 Goran · Alderete
Dr. Ochoa: It’s interesting that any surgical intervention promotes M2 responses
dramatically, and those in severe traumas may last for a month or more; so, it’s really
fascinating that we are seeing that M2 response predominate in our patients.
Dr. Jakicic: When you look at this cascade and when you throw activity or fitness
into this mix, what makes this pathway healthy? Can you really have a healthier obese
person if you put activity into the mix? Is it necessary? And I guess you are thinking
about at least doing some of that.
Dr. Goran: I wasn’t.
Dr. Jakicic: Are you sure? Have you thought about the issue when you talk about the
GLP-1 and the blocking of the insulin that obesity may be blocking that if you are
increasing insulin sensitivity on the other side with activity? Where does activity fit into
this cascade?
Dr. Goran: We have published a few studies showing that exercise, or strength
training can have variable effects on insulin and circulating cytokines for example, and
there are some other studies like that, but I don’t really know what the mechanism is. It’s
an interesting point, and I should probably think about it a bit more. Mostly, I have been
disappointed with our studies with exercise as we have found mixed results. This was not
just in different people but also in different studies that were done in the same way. We
found one effect versus another, so our most consistent observation is that we can’t get a
consistent effect of exercise.
Dr. Drewnowski: I have a question about sugar consumption. The consumption of
added sugar in its many forms, whether solid or liquid, is often associated with lower
incomes. Are there any racial or ethnic groups for whom this combination is particularly
devastating? Can you answer that based on your research? I am saying the consumption
of sugar varies as a function of socioecomic status and is higher among lower income
people than among the upper income people. Are there some low-income groups for
whom sugar-fat mixtures would be particularly damaging?
Dr. Goran: Yes, in fact we have a paper in our Hispanic cohort, and this is a very
specific example but it answers your question. There is some evidence to suggest that
Hispanics have an increased liking for sugars. They have an increased prevalence of the
PNPLA3 gene, though I don’t know why. The prevalence of this gene is almost 50% in
Hispanics versus 20% in Caucasians. This gene promotes fatty liver disease. The substrate
for fatty liver disease is sugar, possibly fructose, because fructose is lipogenic and
promotes lipogenesis in the liver. So, I think this is a perfect storm among Hispanics who
have a genetic predisposition and a liking for sugar. The issue of economics comes into
the equation as well. If you consider the whole story of high fructose corn syrup, you see
that the exports of corn sweeteners by the US are increasing dramatically as a result of a
recent policy shift by the World Trade Organization. So, I think this is a very clear
example of a specific subgroup of the population which is susceptible to the damaging
effects of sugar.
Dr. Ard: How much of this has to do with energy balance in the context of intake of
sugar or other things? Adipocyte biology can change with a bout of exercise or with the
institution of calorie restriction or even bringing people into energy balance; so, how
much of this do you think is related to that milieu of excess caloric intake that then
continues to foster the inflammation?
Dr. Goran: There are limited data to answer this question, especially in humans
which would require a fat biopsy. There are some animal studies that have looked at the
References
1 Chang Y, Ryu S, Suh BS, et al: Impact of BMI 2 Bradshaw PT, Monda KL, Stevens J:
on the incidence of metabolic abnormalities Metabolic syndrome in healthy obese,
in metabolically healthy men. Int J Obes overweight and normal weight individuals:
(Lond) DOI: 10.1038/ijo.2011.247. The Atherosclerosis Risk in Communities
Study. Obesity (Silver Spring) DOI: 10.1038/
oby.2012.173.
60 Goran · Alderete
Drewnowski A, Rolls BJ (eds): Obesity Treatment and Prevention: New Directions.
Nestlé Nutr Inst Workshop Ser, vol 73, pp 61–66,
Nestec Ltd., Vevey/S. Karger AG., Basel, © 2012
Dr. Lovejoy: I would like to ask a controversial question that ties to what most of the
speakers have talked about. Are there specific foods, for example those high in sugar or
fat, that have effects on the brain or body that are parallel to those of addictive drugs? If
that is the case, do we need to consider this in relation to obesity treatment and strategies
for a healthy diet?
Dr. Rolls: I presume you are talking about addiction in a stringent way using a formal
definition that includes withdrawal symptoms.
Dr. Lovejoy: I think that has been a part of the controversy. There are those who
would argue that the stringent definition of addiction applies to food, including
withdrawal. But others disagree and say that it’s not truly addiction with food. But
obviously certain foods are stimulating the same areas in the brain as addictive drugs.
How do we come to grips with that as a field in a way that makes sense to the public?
Dr. Rosenbaum: Some of the brain areas that are more active in individuals following
weight loss and are sensitive to leptin are the same ones that are more active in individuals
who are addictive, so there are certain parallels between addiction and the inability to
sustain weight loss. I don’t know if anyone has looked prospectively at who is going to
gain weight along those lines. Could you identify someone early and say that he is going
to be a food junky down the line? I don’t know the answer.
Dr. Drewnowski: France Bellisle and I published a review entitled ‘Is sweetness
addictive?’ in the British Nutrition Bulletin some years ago. It is worth noting that the
former definition of addiction, as featured in the Diagnostic and Statistical Manual of
Mental Disorders (DSM), is now gone. It used to be based on the criteria of tolerance
and withdrawal. What we have now is something called substance dependence and the
inclusion criteria are very soft. Just about anything fits. Substance use behaviors that are
performed in private, that take precedence over other activities are now part of the
definitions. So, conceivably, eating chocolate would fit the bill. We need to pay attention
to the fact that the definitions of food addiction are changing. Our studies need to keep
pace with the definitions. And then there are very few studies on food addiction or on
the potential underlying mechanisms. Dr. David Kessler in his book, The End of
Overeating, was referring to some studies performed by us at the University of Michigan.
Those studies, conducted with bulimic women used the opiate blocker naloxone to show
that the opioid system was somehow involved in food preferences and the consumption
of good tasting foods. However, that was a far cry from addiction. Now, 15 years later,
there are still no new studies on selective opiate antagonists and their impact on food
consumption. Instead, attention has shifted to dopamine and neuroimaging, but I am
not sure that those studies have come to a consensus on whether or not foods are
addictive. All we know is that something is going on in different areas in the brain
following certain exposures. So, I think the issue of food addiction is still unresolved. I
should note that in all of our research we have been very careful to note that foods were
not truly addictive substances and that liking or even craving was not the same as
addiction. We did not want to use the word addiction; we did not use the word addiction.
Of course, our studies were interpreted in the popular press as a search for a drug that
stops food cravings, which is not what we were all about. So, it’s an interesting issue that
many people raise all the time. There is a big difference between addiction and liking or
preferring or even craving. I don’t think that sugar liking is an addiction.
Dr. Goran: It doesn’t matter what it’s called. If you agree that there are those other
properties that are not healthful, it doesn’t matter whether it’s technically addiction or
not.
Dr. Drewnowski: We are kind of addicted to food since we eat it every day. But that is
not really addiction, there are other factors that you need to very carefully consider as
well.
Dr. Goran: Really, you think we are addicted to food?
Dr. Drewnowski: According to the definition of substance dependence in DSM-IV
that I read very closely, eating food would actually fit.
Dr. Rolls: But really the issue is how we can shift people towards healthier foods and
get them away from the foods that are pushing overconsumption.
Dr. Drewnowski: The economics plays a role too because the foods that you would
identify as targets of cravings and addictions are energy dense, but also they are
inexpensive.
Dr. Rosenbaum: In a rodent study that was just published, Yann Ravussin showed
that if you take mice and overfeed them on the DIL diet, the nice diet that rodents and
lab technicians love, they become obese. If they stay obese long enough, when you
restrict their food, they behave like starving animals and they defend a higher bodyweight.
That’s an analogy to addiction in that somebody who abuses a substance develops a need
for it, and they have a higher set point for how much of that stimulus they need. So, it’s
not a bad model, and perhaps if you sustain an elevated bodyweight long enough or at
specific times in life, your tolerance level in a way that is similar to addiction is
changed.
Dr. Rolls: But an interesting question that comes from your talk is can you ever
reverse that? Can you reduce their weight long enough to get them back to where they
were before the obese phase?
Dr. Rosenbaum: It seems from both the epidemiology and the evolutionary biology
that you can probably jack this set point or whatever you want to call it up, but there is
no way that we know how to jack it down.
Dr. Rolls: Do we really know that?
Dr. Rosenbaum: Animals that are sustained at reduced bodyweight for an extended
period of time remain extraordinarily metabolically efficient, while animals maintained
Abstract
The composition of the gut microbiome is hypothesized to be an environmental factor
that contributes to obesity. Results of several human studies suggest that obesity is asso-
ciated with differences in the gut microbiota composition, reduced bacterial diversity,
and altered representation of bacterial metabolic pathways. The obese phenotype is asso-
ciated with increased microbial fermentation and energy extraction from non-digestible
food components; however, until recently it was not clear how relatively small increases in
energy extraction could contribute to the large and rapid weight gain observed in the
animal studies. Mechanisms by which the gut microbiome may influence metabolism
and energy homeostasis include regulation of energy uptake from diet, interaction with
signaling molecules involved in host metabolism, modification of gut permeability,
release of gut hormones, and low-grade, chronic inflammation, the latter being a hall-
mark of obesity-related diseases. Copyright © 2012 Nestec Ltd., Vevey/S. Karger AG, Basel
Introduction
Results of several human studies, typically with small sample sizes, suggest that
obesity is associated with differences in gut microbial profiles [2, 7–11], as well as
reduced bacterial diversity [9], and altered representation of bacterial metabolic
pathways [9]. It has been hypothesized that, at the phylum level, an increased
ratio of Firmicutes to Bacteroidetes may contribute to the pathophysiology of
human obesity [2, 12]; however, this hypothesis has not been supported consis-
tently across studies [13, 14], even in the studies that detect obesity-associated
differences in gut microbial community [7, 8, 10].
Recently, in a cross-sectional sample of 115 premenopausal women, we
evaluated the association between adiposity and the gut microbial community.
We measured percent body fat on these women by dual-energy X-ray absorp-
tiometry and used several approaches to characterize the microbial commu-
nity in fecal samples. First, using quantitative PCR and terminal restriction
length polymorphism (TRFLP), we found that the abundance of Bacteroidetes
was positively associated with percent adiposity (r = 0.20; p = 0.02) and that
TRFLP peak 473 (indicative of the Bacteroidetes-Prevotella group) was statisti-
cally significantly associated with percent adiposity. Multiple regression testing
of the relationship between adiposity and gut microbial community, adjusting
for energy, carbohydrate, fat and dietary fiber intake, showed that Bacteroidetes
explained 15% of the variation in adiposity in these women. In a subset of
the women, we also compared the gut microbial community composition by
68 Hullar · Lampe
pyrosequencing the V1–V3 region of the16S rRNA gene. In total, we analyzed
239,875 sequences. After trimming the sequences (no ambiguous nucleotides,
homopolymer ≤8, primer mismatch ≤2, barcode mismatch ≤1, quality score
≥35 in 50 nt window), we aligned the sequences in SILVA (http://www.arb-silva.
de) and found that out of 65,492 sequences 13,190 were unique. We found that
the bacterial communities in the obese women (35–46% body fat; n = 35) were
significantly less diverse than those of the women with average percent adipos-
ity (25–32%; n = 27), a trend similar to that reported by Turnbaugh et al. [9] in a
sample of twins. The phylogenetic composition and abundance of the microbial
community were also significantly different between lean and obese individu-
als. Multivariate analysis of the pyrosequencing data, using non-metric multi-
dimensional scaling (NMS), explained 75% variation in our data and NMS axis
2 was positively associated with percent adiposity. In addition, two Prevotella
sp in the phylum Bacteroidetes were positively associated with NMS axis 2 and
increasing adiposity. Given that the Bacteroidetes are a metabolically diverse
group specializing in saccharolytic degradation, this suggests that increased
efficiency of energy extraction from the diet may be associated with specific
microbial metabolic pathways in a few groups of bacteria in obese individuals.
Studies in experimental animal models also support the hypothesis that gut
microbes play an important role in energy regulation and adiposity. Colonization
of axenic (i.e. ‘germ-free’) mice with a gut microbiota derived from conventional
mice resulted in a 60% increase in body fat mass and development of insulin
resistance [15], and the bacterial community composition of the inoculum
influenced amount of fat stored [16]. These effects have been seen in the context
of a chow diet and a semi-synthetic Western diet [17]; however, in a similar
experiment conducted using a high-fat, semi-synthetic diet with the same over-
all proportions of macronutrients as the Western diet, but composed of different
ingredients, axenic mice were not protected from obesity and gained as high or
higher amounts of body fat as the conventional mice [18]. Bacterial community
composition has also been shown to differ by degree of genetically mediated
adiposity; obese ob/ob (i.e. leptin-knockout) mice had 50% lower abundance of
Bacteroidetes and proportionally higher amounts of Firmicutes [1].
The results of the association studies in humans and the animal studies raise
the question whether compositional change in the gut microbiota precedes the
onset of weight gain and whether the microbiota plays a causative role [19].
Unfortunately, no prospective studies have been conducted in humans to exam-
ine this. In a retrospective study, Kalliomäki et al. [20] selected overweight and
obese (n = 25) and normal-weight (n = 24) Finnish children, 7 years of age, and
matching for numerous factors (e.g. gestational age, body mass index, BMI, at
birth, etc.), and assessed gut microbial community by fluorescent in situ hybrid-
ization (FISH) and qRT-PCR in fecal samples collected at 6 and 12 months. They
reported that, in infancy, bifidobacterial numbers were higher in normal-weight
children, and Staphylococcus aureus was greater in overweight children. Some
70 Hullar · Lampe
association and interpretation of the results. DNA extraction [23], PCR primer
biases [24, 25], and accuracy of phylogenetic identification [26] are potential
sources of variation. For example, oligonucleotide probes, used in FISH to iden-
tify shifts in microbiota associated with weight loss [10, 20], are hampered by
cell permeability or probe mismatch issues [27]. In addition, on the adiposity
side of the equation, relying on BMI rather than procedures that more accu-
rately determine body fat percentage (e.g. dual-energy X-ray absorptiometry)
may contribute to misclassification of individuals [28].
The potentially large, but often not considered, contribution of dietary and
other lifestyle differences to some of the obesity-related findings of the cross-
sectional studies is also a concern. Among those studies that evaluate diet,
underreporting of diet, especially associated with studies of obesity, may be
a source of bias [29]. Energy intake is, on average, underestimated, especially
when using memory recall methods [30]. Several behavioral changes may influ-
ence the accuracy of self-report, such as: change in the true intake as a function
of recording, lack of awareness of the amount of food consumed, and reluctance
to disclose amounts or foods eaten. In addition, some studies have shown that,
among dietary items reported, there is evidence for differential underreporting
of certain foods [29], creating further bias in self-reported data.
Liver
ĹSCFA Ļ
ĻFIAF ĹLPL Skeletal
Ĺ
ĹSCFA muscle
ĻFatty acid
oxidation
GPR41/43
LPS
Gram-negative
Blood
bacteria ĹAdipogenesis
Gut lumen
Macrophage
activation
Inflammation
Adipose tissue
Fig. 1. The gut microbiome and its metabolites may influence metabolism and energy
homeostasis through several mechanisms. SCFA, the products of carbohydrate fermenta-
tion, are substrates for lipogenesis, and also are signaling molecules for G protein-coupled
receptors GPR41 and GPR43. The presence of the gut microbiota and SCFA modulate
release of enteroendocrine hormones [e.g. glucagon-like peptides (GLP-1 and -2) and
PYY], affecting hunger and satiety and suppress FIAF. Reduced FIAF in turn decreases
hepatic and skeletal muscle fatty acid oxidation and increases lipoprotein lipase (LPL)
activity and triacylglycerol storage in adipose. LPS translocation across the gut epithelium
leads to higher circulating LPS concentrations and increased low-grade inflammation and
macrophage activation in adipose tissue. Adapted from Bäckhed [31] and Delzenne and
Cani [43].
and their rapid uptake of hydrogen helps to drive the process [38]. Thus, hydro-
gen transfer between hydrogen-producing bacteria and hydrogen-consuming
Archaea is hypothesized to increase fermentation efficiency and SCFA produc-
tion and contribute to increased adiposity – a relationship that has been shown
in an experimental animal model [16]. In humans, the results of one study
showed that higher numbers of Archaea and the order Methanobacteriales were
present in obese individuals (n = 3) as compared to normal-weight individuals
(n = 3) [8]. However, in an earlier study of 1,293 individuals, obesity was statis-
tically significantly less prevalent in methane producers than non-producers,
when producer status was characterized by breath methane excretion [39]. It is
likely that unaccounted dietary factors and other host exposures influence the
observed relationships.
72 Hullar · Lampe
In addition to providing energy, gut microbial metabolites, such as SCFA,
also play a role as signaling molecules, interacting with receptors in pathways
influencing energy uptake [31]. Acetate and propionate and butyrate bind to
two G protein-coupled receptors GPR41 and GPR43, also known as free fatty
acid receptors FFA3 and FFA2, respectively [40]. GPR43 is expressed in immune
cells, adipocytes, and the distal ileum and colon, and Gpr41 is expressed in adi-
pose tissue and several immune function-associated tissues, such as spleen,
bone marrow [40]. Work in GPR41-deficient mice suggests that one aspect of
the interaction between SCFA and GPR41 may be to increase levels of peptide
YY (PYY), an enteroendocrine cell-derived hormone in circulation that reduces
gut motility; reduced gut motility may aid in digestion and allow for increased
absorption of SCFA – acetate and propionate are substrates for hepatic de
novo lipogenesis and gluconeogenesis, respectively [32]. Activation of GPR43
by SCFA inhibits lipolysis and adipocyte differentiation and GPR43-deficient
mice are less prone to high-fat diet-induced obesity and have lower macrophage
numbers in their adipose tissue [41]. These findings would suggest that high
intakes of non-digestible carbohydrate might be detrimental in weight control;
however, prebiotic supplementation of animals overexpressing GPR43 and fed a
high-fat diet has been shown to reduce adiposity [42].
The impact of gut microbial fermentation products, such as the SCFA, on
release of enteroendocrine hormones involved in appetite and bodyweight reg-
ulation has been demonstrated in experimental studies in humans and animals.
Fermentable carbohydrates from several sources, even though they modulate
gut microbial community structure differently, have been shown consistently to
decrease food intake and bodyweight and with concomitant increases in pro-
duction and secretion of two anorexigenic peptides glucagon-like peptide-1 and
PYY [reviewed in 43].
Gut bacteria have also been proposed to increase triacylglycerol storage in
adipose tissue by suppressing fasting-induced adipocyte factor (FIAF) in the
gut epithelium; FIAF suppression in turn stimulates de novo hepatic lipogen-
esis and promotes lipoprotein lipase-directed incorporation of triacylglycerol
into adipocytes [15] and reduces fatty acid oxidation in the liver and in skeletal
muscle [17]. In contrast, another study, also in mice, did not support a crucial
role of the modulation of intestinal FIAF production in mediating fat storage
[18]. Recent in vitro work using coincubations of bacteria and cell lines showed
that some bacteria increased and others decreased FIAF expression in intesti-
nal cancer cell lines [44]. Further, incubation with propionate and butyrate, but
not acetate, increased FIAF expression and cleavage in colon and hepatic cell
lines, suggesting that the SCFA effects on FIAF may not be restricted to the gut
epithelium, but that bacterial effects in vivo may be mediated by gut microbial
metabolites at other tissue sites.
The gut microbiota also influences low-grade, chronic inflammation [43],
a hallmark of obesity-related diseases. Gut microbiome composition has been
Acknowledgements
Disclosure Statement
The authors declare that no financial or other conflict of interest exists in relation to the
content of the chapter.
74 Hullar · Lampe
References
1 Ley RE, Bäckhed F, Turnbaugh P, et al: 14 Mai V, McCrary QM, Sinha R, et al:
Obesity alters gut microbial ecology. Proc Associations between dietary habits and body
Natl Acad Sci USA 2005;102:11070–11075. mass index with gut microbiota composition
2 Ley RE, Turnbaugh PJ, Klein S, et al: and fecal water genotoxicity: an observational
Microbial ecology: human gut microbes study in African American and Caucasian
associated with obesity. Nature 2006;444: American volunteers. Nutr J 2009;8:49.
1022–1023. 15 Bäckhed F, Ding H, Wang T, et al: The gut
3 Gill SR, Pop M, Deboy RT, et al: microbiota as an environmental factor that
Metagenomic analysis of the human distal gut regulates fat storage. Proc Natl Acad Sci
microbiome. Science 2006;312:1355–1359. 2004;101:15718–15723.
4 Wang M, Ahrne S, Jeppsson B, et al: 16 Samuel BS, Gordon JI: A humanized gno-
Comparison of bacterial diversity along the tobiotic mouse model of host-archaeal-
human intestinal tract by direct cloning bacterial mutualism. Proc Natl Acad Sci USA
and sequencing of 16S rRNA genes. FEMS 2006;103:10011–10016.
Microbiol Ecol 2005;54:219–231. 17 Bäckhed F, Manchester JK, Semenkovich CF,
5 Guarner F, Malagelada JR: Gut flora in health et al: Mechanisms underlying the resistance
and disease. Lancet 2003;361:512–519. to diet-induced obesity in germ-free mice.
6 Qin J, Li R, Raes J, et al: A human gut micro- Proc Natl Acad Sci 2007;104:979–984.
bial gene catalogue established by metag- 18 Fleissner CK, Huebel N, Abd El-Bary MM, et
enomic sequencing. Nature 2010;464:59–65. al: Absence of intestinal microbiota does not
7 Armougom F, Henry M, Vialettes B, et protect mice from diet-induced obesity. Br J
al: Monitoring bacterial community of Nutr 2010;104:919–929.
human gut microbiota reveals an increase 19 Bäckhed F: Changes in intestinal microflora
in Lactobacillus in obese patients and in obesity: cause or consequence? J Pediatr
Methanogens in anorexic patients. PLoS One Gastroenterol Nutr 2009;48(suppl 2):S56–S57.
2009;4:e7125. 20 Kalliomaki M, Collado MC, Salminen S, et
8 Zhang HS, DiBaise JK, Zuccolo A, et al: al: Early differences in fecal microbiota com-
Human gut microbiota in obesity and after position in children may predict overweight.
gastric bypass. Proc Natl Acad Sci USA 2009; Am J Clin Nutr 2008;87:534–538.
106:2365–2370. 21 Santacruz A, Marcos A, Warnberg J, et al:
9 Turnbaugh PJ, Hamady M, Yatsunenko T, et Interplay between weight loss and gut micro-
al: A core gut microbiome in obese and lean biota composition in overweight adolescents.
twins. Nature 2009;457:480–484. Obesity (Silver Spring) 2009;17:1906–1915.
10 Collado MC, Isolauri E, Laitinen K, et al: 22 Ravussin Y, Koren O, Spor A, et al:
Distinct composition of gut microbiota dur- Responses of gut microbiota to diet composi-
ing pregnancy in overweight and normal- tion and weight loss in lean and obese mice.
weight women. Am J Clin Nutr 2008;88: Obesity (Silver Springs) 2012;20:738–747.
894–899. 23 Nechvatal JM, Ram JL, Basson MD, et al:
11 Schwiertz A, Taras D, Schafer K, et al: Fecal collection, ambient preservation, and
Microbiota and SCFA in lean and overweight DNA extraction for PCR amplification of
healthy subjects. Obesity 2010;18:190–195. bacterial and human markers from human
12 Turnbaugh PJ, Ley RE, Mahowald MA, et al: feces. J Microbiol Methods 2008;72:124–132.
An obesity-associated gut microbiome with 24 Polz MF, Cavanaugh CM: Bias in template-
increased capacity for energy harvest. Nature to-product ratios in multitemplate PCR.
2006;444:1027–1031. Appl Environ Microbiol 1998;64:3724–3730.
13 Duncan SH, Lobley GE, Holtrop G, et al: 25 Huys G, Vanhoutte T, Joossens M, et al:
Human colonic microbiota associated with Coamplification of eukaryotic DNA with
diet, obesity and weight loss. Int J Obes 2008; 16S rRNA gene-based PCR primers: possible
32:1720–1724. consequences for population fingerprinting
of complex microbial communities. Curr
Microbiol 2008;56:553–557.
76 Hullar · Lampe
Discussion
Dr. Birch: My question is whether you have information about whether participants
were breastfed or formula fed as infants?
Dr. Lampe: This was the study of Kalliomäki et al. [1]. Two groups of 7-year-old
children, normal weight and overweight or obese, were selected for retrospective
evaluation of their gut microbial communities in stool samples collected within the first
year of life. The children were part of a follow-up of a cohort of Finnish children who
had participated pre- and postnatally in a randomized trial of probiotics and atopic
disease.
Dr. Birch: Do you know if obesity status at 7 years of age was associated with
differences in infant feeding mode?
Dr. Lampe: There were no statistically significant differences in infant feeding mode
between the normal weight and overweight children. The two groups were matched for
several factors, including gestational age, BMI at birth, mode of delivery, duration of
breastfeeding, use of antibiotics during infancy, intervention group, and frequencies of
atopic diseases and atopic sensitization at 7 years of age. Other studies in breastfed and
formula-fed infants show differences in gut microbial populations, which is why
matching was important.
Dr. Haschke: The Finnish study was a secondary outcome study looking at allergies.
All the infants were breastfed until 6 months of age, and there was some intervention in
terms of giving mothers probiotics prenatally, so this study has been reported 3 times. It’s
from Erica Isolauri’s group (Department of Pediatrics, University of Turku, Finland),
and it could not be reproduced by any other group that there is some difference
continuously there, so we have to wait for confirmation. The clinical data have been
really weak until now. The hypothesis is attractive, but we still don’t have a prospective
study showing that certain microbiota could be preventive.
Dr. Lampe: Thanks for that clarification. I think you raise an important point, and
that is that we really do not have any good prospective studies. We need to encourage
cohort studies to collect and store stool samples, so that we can do the robust studies that
are needed in order to answer these questions more effectively.
Dr. Goran: I always get a little confused with some of these aspects because short-
chain fatty acids are being protective in some situations, so resistant starch for example
is promoted as a vehicle to induce insulin resistance and it starts to act through short-
chain fatty acids. Can you clarify whether there are positive and negative effects of some
of these things, especially in the context of your study where you showed that fiber and
starch were predictive of the biome? Are they predictive of a beneficial or a harmful
profile?
Dr. Lampe: All of the individuals in our observational study of diet and gut microbial
community were healthy, so we do not have information on differential relationships
dependent on metabolic syndrome or other disease phenotypes. Although the dietary
data are predictive of the microbiome, they cannot distinguish between physiologically
positive and negative effects. Looking at the totality of the predictive value of diet, these
diet components help to define different groups of individuals who have different
microbiomes. Whether or not one is better from a health standpoint, we don’t know in
the context of this study. Intervention studies of dietary fiber and gut microbial
community show shifts in different groups of bacteria and in directions that typically
78 Hullar · Lampe
Dr. Lampe: As we saw from the one slide of the study of Koenig et al. [2] – and
granted that this was just the example from one child – the gut microbiome is pretty well
established by weaning. I think the estimate is that by age 2 or 3, the gut microbiome has
the functional attributes of the adult microbial community.
Dr. Lovejoy: What would happen if you dramatically changed your diet, given the
examples of children in Africa versus children in Italy? If somebody went from a largely
meat-eating diet to becoming a vegan and staying a vegan, would their microbiome
change and stay different or would they maintain the biome that they had since 2 years
of age?
Dr. Lampe: Keep in mind that in that study, the children were born and raised to
their particular environments and diets. Nonetheless, in adults, there would be changes
to the microbiome as a result of what the person is eating and therefore what substrates
are available to the bacteria. At the same time though, particularly at the genetic level,
you are still likely to find those other bacteria that were part of the microbiome prior to
the major change in diet. It’s just that there are going to be fewer of them, but they are
still likely to be detectable unless there is some major perturbation as a result of antibiotic
treatment or disease.
References
1 Kalliomäki M, Collado MC, Salminen S, et 2 Koenig JE, Spor A, Scalfone N, et al:
al: Early differences in fecal microbiota com- Succession of microbial consortia in the
position in children may predict overweight. developing infant gut microbiome. Proc Natl
Am J Clin Nutr 2008;87:534–538. Acad Sci U S A 2011;108(suppl 1):
4578–4585.
Abstract
Obesity prevalence among infants and young children has increased rapidly during the
past 4 decades, a disturbing trend given early obesity’s association with later life obesity
and its comorbidities. Fortunately, infancy is a period of great behavioral and metabolic
plasticity offering numerous targets for preventive interventions. Modifiable factors that
may affect early rapid weight gain and obesity risk include infant sleep duration, feeding
to soothe infant distress, and the introduction of solid foods and transitional feeding. We
discuss evidence linking these factors to weight outcomes, as well as results from behav-
ioral obesity interventions in infancy, from our laboratory and others’. For example, in a
recent pilot intervention, we focused on helping new mothers address three areas of
infant behavior hypothesized to affect weight gain and early obesity risk: infant sleeping,
crying, and feeding. First-time mothers were randomly assigned to receive either a
Soothe/Sleep intervention, an Introduction of Solids intervention, both interventions, or
no interventions. The interventions were delivered via home visits and showed positive
effects on infant behaviors and weight outcomes at 1 year. Based on evidence from such
pilot interventions, we assess the plausibility of targeting behavioral factors in infancy
and suggest next steps for early prevention research.
Copyright © 2012 Nestec Ltd., Vevey/S. Karger AG, Basel
The prevalence of obesity among young children has risen rapidly during the
past 4 decades. Approximately 20% of 2- to 5-year-olds are currently over-
weight, and approximately 10% are obese. Similarly, 10% of children under age
2 have a weight status at or above the 95th percentile for age and sex [1]. These
statistics, combined with the limited success of obesity interventions target-
ing school-age children [2], highlight early obesity prevention as an important,
emerging research area. Two growth trajectories can result in obesity early in
life [3]. First, through prenatal influences (e.g. maternal pre-pregnancy weight,
gestational weight gain, smoking during pregnancy), babies can be born large
and stay large during the postnatal period. Second, babies with normal or low
birthweights can gain excessive weight during infancy. Rapid weight gain dur-
ing infancy predicts obesity and its comorbidities later in the lifespan, even after
adjustment for prenatal factors. These two distinct pathways suggest two ave-
nues for intervention: prevention efforts beginning during the prenatal period
and targeting the diet and behaviors of the mother, and postnatal prevention
efforts targeting infant factors linked to early weight gain. In this paper, the lat-
ter is our focus. However, it is important to remember that birthweight affects
subsequent growth, and in general, infants who are born heavier are more likely
to regress toward the mean and grow more slowly than those born at the lower
end of the birthweight distribution.
Rapid weight gain during infancy is associated with increased risks of obesity
and its cardiovascular comorbidities from early childhood [4] through adult-
hood [5]. In addition to the epidemiological evidence, research with animal
models reveals that early rapid weight gain can have epigenetic effects that may
alter developing metabolic systems to increase risks of obesity, metabolic syn-
drome, cardiovascular disease, and diabetes [6]. This evidence indicates that,
contrary to popular belief, a chubby baby is not necessarily a healthy baby who
will ‘grow out of it’ but may be more likely to ‘grow into’ obesity later in life.
Despite extensive evidence that early rapid growth increases obesity risk, until
very recently, few obesity prevention efforts have been focused on infancy [7].
There are many reasons to consider obesity prevention during infancy.
Obesity rates are high by early childhood, and epigenetic effects of obesogenic
environments on early growth indicate that infancy is a period of behavioral and
metabolic plasticity that can have life-long effects on health. The rapid dietary
transition from an exclusive milk diet to a modified adult diet by age 24 months
is one example of the flux and instability of infancy [8], highlighting this period
as an opportunity to shape behaviors that affect growth and obesity risk. Basic
research, in combination with a small number of pilot intervention studies, shed
light on the efficacy of targeting behavioral factors linked to infant growth.
The current state of obesity research in early life suggests that moderating early
rapid growth could be a successful first step in preventing obesity. What infant
factors should be targeted in order to achieve this goal? In a recent review, Paul
et al. [9] identified several modifiable factors, implicated by basic research
as predictors of early rapid weight gain and obesity risk. These factors were
infant feeding mode (breastfeeding or formula feeding), infant sleep duration,
parental use of food to regulate infant distress, the timing of the introduction
Infant sleep Educate parents on methods + Intervention infants had greater increases in nocturnal
to lengthen sleep duration sleep compared to controls, as well as greater
and soothe at night without reductions in settling time and night-time waking [23]
feeding as a first response to + Breastfed intervention infants showed increased
nocturnal crying nocturnal sleep duration [24]; effect not seen in infants
transitioned to formula before 16 weeks
Parental regulation of Respond quickly to crying in + Intervention infants were less likely to be given food as
infant distress early infancy, but use a reward [22]
(i.e. use of food to alternative methods to – Intervention did not affect maternal beliefs about
soothe or change soothe than feeding using food to calm baby or responsiveness to satiety
behavior when child cues [23]
is not hungry) + In [24], fussing/crying episodes were more likely to be
followed by awake/calm (as opposed to feeding; [27])
Introduction of solid – Delay introduction of + Intervention group introduced solid foods later [21, 22]
foods complementary foods + Intervention group introduced solid foods later [23]
until at least 4 months of + Tended to introduce solids later (trend level) and were
age more likely to accept novel foods [24]
– Avoid placing cereal in a
bottle
– Use repeated exposure to
healthy foods as a
response to normal infant
neophobia
Transitional feeding – Emphasize healthy + Earlier introduction of cup [21, 22] in intervention
(i.e. introduction of dietary choices group
developmentally – Use low-fat cow’s milk + Less juice feeding in intervention group [21]
appropriate foods after introduction at 1 + Less likely to be put to bed with a bottle [22]
and beverages) year – No effects on maternal intake of healthy foods [23]
– Do not any give juice to
children <6 months; then,
limit daily consumption of
100% fruit juice to <6 oz
– Give juice only in a cup;
do not allow children to
easily transport juice
– Completely avoid fruit
drinks/soft drinks
1
Intervention targets selected from Paul et al. [9], based on the existing evidence base from pilot interventions
targeting infants.
2
This column shows whether these variables were affected in the pilot studies of three groups whose studies were
reviewed herein [21–24]. If one of the studies is not mentioned in a given row, then that behavior was not targeted
in that study.
60
*
40
20
0
No Feed Sleep Both
Intervention group
Fig. 1. Infants who received both interventions [24] had lower weight-for-length percen-
tiles at age 1 year (n = 110), relative to the WHO growth standards. No = No intervention;
Feed = Introduction to Solids intervention; Sleep = Soothe/Sleep intervention; Both =
both interventions.
‘when’, ‘how’, and ‘which’ foods to introduce to infants and provided systematic
experiences with new foods between ages 4–6 months. All participants were
given a standard infant parenting book, and nurses answered questions about
general infant care. Mothers reported on infant behavioral states (sleeping, fuss-
ing/crying, awake/calm, and feeding) in 15-min intervals over 4 days at infant
ages 3 and 16 weeks. At age 1 year, infant weight and length were measured.
Infants receiving both interventions had significantly lower weight-for-
length percentiles at 1 year compared to other groups. This result is depicted in
figure 1, where the y-axis depicts weight-for-length percentiles using the World
Health Organization (WHO) growth charts, in accordance with the recent CDC
recommendation to plot the growth of American children younger than 2 using
these charts [25] and in contrast with our previously published results, which
were plotted relative to the CDC growth charts [24]. The WHO charts depict
optimal growth of breastfed infants living in families with adequate resources,
so that infant growth was not limited by food availability. These growth stan-
dards are appropriate for our sample where all mothers intended to breastfeed.
Breastfed infants who received the Soothe/Sleep intervention also slept lon-
ger at night and had fewer nightly feedings from 3 to 16 weeks compared to
infants who did not receive this intervention [24], and there was evidence that
mothers in this group were less likely to indiscriminately use food to soothe
Conclusions
Disclosure Statement
The authors declare that no financial or other conflict of interest exists in relation to the
content of the chapter.
References
1 Ogden CL, Carroll MD, Curtin LR, et al: 3 Cole TJ: Children grow and horses race: is
Prevalence of high body mass index in U.S. the adiposity rebound a critical period for
children and adolescents, 2007–2008. J Am later obesity? BMC Pediatr 2004;4:6.
Med Assoc 2010;303:242–249. 4 Taveras EM, Rifas-Shiman SL, Belfort MB, et
2 Summerbell CD, Waters E, Edmunds L, et al: Weight status in the first 6 months of life
al: Interventions for preventing obesity in and obesity at 3 years of age. Pediatrics 2009;
children. Cochrane Database Syst Rev 2005; 123:1177–1183.
CD001871.
Abstract
Obesity in the United States does not affect all segments of the population equally. It is
more prevalent in deprived neighborhoods and among groups with lower education and
incomes. Inequitable access to healthy foods is one mechanism by which socioeconomic
factors can influence food choice behaviors, overall diet quality, and bodyweight. Having
a supermarket in the immediate neighborhood has been linked to better diets and to
lower obesity rates. However, the affordability of healthy foods may have more of an
impact on food patterns than does distance to the nearest store. Grains, added sugars,
and added fats are inexpensive, good-tasting, and convenient. Their consumption has
been linked to lower quality diets, lower diet costs, and lower socioeconomic status. By
contrast, the recommended healthier diets not only cost more but were consumed by
more affluent groups. New techniques of spatial analysis are a promising approach to
mapping obesity rates and linking them with measures of socioeconomic status based on
diverse social and economic aspects of the built environment. Low residential property
values predicted bodyweights of women better than did either education or incomes.
Shopping in low-cost supermarkets was another powerful predictor of bodyweight.
Bodyweight gain may be best predicted not by any one nutrient, food or beverage but by
low diet cost. Higher obesity rates in poor neighborhoods may be the toxic consequence
of economic insecurity. Alleviating poverty may be the best, if not the only, way to stop
the obesity epidemic. Copyright © 2012 Nestec Ltd., Vevey/S. Karger AG, Basel
Introduction
The obesity epidemic in the United States has been linked to the changing food
environment. Studies have linked rising obesity rates with a growing consump-
tion of energy-dense foods, sweetened beverages, and selected dietary ingredi-
ents [1, 2]. Much research attention has focused on the consumption of added
sugars and fats and on the role of snacks, beverages, fast foods, large portion
sizes, and eating away from home. However, pinpointing which dietary factors
are directly responsible for obesity has proved difficult. Food patterns are the
result of complex interactions between the person and the social and economic
environment [3]. Although individual behavior is clearly involved, diet qual-
ity is reliably predicted by education, occupation, incomes and by other, often
unobserved, indices of social class.
Obesity rates in the US are not equally distributed across all social strata
[4]. Instead, higher rates are observed among some minorities and groups with
lower education and incomes [1, 4, 5]. Minorities and the poor are clearly at a
disadvantage when it comes to the adoption of healthier eating habits [6]. Local
disparities in access to healthy foods may be one problem. Studies on ‘food des-
erts’ [7] have shown that lower income neighborhoods attracted more fast-food
outlets and convenience stores as opposed to full-service supermarkets [8]. By
contrast, wealthier areas had access to better restaurants, fresher produce, and
more opportunities for physical activity.
Socioeconomic variables, including those in the built environment, have a
profound impact on bodyweights and health [9, 10]. Yet, greater distance to the
nearest supermarket may not fully explain why obesity rates are much higher in
poor neighborhoods [7]. That explanation may involve food affordability and
food cost. Simply put, on a per calorie basis, grains, fats and sweets cost less,
whereas many healthier and more nutrient-dense foods cost more [1, 11]. In
recent analyses of US federal data sets, higher quality diets were associated with
higher per calorie diet costs, and were more likely to be consumed by wealth-
ier and better-educated persons [12]. In other studies, the observed influence
of socioeconomic status (SES) variables on diet quality, so often attributed to
nutrition education, was partly explained by diet cost [13].
Obesity research in the US has steered clear of the complex issues of pov-
erty and social class, preferring to deal with individual-level genetics, metabo-
lism, physiology, or behavior. One concern has been that if obesity becomes a
problem of the disadvantaged and the poor, it will no longer command national
attention. However, effective obesity prevention and treatment strategies criti-
cally depend on knowing the environmental context of the national obesity
epidemic. Knowing who is most likely to become obese, where and why, is an
essential prerequisite to designing effective strategies for obesity prevention and
treatment.
Obesity rates in the US are linked inversely to education and incomes [4].
However, the observed socioeconomic gradient has not been very steep and,
other than for white women, not always readily apparent. Some researchers
96 Drewnowski
have emphasized that obesity trajectories were similar for all groups, with
obesity rates increasing steadily in both sexes, across all ages and races, and at
all incomes and educational levels [4]. By contrast, others have noted diver-
gent trajectories and a growing social gap in obesity rates among children.
Whereas obesity rates among children from the highest SES group declined
between 2003 and 2007, children from the lowest SES group continued to gain
weight [5].
Measuring the social gradient in health and bodyweight presents many
challenges. Whether past education and current income measures adequately
capture the multiple aspects of SES is unclear. Several researchers believe that
these two measures are insufficient to explain the observed influence of social
position on long-term bodyweight [14, 15]. Some have tried to supplement the
existing measures with new metrics of economic insecurity and with measures
of area-based deprivation versus long-term wealth.
Geographic mapping of obesity rates at a sufficiently fine geographic scale
offers new insights into the social and economic determinants of health [16].
Existing approaches to mapping obesity rates by state, county, or metropolitan
area tend to obscure SES distinctions by neighborhood. Whereas state- and
county-level obesity maps issued by the Centers for Disease Control are well
known, fewer studies have mapped obesity rates by political districts, health
planning areas, zip codes, census tracts or by neighborhoods. Where such data
do exist, the link between high-obesity and high-poverty census tracts becomes
more apparent.
Figure 1 shows the joint distribution of obesity and 150% poverty by census
tract for Seattle King County. Although King County is reputed to be healthy
overall, the local disparities in obesity rates by census tract ranged from 5 to
over 30%, a 6-fold difference. The map also makes it clear that high poverty and
obesity rates shared the same geographic location.
One problem with area-based data is that the links between obesity and
the built environment depend on the type of geographic aggregation and may
require complex multilevel analysis. Whereas data on poverty and wealth at the
census tract level can be readily obtained from the US Census, health and weight
data at that level of geographic resolution are exceedingly rare. As a result, we
still have an imprecise understanding of the spatial distribution of obesity and
its links to poverty and social disadvantage.
New techniques of spatial analysis may help remedy this problem. The geoc-
oding of addresses of health survey respondents allows for more sophisticated
spatial analyses of obesity at the individual level [16, 17]. For example, the
addresses of participants in the Seattle Obesity Study (SOS) were geocoded to
the centroid of the home parcel using the 2008 King County Assessor parcel
data. Geocoding followed standard methods in ArcGIS, version 9.3.1. Spatial
cluster detection analyses were then used to identify significant obesity clus-
ters in lower income neighborhoods. Relevant neighborhood features included
km 0 5
Puget
sound
Fig. 1. Joint geographic distribution of obesity (BMI >30) and 150% poverty rates in
Seattle King County by census tract.
98 Drewnowski
In other words, studies of the impact of neighborhood variables on bodyweights
and health can now be conducted at the individual level.
Residential property values became the variable of most interest, given that
home equity for most Americans represents the bulk of their wealth. Obtained
from county tax rolls, property values for study respondents may be a more
accurate measure of individual socioeconomic position than provided by
either education or income [16]. Based on objective tax data, rather than on
self-report, residential property values provide an additional link to neighbor-
hood resources, including access to food sources and local opportunities for
physical activity.
Spatial analyses, based on individual-level metrics of the built environ-
ment offer a new way to map the geographic distribution of obesity and health
behaviors across neighborhoods [17–19]. Such methods also permit a new
look at the social, economic and environmental determinants of obesity and
self-reported health. In the SOS, consistent inverse associations were obtained
between obesity and low residential property values [17]. In analyses based on
standard regression models, residential property values were the best predictor
of bodyweight of women, adjusting for individual-level education and incomes.
In contrast, and consistent with other data, property values had no impact on
bodyweights of men.
The disparity in obesity rates among women by property values was more
than 3-fold (300%). By contrast, the observed disparities in obesity rates by
race/ethnicity, education, or incomes are normally in the order of no more than
20–50%. Property values are a potentially useful and novel metric of wealth for
health studies. These new measures point to a strong social gradient in obesity
rates across neighborhoods.
100 Drewnowski
Legend
Kernel density of grocery stores
2
0.8
0
Fig. 2. Distribution of grocery stores in Seattle King County. Store clusters are identified
using Kernel Density analysis.
The SOS data run counter to the overwhelming consensus is that physical
proximity to supermarkets has a major influence on diet quality and health. In
the SOS sample, most people did not shop in the immediate neighborhood, such
that mere physical proximity to a store was not an accurate index of exposure.
Furthermore, it appeared that the poor and the wealthy shopped farther than
absolutely necessary, going up to three miles beyond the nearest store. The inter-
pretation was that lower income groups drove farther in search of food bargains,
whereas the wealthy drove to more expensive destination stores more commen-
surate with their SES.
These Seattle-based data need replicating in cities with different patterns of
food shopping. Arguably, communities may be vulnerable to obesity and chronic
disease not because the nearest supermarket is more than a mile away, but
Eggs
Vegetables
100
Fruit
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Mean cost/100 kcal (USD/100 kcal)
Fig. 3. Relation between energy density (kcal/100 g) and energy cost (USD/100 kcal)
denoting the relative positions of the different food groups. Median data by group are
from US Department of Agriculture Food and Nutrient Database for Dietary Studies
(FNDDS 2.0).
because healthy food choice is not always the most affordable choice. Systematic
efforts to improve diet quality by improving access to healthful foods will need
to take economic inequalities into account.
Food choices are made on the basis of taste, cost, convenience, health and variety
[22]. Taste refers to the sensory appeal of foods, such as palatability, aroma, and
texture. The concepts of taste and energy density are intertwined, since the most
energy-dense foods are usually the most palatable and vice versa. Energy density
of foods is defined as energy per unit weight or volume (MJ/kg) [23, 24]. Cost
refers to the purchase cost per unit of energy (USD/10 MJ) or the purchase cost of
a daily diet (USD/day). Convenience refers to the time spent on buying, preparing,
and cooking food. Variety refers to the innate drive to secure a varied diet, whereas
health refers to concerns with nutrition, chronic disease, and bodyweight.
The low cost and high palatability of energy-dense foods could help explain
why higher obesity rates are found among lower income groups [23, 24]. Grains,
fats and sweets are good-tasting, satisfying, accessible, and convenient. In general,
grains, fats and sweets cost less per calorie than do lean meats, vegetables and fruit
(fig. 3). Clinical and laboratory studies suggest that energy-dense foods have a
lower satiating power, and may lead to passive overeating and weight gain.
For those reasons, rising obesity rates have been blamed on the food envi-
ronment. Energy-dense diets, increasing portion sizes and the consumption of
102 Drewnowski
fast foods, snacks and beverages have all been linked at one time or another
to obesity risk. Again, physiological mechanisms regulating food intake were
generally thought to be at fault. Whereas some studies suggested that humans
failed to compensate for calories in liquids, other studies invoked incomplete
compensation for solid energy-dense foods.
By contrast, there has been relatively less research emphasis on the obese per-
sons’ economic environment. With the exception of studies on healthy food pat-
terns conducted by the US Department of Agriculture [25], US-based research
on diet cost in relation to health outcomes is very limited [11, 12, 26]. Existing
studies, some based on econometric modeling, do suggest that constraints on
food expenditures may contribute to the obesity epidemic, especially among
lower income groups. One hypothesis, grounded in the economics of food choice
behavior, is that individual weight gain is best predicted not so much by the con-
sumption of any one food or any one nutrient but by low overall diet cost.
Diet quality, both in the US and elsewhere, is a function of SES. It is well known
that older and wealthier consumers have higher quality, healthier, and more var-
ied diets, with more high-quality meats, seafood, vegetables and fruit [24]. In
contrast, lower income households tend to select diets with lower cost meats,
inexpensive grains, and more added sugars and fats. The observed influence of
SES on diet quality may be explained, in part, by diet cost.
In the SOS, usual dietary intakes of a representative sample of 1,295 adults
in King County (WA) were assessed based on a food frequency question-
naire [13]. Energy density (kcal/100 g) was calculated using food composition
tables. The monetary value of individual diets was estimated using local retail
supermarket prices for 384 foods. A column of prices in USD/100 g edible
portion was added to the nutrient composition database. Local prices were
attached to 384 FFQ component foods. Prices were obtained for those foods
that were most frequently consumed and for the lower cost options, includ-
ing frozen and canned foods. The underlying assumption in calculating diet
costs was that all foods were purchased at retail and then prepared and con-
sumed at home. Analogous assumptions are made by the US Department of
Agriculture in calculating the cost of healthful diets, including the Thrifty
Food Plan [25].
Mean cost per edible portion of food was calculated, after adjusting for prepa-
ration and waste, and was used to estimate the cost of daily diet. Nutrient quality
of the diet was based on nutrients of concern as identified by the 2005 Dietary
Guidelines: fiber, vitamins A, C and E, calcium, magnesium and potassium. The
more costly diets were associated with a higher consumption of fruit and veg-
etables and with lower consumption of grains, fats and sweets, after adjusting
104 Drewnowski
in common is their low cost. Diets of lower income households provide cheap,
concentrated energy from fat, sugar, cereals, potatoes and meat products – but
offer little in the way of whole grains, vegetables and fruit. Low-income con-
sumers are more likely to live in areas with limited access to healthier foods
and to be users of fast-food as opposed to full-service restaurants. The failure
to select healthy diets has been explained in terms of economic conditions that
include limited physical access to supermarkets and grocery stores and the time
spent commuting to work.
There is accumulating evidence that obesity tends to cluster in poorer neigh-
borhoods. Going beyond education and incomes, the SOS examined the influ-
ence of neighborhood type, property values, and supermarket choice on the
participants’ bodyweight. Preliminary analyses suggest that SES variables were
extremely powerful and were more strongly linked to obesity than were diet-
related variables. In other words, the new measures of SES accounted for more
variance in obesity rates than did energy density or the macronutrient composi-
tion of the diet.
Adopting a healthy diet may pose an economic as well as a behavioral chal-
lenge. Some of the current strategies for obesity prevention do not recognize
that healthier diets can cost more. Some years ago, the NIH Obesity Education
Initiative advised obese patients to look for guavas, persimmons, star fruit, kiwi,
and papaya as opposed to bologna and American cheese. Dietary Guidelines
2010 [27] recommended a healthful assortment of foods that included vegeta-
bles, fruit, whole grains, low-fat milk products, and fish, lean meat, poultry or
beans. The 2010 Guidelines further emphasized foods that were unprocessed,
fresh, and contained little sodium and no added sugars and fats. It is a matter
of some concern that those obesity prevention strategies are largely based on
recommending high-income diets to low-income people.
Studies conducted in Australia, Canada, and the EU contrast with the pre-
vailing US view that healthful diets do not represent any additional expendi-
ture to the consumer. In a French study, diets with a higher content of vitamins
and minerals were associated with higher diet costs. In fact, lower energy
density and higher nutrient density were each independently associated with
higher energy adjusted diet costs. In other modeling studies, based on linear
programming, attempting to reduce daily diet costs without taking diet qual-
ity into account led to energy-dense diets composed of grains and sweets that
were similar to those already consumed by lower income groups. Although
spending more does not assure a good diet, reducing diet costs below a certain
minimum virtually assures that the resulting diet will be energy rich but nutri-
ent poor.
Based on recent analyses of federal data sets, evidence is emerging that higher
quality diets as measured by the Healthy Eating Index (HEI) cost more. Higher
HEI scores were associated with higher diet costs, higher incomes, more educa-
tion, and with lower rates of obesity.
Economic and food policy interventions at the national and international lev-
els are the most promising approach to obesity prevention. The UK Foresight
Report [28] outlined a multisector multilevel strategy that involved all branches
of government. However, stemming the obesity epidemic cannot be sepa-
rated from stemming the tide of poverty. The rising obesity rates may reflect
the increasingly unequal distributions of incomes and wealth [29]. Evidence is
emerging that obesity in America is a largely economic issue.
Acknowledgements
Disclosure Statement
The author declares that no financial or other conflict of interest exists in relation to the
content of the chapter.
References
1 WHO Report. Global strategy on diet, physi- 7 White M: Food access and obesity. Obes Rev
cal activity and health. Geneva, World Health 2007;8(suppl 1):99–107.
Organization, 2004. 8 Morland K, Wing S, Diez RA: The contextual
2 Drewnowski A, Specter SE: Poverty and effect of the local food environment on resi-
obesity: the role of energy density and energy dents’ diets: the atherosclerosis risk in com-
costs. Am J Clin Nutr 2004;79:6–16. munities study. Am J Public Health 2002;
3 Treuhaft S, Karpyn A: The grocery gap: who 92:1761–1767.
has access to healthy foods and why it mat- 9 Grow HM, Cook AJ, Arterburn DE, et al:
ters. http://www.policylink.org/site/apps/ Child obesity is associated with social disad-
nlnet/content2.aspx?c=lkIXLbMNJrE&b=51 vantage of children’s neighborhoods. Soc Sci
36581&ct=8079863. Med 2010;71:584–591.
4 Ogden CL, Lamb MM, Carroll MD, Flegal 10 Singh GK, Siahpush M, Kogan MD:
KM: Obesity and socioeconomic status in Neighborhood socioeconomic conditions,
adults: United States 2005–2008. NCHS Data built environments and childhood obesity.
Brief 2010;50:1–8. Health Aff (Millwood) 2010;29:503–512.
5 Singh GK, Siahpush M, Kogan MD: Rising 11 Drewnowski A: The cost of US foods as
social inequalities in US childhood obesity related to their nutritive value. Am J Clin
2003–2007. Ann Epidemiol 2010;20:40–52. Nutr 2010;92:1181–1188.
6 Wang Y, Chen X: How much of racial/ethnic 12 Rehm CD, Monsivais P, Drewnowski A: The
disparities in dietary intakes, exercise, and quality and monetary value of diets con-
weight status can be explained by nutrition- sumed by adults in the United States. Am J
and health-related psychosocial factors and Clin Nutr 2011;94:1333–1339.
socioeconomic status among US adults? J
Am Diet Assoc 2011;111:1904–1911.
106 Drewnowski
13 Aggarwal A, Monsivais P, Cook AJ, 21 Jiao J, Moudon AV, Hurvitz PM, Drewnowski
Drewnowski A: Does diet cost mediate the A: How to identify food deserts: measuring
relation between socioeconomic position and physical and economic access to super-
diet quality? Eur J Clin Nutr 2011;65: markets in King County, WA. Am J Public
1059–1066. Health, Epub ahead of print.
14 Braveman PA, Cubbin C, Egerter S, et al: 22 Glanz K, Basil M, Maibach E, et al: Why
Socioeconomic disparities in health in the Americans eat what they do: taste, nutrition,
United States: what the patterns tell us. Am J cost, convenience, and weight control con-
Public Health 2010;100(suppl 1):S186–S196. cerns as influences on food consumption. J
15 Cubbin C, Pollack C, Flaherty B, et al: Am Diet Assoc 1998;98:1118–1126.
Assessing alternative measures of wealth in 23 Darmon N, Briend A, Drewnowski A:
health research. Am J Public Health 2011; Energy-dense diets are associated with lower
101:939–947. diet costs: a community study of French
16 Moudon AV, Cook AJ, Ulmer J, et al: A adults. Publ Health Nutr 2004;7:21–27.
neighborhood wealth metric for use in health 24 Monsivais P, Drewnowski A: Lower-energy-
studies. Am J Prev Med 2011;41:88–97. density diets are associated with higher mon-
17 Rehm CD, Moudon AV, Hurvitz PM, etary costs per kilocalorie and are consumed
Drewnowski A: Residential property values by women of higher socioeconomic status. J
are associated with obesity among women Am Diet Assoc 2009;109:814–822.
in King County, WA. Soc Sci Med 2012;75: 25 Carlson A, Lino M, Gerrior S, Basiotis P:
491–495. Revisions of USDA’s low-cost, moderate cost,
18 Leal C, Bean K, Thomas F, Chaix B: Are and liberate food plans. Fam Econ Nutr Rev
associations between neighborhood socioeco- 2003;15:43–68.
nomic characteristics and body mass index or 26 Monsivais P, Drewnowski A: The rising
waist circumference based on model extrapo- cost of low-energy-density foods. J Am Diet
lations? Epidemiology 2011;22:694–703. Assoc 2007;107:2071–2076.
19 Casey R, Chaix B, Weber C, et al: Spatial 27 US Department of Agriculture. Dietary
accessibility to physical activity facilities and Guidelines for Americans 2010. http://www.
to food outlets and overweight in French cnpp.usda.gov/dietaryguidelines.htm.
youth. Int J Obes (Lond), Epub ahead of 28 Kopelman P: Symposium 1: Overnutrition:
print. consequences and solutions. Foresight
20 Drewnowski A, Aggarwal A, Hurvitz PM, et report: the obesity challenge ahead. Proc
al: Obesity and supermarket access: proxim- Nutr Soc 2010;69:80–85.
ity or price? Am J Public Health, Epub ahead 29 Offer A, Pechey R, Ulijaszek S: Obesity under
of print. affluence varies by welfare regimes: the effect
of fast food, insecurity and inequality. Econ
Hum Biol 2010;8:297–308.
Discussion
Dr. Finegood: I think the take-home message I just got is that if I am poor, I should
still try to own a home and shop at Whole Foods so I don’t have the money to eat, is that
it?
Dr. Drewnowski: No.
Dr. Finegood: Obviously, I am being a bit facetious.
Dr. Drewnowski: We found that people’s attitudes towards healthy food were the
most important predictor of diet quality. If people wanted their diets to be healthy, then
they had more nutrient-dense diets, regardless of where they shopped. In other words, it
is possible to select nutrient-dense foods even within a low-cost supermarket. We are
now doing second-level analyses trying to determine whether or not health-related
attitudes determine food choices within the supermarket at all levels of income.
108 Drewnowski
are less obese. This pattern is reversed in developing countries. For minorities in the US,
things become more complicated because in some cases the more affluent minority
numbers are more likely to be overweight or obese. There is also the issue of obesity
trajectories, and here opinions vary. In the US, obesity data seem to show that all SES
groups are getting obese at the same rate, so that all the trajectories are rising in parallel.
Data from Europe suggest, to the contrary, that the social disparities are actually
increasing. So whom do you believe? In France and in other countries, the poor are
becoming obese but the rich are not, or not nearly as fast. So the divergence by SES
between the US and the EU is interesting, I am not sure what the final word is. The
consensus in the US seems to be that everyone is getting more overweight at the same
rate.
Dr. Johnson-Askew: Have you had an opportunity to look at those people in your
region that are poorer, or look at those who own their home versus those who do not?
The other question is whether you have adjusted for fast food density in your study.
Dr. Drewnowski: We have data on lower income groups. Lower income households
live in areas with lower property values, and we can track those by their geographic
location. Seattle is not particularly segregated, so that low-income areas cannot be
associated with any particular race or ethnicity. We do have geocoded data on the
locations and the density of fast foods. We can also distinguish between the fast food
restaurants that were closest to our participants’ homes and the ones that they actually
went to on a regular basis. The two were not the same. The fast food restaurants that
were actually frequented were not even in the immediate neighborhoods. So that is very
interesting from the standpoint of public policy. The current strategies to build more
supermarkets and take fast foods away are all based on the premise that people shop and
eat near where they live. That may not be the case. People make shopping and eating
decisions for many reasons. Physical proximity is just one of them.
Dr. Oppert: If I understand correctly, you don’t believe in the food desert concept.
Dr. Drewnowski: No, I don’t believe in food deserts in Seattle. I am prepared to
believe that the concept of food deserts is very real in places like New Orleans after
Katrina or in Detroit, especially if people have to walk to get the food. It’s just that in
Seattle, with our distribution of food sources and our study population, we did not see
that. We have submitted a manuscript where we added the transportation component
and car ownership to the definition of food deserts – currently thought of as census
tracts with low income and distance of more than a mile to the nearest supermarket. We
added the mode of transportation: food deserts expand or shrink depending on whether
you walk, bike, take the bus, or drive. So, if you walk, there will be places where the
nearest supermarket is more than 10 min away. But if you bike, take public transport or
go in a car, then pretty much every residential address in Seattle can access a supermarket
within 10 min. It may not be the right supermarket but there is a supermarket.
Dr. Oppert: You also showed that some fast food outlets were clustered.
Dr. Drewnowski: Driving influences development.
Dr. Oppert: So, don’t you feel that we are just following what people have done when
they have built these food outlets along the major roads?
Dr. Drewnowski: Most of the fast food restaurants were clustered along big arterial
roads in a ribbon pattern and did not necessarily target middle schools or low-income
areas. Restaurants, including fast foods, tend to locate where the people are, which is
something that people in the business world already know about. The location of food
110 Drewnowski
bypassing many others on the way, so there must have been something special about
those stores. I think that every supermarket would like to distinguish itself in some way
and become the preferred shopping destination for people coming in from all over the
place. We were able to create some interesting shopping polygons to map each
supermarket’s service area. We found that adjoining supermarkets could serve customers
of very different SES.
Dr. Barclay: I am trying to understand why low-income populations are choosing
not only cheaper but also less healthy foods. Is it about education? Should supermarkets
help people to monitor their health? Should the processed food industry focus on
improving the nutritional quality of lower cost foods?
Dr. Drewnowski: It was not all about money. There were people who were going to
low-cost supermarkets but who were still selecting a nutritious diet. There are other
aspects of culture, lifestyles and attitudes that affect food selection. Whether the answer
lies in education, health promotion, price interventions, supermarket interventions, or
food fortification, I can’t tell you at this point.
Whatever the best solutions to obesity prevention are, they need to come from the
ground up and be a part of a systems approach. Grassroots initiatives are better than
government-imposed taxation, prohibition, or legislation. Something has to happen to
shape consumer behavior for the better. In our research, we are just beginning to
delineate food shopping behavior. I know that the easy answer is to build more
supermarkets, but bringing a supermarket half a mile closer may not affect shopping or
eating behavior all that much.
Dr. St. Jeor: This is fascinating data, and how it comes together is quite complex. I
was just curious if you are working with the food industry. What about commercial
development done by city planners because they must be ahead of this in one way or
another? And finally, how do you plan to interpret your findings to benefit the obese or
the non-obese population?
Dr. Drewnowski: We have not worked with a supermarket chain on this research
project, but I think that will come. We do work with researchers in architecture and in
urban planning and with the local transportation people and with the Seattle city
planners. That is our new audience in the area of obesity prevention. We are interested in
working with supermarkets and grocery stores as potential partners in fighting the
obesity epidemic. National statistics show that 62% of calories in the diet of adults were
purchased in a grocery store. By contrast, the amount of calories supplied by fast foods
or vending machines is very small. National statistics also show that 72% of added sugars
are purchased in a grocery store. Supermarkets and grocery stores are the main sources
of calories for obese and lean people.
Dr. St. Jeor: But do you have a plan? How can you put some of your findings into
action to help with this problem?
Dr. Drewnowski: Our research relates to food systems and food policy. It tells us that
looking at food retail is very important. Showing where the calories are coming from is
an important and overlooked component. The environmental and socioeconomic
factors can also supplement the data on the physiology of obesity that we already have.
In the systems approach, everything is interconnected. There are many leverage points
that can be used to improve diets.
Dr. Finegood: I am still stuck on that 8-fold difference in obesity rates by supermarket
chain, and I am wondering whether the obese people who live in high-income
112 Drewnowski
New Directions for Prevention
Drewnowski A, Rolls BJ (eds): Obesity Treatment and Prevention: New Directions.
Nestlé Nutr Inst Workshop Ser, vol 73, pp 113–121,
Nestec Ltd., Vevey/S. Karger AG., Basel, © 2012
Abstract
There is increasing interest in identifying characteristics of neighborhood environments
(physical, social, economical) that might favor unhealthy dietary and physical activity pat-
terns leading to excess weight at population level. Measurement of characteristics of the
physical environment in relation to food and physical activity has greatly improved in
recent years. Methods based on assessment of perceptions by residents of their neigh-
borhood or on objective assessment of the actual built environment (such as provided by
Geographic Information Systems tools) would benefit to be combined. A number of
recent systematic reviews have updated our knowledge on relationships of food and
physical activity environments with relevant behaviors and obesity. Available evidence
appears to show more consistent evidence of association between built environment
characteristics related to physical activity (‘walkability’ indices, land use mix, variety of
transports. . .) with physical activity behavior than with weight status. In contrast, built
environment characteristics related to food habits (accessibility to different types of food
outlets, availability of healthy foods. . .) would be more consistently associated with
weight status than with eating behavior. The need for data from different countries and
cultures is emphasized, as much as the importance of transdisciplinary research efforts for
translation of these findings into our living environment.
Copyright © 2012 Nestec Ltd., Vevey/S. Karger AG, Basel
Introduction
Recent systematic reviews have updated the status of knowledge regarding rela-
tionships of food and physical activity environments with relevant behaviors and
obesity [13–15]. For the food environment, in the review by Giskes et al. [13]
focused on obesogenic dietary intakes in adults (≥18 years), 28 original stud-
ies (from 2005 to 2008) were reviewed. Number of participants ranged from 22
to 1,000. Most studies originated from the US (n = 16), followed by Australia
(n = 6), Japan (n = 2), the UK (n = 2), The Netherlands (n = 1), and New Zealand
(n = 1). All but one were cross-sectional studies. Interestingly, among a total
of 18 accessibility factors assessed, 16 were objective measures. An important
finding emerging from this review is that current literature in adults appears
to show more consistent evidence of associations between environmental fac-
tors and weight status than between environmental factors and obesity-related
dietary intakes. Greater accessibility to supermarkets and lower access to take-
away outlets were found associated with lower BMI or prevalence of overweight/
obesity. No consistent association was found between fruit and vegetable con-
sumption and access to supermarkets or takeaway outlets, or availability/shelf
space of fruits and vegetables. In contrast, area-level socioeconomic status was
more consistently associated with healthier dietary behaviors. In children, based
on objective measures of environmental factors, available evidence suggests that
weight status is positively related to spatial accessibility to convenience stores, but
findings with other food retail outlets and restaurants appear mixed [14].
For the physical activity environment, Ding and Gebel [15] performed a
review of reviews on associations between built environment characteristics,
physical activity behavior and obesity based on reviews published between 1990
and 2011. Among 37 reviews included for examination of their key characteris-
tics, a vast majority (n = 27) dealt with physical activity and only a few with obe-
sity (n = 5) or both physical activity and obesity (n = 5). Most reviews focused
on youth (n = 12), and only 5 targeted adults and 2 senior residents. Very few
reviews considered specific populations such as African-Americans, low socio-
economic or rural residents (n = 1 each) pointing to the need for more data in
these groups.
Another recent review also emphasizes the fact that fewer studies have
assessed associations of physical environment attributes with weight status
than with physical activity [16]. In this latter review by Durand et al. [16], 5–10
times more studies assessed relations with walking behavior compared to BMI/
weight status for environmental characteristics such as ‘walkability’ (composite
indices usually including residential density, street connectivity and land use
mix), mixed land use, variety of transports available, or density of built space.
Moreover, if 30 to about 60% of associations between built environment char-
acteristics were found in the expected direction with walking behavior, this
Food ± +
Physical activity ++ ±
proportion was reduced to only 10–20% for associations with BMI. Positive
associations between accessibility to green space and both physical activity and
weight status are also increasingly being reported [17]. Finally, in older adults,
current literature appears inconsistent on the relation between walkability indi-
ces and physical activity [18].
Altogether, available evidence more consistently shows an association
between built environment characteristics related to physical activity with
physical activity behavior than with weight status, whereas built environment
characteristics related to food habits would be more consistently associated with
weight status than with eating behavior (table 1). The importance of combin-
ing perceived and objectively determined aspects of the environment has again
been recently emphasized. Gebel et al. [19] observed that residents who per-
ceived neighborhoods to be less walkable than objectively determined (through
GIS) were less active, more obese and more likely to decrease physical activity
and to gain weight over time than those with a more accurate environmental
perception.
Significant advances have been made in recent years regarding the theories and
methods used to study the food and physical activity environments. However,
we have to acknowledge that major challenges are ahead to better understand
the complex pathways through which attributes of the built environment may
impact weight status, in conjunction with neighborhood and individual socio-
economic characteristics.
In terms of measurement, defining the size of the neighborhood in which the
relation between environment and behavior operates remains a methodological
issue, as much as ways to combine refined objective spatial measures (such as
GPS) with assessment of how residents perceive their environment. The combi-
nation of data from multiple sources and obtained with different types of sensors
will require the development of new data platforms for integration and analysis.
Designing and implementing longitudinal studies are on the list of priorities,
Acknowledgments
Part of the work of the authors reported in this paper is from the ELIANE study. ELIANE
is a project supported by the French National Research Agency (ANR-07-PNRA-004,
coordinator J.M. Oppert).
Disclosure Statement
J-M. O serves on scientific advisory boards for Vivus, Tanita, Institut Benjamin Delessert
and has received funding from Fondation Le Roch (France).
References
1 Richard L, Gauvin L, Raine K: Ecological 5 Penchansky R, William Thomas J: The
models revisited: their uses and evolution in concept of access. Definition and relation-
health promotion over two decades. Annu ship to consumer satisfaction. Med Care
Rev Public Health 2011;32:307–326. 1981;XIX:127–140.
2 Brug J, van Lenthe FJ, Kremers SPJ: 6 Glanz K: Measuring food environments.
Revisiting Kurt Lewin: how to gain insight A historical perspective. Am J Prev Med
into environmental correlates of obesogenic 2009;36(suppl 4):S93–S98.
behaviors. Am J Prev Med 2006;31:525–529. 7 Sallis J: Measuring physical activity envi-
3 Foresight: Tackling Obesities: Future Choices ronments. A brief history. Am J Prev Med
– Project Report. London, Government 2009;36(suppl 4):S86–S92.
Office for Science, 2007. 8 Charreire H, Casey R, Salze P, et al:
4 Bergeron P, Reyburn S: Impact of built envi- Measuring the food environment using
ronment on physical activity, food intake and geographical information systems: a meth-
body weight (in French). Québec, Institut odological review. Public Health Nutr 2010;
National de la Santé Publique, 2010. 13:1773–1785.
Discussion
Dr. Drewnowski: I was very interested to see the major differences in urban form
between Berlin, Shanghai and New York. Context can be very important, and what we see
in Seattle may not apply to Shanghai or to Paris. We need to do this kind of transdisciplinary
transcontinental research to figure out which features of the food environment are
common and which are unique to specific cities. It’s an interesting question.
Dr. Oppert: Right, but you have to convince the funders.
Dr. Finegood: It seems that all this work on the causal mechanisms of obesity is not
going to get us very far because it’s not terribly transferable from one place to another.
Shouldn’t we get more focused on solution-oriented research which isn’t worried about
the causes of the problem but is more focused on what solutions are going to work in
Shanghai versus New York versus another place? Your argument about causality – I
would say forget it and move on.
Dr. Oppert: Yes, of course we emphasize cultural specificity; however, we can
anticipate that some associations will be transferable from one place to the other, or at
least this is what we are trying to do with Dr. Drewnowski, comparing shopping in Paris
with Seattle supermarkets.
Abstract
Obesity is clearly a complex problem for both the individual and for society. Complex or
‘wicked’ problems have common characteristics such as heterogeneity, nonlinearity,
interdependence, and self-organization. As such they require solutions appropriate for
complex problems, rather than a reductionist search for the causes. ‘Systems thinking’
provides new ways to consider how to collectively address complex societal problems like
obesity, where biology interacts with social, cultural and built environmental factors in
infinite permutations and combinations. The systems that give rise to the obesity epi-
demic function at multiple levels, and there are important interactions between these
levels. At any given level, individual actors and organizations matter and system function
is optimized when individual and organizational capacity to respond is well matched to
the complexity of individual tasks. Providing system supports to help networks of indi-
viduals become ‘communities of practice’ and ‘systems of influence’ may also help to accel-
erate the pace of effective action against obesity. Research efforts need to move away
from the relentless search for the specific isolated causes of obesity and focus on solu-
tions that have been shown to work in addressing other ‘wicked’ problems.
Copyright © 2012 Nestec Ltd., Vevey/S. Karger AG, Basel
For many years, the obesity ‘problem’ has been framed within the paradigm that
people need to be individually responsible to eat less and move more [1]. While
public discourse remains mostly rooted in this paradigm, academic literature
has shifted to include recognition of the importance of many social, psychologi-
cal and physical environment variables in causing obesity. This has helped to
move the dialogue somewhat from individual to societal responsibility [2], but
Table 1. Characteristics of simple, complicated and complex systems
blame is still ascribed in the tone of moral panic which supports the notion that
people must be individually responsible for their weight [3].
Obesity research is usually based on conceptual models that have focused
more on working out the causes of obesity than on developing solutions, espe-
cially solutions appropriate for complex problems. Biological and physiological
mechanisms associated with obesity were the exclusive focus of obesity research
when the first journal devoted to obesity was introduced in the late 1970s [4].
The deeply held belief of this biomedical focus is that working out the causes
of a problem will lead to solutions. This paradigm remained dominant until
the late 1990s when papers on population and public health began to appear
in the three obesity research journals that existed by that time [4]. The rise in
population level research helped to increase the number of identified causes and
has opened the door to a discussion of societal level responsibility, but has not
yet led to a shift from the basic paradigm that solutions need to be rooted in an
understanding of the causes [5].
In the last few years, obesity has been labeled a complex or ‘wicked’ problem
[6]. Characteristics giving rise to this complexity include the heterogeneity of
our genes and environments, nonlinearities in processes like weight loss, the
importance of triggers to sustained behavior change, and the many reinforcing
feedback loops that drive individuals towards less healthy behaviors (table 1).
Current scientific approaches, especially those rooted in an understanding of
the causes, are better suited to problems that are simple or complicated, not ones
that are complex. Simple problems have simple solutions based on the causes of
the problem and sometimes complicated and complex problems can have simple
solutions (or a set of simple solutions), but these solutions need to be based on
a systems approach and not necessarily on the causes. As Wagner [7] points out,
causality can only be meaningfully defined for systems with linear interactions.
Rittel and Webber [8] suggest we need a different way to approach wicked
problems. They suggest that scientific methods have been developed to address
124 Finegood
‘tame’ problems, not ones where the problem cannot be definitively described,
where there are no ‘optimal’ solutions, and it is impossible to define a ‘stopping
rule’, or to know when the problem is solved [8]. Solutions appropriate for com-
plex problems tend to be more process (as opposed to outcome) oriented and
focus on the interactions and interdependencies between individuals, organiza-
tions or levels in the system [9].
New streams of research rooted in systems thinking are needed to build
novel and effective approaches to address obesity and other complex health and
societal problems [6]. But achieving this goal will require several fundamental
paradigm shifts. We need to focus on solutions rather than the causes of the
problems [5], be holistic and integrative rather than reductionist, focus on pro-
cesses rather than outcomes, and understand both our collective and individual
roles and responsibilities.
The introduction of the Foresight Obesity System Map in 2007 [10] helped initi-
ate a dialogue about the complexity of obesity and how systems thinking and
systems science can help to change the way we approach looking for solutions
[6]. The Obesity System Map is a causal loop diagram which helps communi-
cate the complexity of the system as a whole and emphasizes the importance of
feedback loops and interdependencies to the development of obesity [10]. The
map, a product of a stakeholder engagement process, illustrates many possible
connections between important subsystems including food production, social
psychology and physical activity environments [11]. The map was built both on
available evidence and the experience of the stakeholders involved in its con-
struction; specific linkages may only be relevant to some individuals in the con-
texts in which they live, learn, work and play.
Common responses to complex problems include despair, retreat, believing
the problem is beyond hope and assigning blame [9]. But accepting complexity
does not mean we need to give up, rather it means we need to turn to solutions
appropriate for complex problems. When viewing solutions from a reductionist
paradigm, the natural tendency is to think solutions for complex problems need
to be multi-level and comprehensive [12, 13]. There is evidence that compre-
hensive interventions are more effective, but comprehensive interventions can
also overwhelm the systems that support them, the resources available and they
are difficult to evaluate [14].
Can accepting complexity help us identify new ‘simple’ solutions? Many sys-
tems thinkers, researchers and writers articulate ‘simple’ principles that under-
lie systems approaches. Bar-Yam [9] reminds us we need to look for solutions
in the interactions between a system and its environment, such as the capacity
of actors to act relative to the complexity of their tasks. He also suggests that
Fail
Complexity of environment
Fig. 1. Matching capacity and complexity. When the capacity of an individual or organi-
zation is greater than the complexity of their environment (or tasks), the individual or
organization will survive. When capacity is less than complexity the individual or organi-
zation will fail.
126 Finegood
While this idea that capacity and complexity need to be well matched is
a theoretical construct, it also makes sense intuitively and can be applied
directly to thinking about solutions. Rather than asking ‘what are the various
biological, social and environmental factors causing obesity?’ it suggests that
research should focus on where in the complex system (described for example
by Foresight Obesity System map) there are mismatches between capacity and
complexity. More importantly, it leads us to consider interventions that reduce
complexity (rather than increase it) or increase capacity.
One could argue that the dominant focus on educational approaches to
addressing obesity has been an attempt to increase people’s capacity for healthy
eating and active living. Yet, the fact that education has largely failed to halt or
reverse the epidemic suggests either that the marginal increase in capacity that
might derive from an educational approach is insufficient, or that the focus of
most educational programs is not actually helping to increase capacity. Different
educational targets may be better able to increase capacity, but it is also likely
that education is not sufficient to overcome other drivers that make it more
complex to act according to what we know [18].
The need to match capacity and complexity also suggests we need to
reduce the complexity of healthy eating and active living. Many environmental
approaches and the notion that ‘we need to make the healthy choice the easy
choice’ are consistent with reducing complexity, but some changes may inad-
vertently increase complexity. Consider for example the impact of increasing
the availability of healthy food choices. Behavioral economists have shown that
as the number of choices increase, we are less likely to actually make an active
choice and instead will make our decisions by default [19]. For this reason, it
is not enough to make the healthy choice an easy choice; we must also make
unhealthy choices more difficult or non-existent.
Lastly, when we consider variables like our capacity for behavior change and
the complexity of the change, it is not enough to only consider the rational com-
ponents of behavior. Emotional drivers can easily outweigh rationality [18]. As
such, it is not enough to make the healthy choice the easy choice; we must also
make it the desirable choice [20]. Having more desirable defaults when choices
are being made will reduce the complexity of making healthy choices and ensure
more individuals have the capacity to succeed.
Level Definition
such as subsidies, taxes and standards) and (11) buffers (the size of stabilizing
stocks relative to their flows, e.g. the number of people on earth relative to the
number who are born and die each day). Other leverage points include (7 and
8) balancing and reinforcing feedback loops, (4) self-organization, (2) the par-
adigm or mind-set of the system (the deeply held beliefs) and (1) transcend-
ing the paradigm (letting go of beliefs and assumptions about the system). As
Meadows points out, the leverage points with the higher numbers are relatively
easier to implement, but also are less effective, whereas changing the deeply held
beliefs under which a system operates is much harder, but also more effective.
While the 12 leverage points Meadows described provide a helpful frame-
work for a systems approach to complex problems, with 12 levels, this framework
was difficult to translate into a tool for planning and/or analyzing intervention
approaches [21]. We collapsed the 12 levels into a 5-point intervention level
framework (table 2). The 5 levels are paradigm, goals, structure (as a whole),
feedback loops/delays, and structural elements, and include all 12 of Meadows
levels [22]. With only 5 levels, this framework has the potential to be applicable
to a variety of challenges associated with complex problems including under-
standing how a system operates to assessing the compatibility of actions at dif-
ferent levels of a system and across a range of goals.
Our first application of this framework was to a set of materials provided as
pre-conference reading for a meeting on food systems and public health [23].
The actions recommended within these readings that spoke to making food sys-
tems healthy, green, fair and affordable were sorted into the intervention level
framework [22]. This analysis suggested that some actions to achieve all four
goals are compatible, including broad public discussion and implementation
of policies and programs that support sustainable food production and distri-
bution. At the level of paradigm and goals, however, the challenge of making
healthy and green food also affordable becomes apparent as some actions at
some levels may be in conflict.
Current and future uses of the intervention level framework include other
analyses of qualitative data, supporting reflective practice to enable cross-sector
dialogue, program planning, research and evaluation. As Meadows points out
128 Finegood
in her ‘guidelines for living in a world of systems’, it is very important to ‘get
the beat of the system’ and to ‘expose your mental models to the light of day’.
Recognizing the multiple levels at which a system of interest operates and where
the changes are needed is an important step to tackling a complex problem for
the good of the whole. Asking questions about what is happening and what
needs to change at each level of a system and in the interaction of the levels is a
good place to start.
Influencing Emergence
Emergence refers to the arising of novel structures, patterns, and properties
during the process of self-organization in complex systems [24], and is often
thought of as ‘the whole is greater than the sum of the parts’. Bar-Yam [25]
defines multiple types of emergence depending on the nature of the relationship
between the parts and the whole. ‘Weak’ emergence is the difficult to understand
micro-to-macro relationship between the parts and the whole, whereas ‘strong’
emergence can arise either through system constraints or from a global-to-local
causality. The Foresight Obesity System map suggests that obesity results from
weak emergence of a large number of factors. The map illustrates how more
than one hundred different ‘micro’ level variables and their interdependencies
give rise to obese individuals. It is also likely that strong emergence contributes
to the epidemic of obesity. For example, globalization of the food supply clearly
has an effect on local food environments. A better understanding of system con-
straints and global-to-local relationships could provide new insights into how to
influence the emergence of obesity.
Wheatley and Frieze [26] suggest ways to influence emergence even in the
absence of a deeper understanding of system constraints or causality. They con-
sider that networks are the only form of organization used by living systems
and since they are mostly interested in social systems, this frame works well and
enables them to describe emergence as ‘the fundamental scientific explanation
for how local changes can materialize as global systems of influence’ [26]. They
suggest that as a change theory, this approach offers methods and practices to
influence emergence and accomplish system-wide changes.
Networks are defined as the first stage in the life cycle of emergence. They
tend to be based on self-interest, are self-organized and have fluid membership.
The next stage is ‘communities of practice’. These can spring from networks, are
also self-organized, but motivation for participation goes beyond self-interest.
In communities of practice, people participate to serve their own needs, but also
to serve the needs of others; there is an intentional commitment to advance the
field of practice and to share learning with a wider audience. The speed with
which people exchange knowledge, learn and grow can be rapid in a community
of practice. The last stage in the life cycle of emergence is the ‘system of influ-
ence’ where efforts and ideas that were expressed by a few suddenly become the
norm. Practices developed by courageous communities suddenly become the
Conclusions
Acknowledgements
The author wishes to gratefully acknowledge feedback on the manuscript from Dr.
Carrie L. Matteson, Philippe Giabbanelli, Lee Johnston and Penny Deck and grant sup-
port from the Canadian Institutes of Health Research.
130 Finegood
Disclosure Statement
The author of this chapter does not have any relationships to disclose.
References
1 Saguy A, Almeling R: Fat in the fire? Science, 12 Huang TT-K, Glass T: Transforming research
the news media, and the ‘obesity epidemic.’ strategies for understanding and preventing
Sociol Forum 2008;23:53–83. obesity. JAMA 2008;300:1811–1813.
2 Ries NM, Rachul C, Caulfield T: Newspaper 13 Gortmaker SL, Swinburn BA, Levy D, et
reporting on legislative and policy inter- al: Changing the future of obesity: science,
ventions to address obesity: United States, policy, and action. Lancet 2011;378:838–847.
Canada, and the United Kingdom. J Public 14 Wakefield M, Chaloupka F: Effectiveness of
Health Policy 2011;32:73–90. comprehensive tobacco control programmes
3 Boero N: All the news that’s fat to print: the in reducing teenage smoking in the USA.
American ‘obesity epidemic’ and the media. Tobacco Control 2000;9:177–186.
Qual Sociol 2007;30:41–60. 15 Bar-Yam Y: Complex systems and sports:
4 Bedoya DL, Matteson CL, Finegood DT: complex systems insights to building effec-
Evolution of conceptual models of obesity: tive teams. http://necsi.edu/projects/yaneer/
from simple to complex. Can J Diabetes SportsBarYam.pdf.
2011;35:216. 16 Meadows DH, Wright D: Thinking in
5 Robinson TN, Sirard JR: Preventing child- Systems. Vermont, Chelsea Green, 2009.
hood obesity: a solution-oriented research 17 Wheatley MJ, Kellner-Rogers M: Bringing
paradigm. Am J Prev Med 2005;28(suppl 2): life to organizational change. J Strategic
194–201. Perform Manag 1998 April/May:5–13.
6 Finegood DT: The complex systems science 18 Heath D, Heath C: Switch: how to change
of obesity; in Cawley J (ed): The Oxford things when change is hard. Toronto,
Handbook of the Social Science of Obesity. Random House Canada, 2010.
Oxford, Oxford University Press, 2011, pp 19 Ariely D: Predictably Irrational: The Hidden
208–236. Forces That Shape Our Decisions. New York,
7 Wagner A: Causality in complex systems. HarperCollins, 2009.
Biol Philos 1999;14:83–101. 20 French J, Blair-Stevens C, McVey D, Merritt
8 Rittel HWJ, Webber MM: Dilemmas in a R: Social marketing and public health: theory
general theory of planning. Policy Sci 1973;4: and practice. Oxford, Oxford University
155–169. Press, 2010.
9 Bar-Yam Y: Making Things Work: Solving 21 Malhi L: Places to intervene in the obesity
Complex Problems in a Complex World. system; undergraduate thesis; Burnaby,
Cambridge, NECSI Knowledge Press, 2004. Simon Fraser University, 2009.
10 Vandenbroeck P, Goossens J, Clemens 22 Malhi L, Karanfil Ö, Merth T, et al: Places
M: Foresight Tackling Obesities: Future to intervene to make complex food systems
Choices – Building the Obesity System more healthy, green, fair, and affordable.
Map. London, Government Office for J Hunger Environ Nutr 2009;4:466–476.
Science, 2007, pp 1–76. http://www.bis.gov. 23 Story M, Hamm MW, Wallinga D: Food
uk/assets/bispartners/foresight/docs/obe- systems and public health: linkages to
sity/12.pdf. achieve healthier diets and healthier
11 Finegood DT, Merth TDN, Rutter H: communities. J Hunger Environ Nutr
Implications of the foresight obesity system 2009;4:219–224.
map for solutions to childhood obesity. 24 Goldstein J: Emergence as a construct: his-
Obesity 2010;18(suppl 1):S13–S16. tory and issues. Emergence 1999;1:49–72.
Discussion
Dr. Drewnowski: Let me begin by asking you a question about your experience at the
Canadian Institutes of Health Research (CIHR) and the emerging interdisciplinary
networks in the area of obesity.
Dr. Finegood: Networks and teams are helpful ways to advance obesity research
because they create an interface where different disciplines can learn about each other
and can bring complimentary skill sets together to work on a problem. At the CIHR, we
created several opportunities for multi- and transdisciplinary teams to form and to
enable both basic research and research on natural experiments. By creating a competition
for funding, we enabled the cooperation of researchers across their usual silos. By
supporting research on natural experiments, where the researcher does not have control
over the intervention, we have begun to learn more about policy level interventions like
the return of PartipAction and the implementation of a physical activity tax credit. Those
projects have been fairly effective at emphasizing the value of intervention and
interdisciplinary research and giving it a foothold so that it could grow.
Dr. Drewnowski: Our experience was similar in that our work with urban planners
and geographers was sparked by a planning grant from the National Institutes of Health
for a transdisciplinary exploratory center for obesity research. Putting money on the
table is the best way to promote interdisciplinary research between people who would
never talk to each other otherwise.
Dr. Finegood: By creating competition at the right level, you create cooperation at
the next level down. Think sports teams and leagues, the most competitive team in the
league has the most cooperative players. This systems idea can be used in the research
funding domain to stimulate cooperation, but also elsewhere in government.
Dr. Bray: We had a program in place, briefly, where we had the police department,
the university system, and government people with all of us donating our time. The
mayor was actually quite engaged in community wide health. The program was really
comprehensive, but the problem was that it wasn’t self-sustaining. There are a lot of
people who acknowledge that obesity is a complex problem and there are people who
want to participate. But those of us, who want to do it, can’t do it endlessly without some
sort of support and somehow that self-sustaining component is often not there.
Dr. Finegood: Many of us understand that obesity is a big complex problem, and we
need to address many different facets of it. Our reductionist mindset has led us to huge
and comprehensive solutions and those are very hard to self-sustain. But if we shift our
mindset a little bit and zero in on integration and integrative structures, without trying
to be comprehensive in everything we do, then we can be more self-sustaining and
132 Finegood
adaptable. This is where supporting emerging networks can be helpful; by taking people
who are on the ground doing things separately and making strategic connections, you
enhance the ability of the structures to function. You do not necessarily make people do
more, but by making strategic links you begin to move from a network to a ‘community
of practice’ and ultimately to a ‘system of influence’. These are things that we can do if we
get away from the mindset that everything has to be done together in one big effort.
Does that make sense?
Dr. Bray: Yes it does, and I also think that sustainability has to also be part of the
plan.
Dr. Finegood: If we are thinking that the solution to a complex problem lies in
working out the reasons for the problem, linking these causes to their specific outcomes
and then intervening at the cause, then we are ignoring the potential for intervention
through integrative approaches. The path to workable solutions through working out
the causes of a problem is slow and very costly due to its complexity. I believe that simple,
more sustainable solutions can be developed within whatever resources the actors bring
to the table if the focus for the group is problem solving around their shared goals and
integrative processes.
Dr. Rosenbaum: But doesn’t the importance of working out causality depend on
what your goal is? Isn’t it different when your goal is ‘health improvement’ as compared
to ‘reducing body fatness’, not that the two are necessarily separate? Isn’t causality
important to work out treatment as compared to prevention? It’s like looking for ways to
treat measles or for ways to prevent it, that’s the question. I guess the official topic of this
session was prevention, so I was thinking that causality is very important. Prevention
and treatment are completely different problems.
Dr. Finegood: There probably are differences in the importance of working out
causality for prevention and treatment. Tom Robinson argues that we will speed up the
finding of effective solutions if we test hypotheses not about the ‘causes of the problem’,
but instead we focus on the ‘causes of the solutions’. Both health improvement and
reducing body fatness are complex problems and chasing causality in both cases isn’t
necessarily all that helpful. And yes, there probably are differences between prevention
and treatment outcomes for variables like physical activity or the nature of the food
environment.
Dr. Rosenbaum: The relationship between food and physical activity is very different
in someone at their usual weight than in somebody who is reduced weight. I like your
idea of capacity and complexity because I think that the kind of physical activity that is
necessary to lose weight may well be beyond the capacity of most of us. We just can’t do
it, we don’t have the time, and we don’t have the capacity.
Dr. Finegood: This really strikes me as being a view about the individual. The
difference between prevention and treatment may apply to an individual. But the same
argument about capacity and complexity also applies to the food environment. I don’t
think that the changes that we need to make in our food environment are going to be
different for obesity prevention and obesity treatment necessarily. So, in that sense I
don’t think they are different. It depends on what level you are at as to whether that
distinction is important or not.
Dr. Rosenbaum: I would say that they are very different. In one case, your body is
fighting against you in terms of treatment; there are a million things that fight against
134 Finegood
Dr. Finegood: I didn’t tell you anything about the CAPTURE Project, which is
another half of my life at the moment. CAPTURE stands for a CAnadian Platform To
increase Usage of Real-world Evidence. It is an effort to build a system to support the
collection, sharing, and use of practice-based or real-world evidence. It is intended to
support communities and people on the ground. The goal is to figure out what data
would help them and then use that data to figure out what they want to do. If we were
better at doing that kind of work, then we could solve some of these problems without
necessarily spending huge amounts of money. It is the people on the ground that need to
explore the diversity of ideas and come together to figure out what is possible with the
resources they have and the context that they are in.
Dr. Drewnowski: One way to provide helpful data is to aggregate obesity statistics by
political district. The moment you do that, it becomes somebody’s responsibility. It’s just
another way of analyzing geographic data but everything changes because now it can be
brought to the attention of politicians and policy makers. So, it’s one way of making best
use of local data.
Dr. Bray: There is no way that obesity prevention or intervention projects can depend
on an endless flow of money. They have to be community oriented, it has to be something
intrinsic to the program itself, and it probably will need to arise from the community.
Dr. Barclay: I think there has been at least one reasonably successful multisectorial
collaboration which started in France (EPODE: http://www.epode-european-network.
com). Although it hasn’t been able to reverse obesity, it has been able to stabilize obesity
in children. The program brings together players from government, education, and
industry, etc.
Dr. Finegood: Funded if I am not mistaken by Nestlé, is that correct?
Dr. Barclay: Nestlé is one of the co-funders.
Dr. Finegood: It’s a good example of what can happen if you can engage people at
multiple levels and in multiple sectors.
Dr. Drewnowski: I have a question for you based on your experience as a former
head of a research institute. We have heard about various aspects of science at the
individual level and environmental level; we have heard about the microbiome,
inflammation, the brain, and the environment. Based on the people in this room, who
would you put together with whom and why to create an emerging network?
Dr. Finegood: What I would say to that is it’s not up to me because it’s not a command
and control kind of problem. If I can create the competition that stimulates cooperation
then it’s up to the people in the room to figure out who they can work with and what
makes the most sense. I have led enough multidisciplinary research teams to know that
it’s a challenging thing to do. It takes time for people to learn each other’s language. The
important thing is to create the opportunity and to give them the time and maybe
financial support. But there are some multidisciplinary questions and problems that we
do need to resolve.
Dr. Drewnowski: I think you are right; it’s a question of time and trust. At one point
granting agencies had a list of bullets saying that you must have one person from this
area and one person from that area.
Dr. Finegood: Forcing it doesn’t work that well. People have to know each other. This
was a technique used at CIHR over the years. Sometimes what you get are researchers
going to people in government on the day that the grant is due saying: will you sign this
letter of support. That is not what I call authentic engagement. But at other times, you
136 Finegood
image of a healthy food as there is developing the actual healthy food. Subway has done
an excellent job of marketing themselves as a supplier of healthy food and created a
health halo for their products, but not all of their sandwiches are in the same healthy
category.
Dr. Finegood: I am not going to stand up here and defend the food industry for its
practices because many of those practices are questionable, and yet I would defend them
for wanting to sell their food. If we want to stay in our corners and not have conversations
about this, how can we support the shift that the companies need to make, the iterative
gradual shifts that they need to make in order to deliver a much healthier food supply
that is honestly healthy and is not just capitalizing on the apparent demand for healthy
food? One of the things that we need to do is to get our government policies to change.
Dr. Johnson-Askew: I would like to know what role you think systems modeling has
as we approach this very complex problem of obesity.
Dr. Finegood: One thing systems modeling can do is to help us integrate the data we
do have so as to create a better understanding of the big picture, although there is a part
of me that thinks it’s just the way reductionist scientists need to go about addressing a
complex problem. Josh Epstein answers the question of ‘why model?’ with many different
points about how it improves the rigor of our consideration of fact and assumption,
helps us test predictions, and can illuminate core dynamics and uncertainties.
Dr. Lovejoy: I wanted to bring up something that we haven’t talked about much
today, which is the role of stress. As we are thinking about obesity treatment and
prevention, we should look at the physiological effects of stress, which certainly impact
the brain, the gut, and inflammation, and at the behavioral effects which impact people’s
desire to eat different foods and their willingness to engage in physical activity. Research
has been done showing that the built environment contributes to stress and lack of sleep,
so I think that stress is the often overlooked third leg of the obesity story. We talk about
diet and physical activity all the time, but we need to consider that stress may be just as
important in terms of the big picture.
Dr. Drewnowski: I am glad that you mentioned stress. People who work on the
relation between the built environment and obesity have been using the stress hypothesis
a lot. Living in a low-income neighborhood with no access to facilities and resources can
increase stress levels and alter work patterns, sleep patterns, and food choice. So, the
stress hypothesis has been quite prominent, and there is increasing literature on that. We
did not talk about it today, but that is not to understate its importance, it’s a major
factor.
Dr. Warshaw: Building on this, I want to introduce another topic that relates both to
treatment and prevention – the importance of support. Study after study suggests that
support has to be a critical issue. In the national diabetes prevention program which is a
CDC project from the affordable care act, they are going to be using some community
health workers. There is a lot of interest in using community health workers and
paraprofessionals as supporters since they are less expensive than trained professionals.
Such support seems particularly important for weight loss maintenance. We tend to
support people intensively during the weight loss phase and then there often is withdrawal
of support, but yet so much research suggests that’s when people need support. We need
to communicate to the public just how hard it is to keep weight off and that they likely
will continue to need support.
Dr. Drewnowski: I agree with you. There are two important issues related to social
support. Having spouse, family, or peer support during the weight loss phase is critical to
its success. The other issue is having sufficient social support to prevent weight gain in
the first place. The latter is sometimes called social capital as distinct from income and
material resources. Basically, it’s who you know that can provide you with support and
with social resources. Even if your income is low, but you still have parents, family,
neighbors, a house and some assets, you will probably do okay. But there are people who
are completely on their own, without a social support network. Such issues are not picked
up in studies on obesity and SES because many just focus on incomes and nothing else.
There are other aspects of social context and social capital which are very important, and
we need a better handle on those.
Dr. Finegood: We heard lots of good data this morning that would argue that it is
more important to support people after the first 6 months of weight loss than it is
during the first 6 months. You can do almost any treatment in those first 6 months, and
people will lose weight. It is after the first 6 months that people are struggling and need
support. I have often thought that the challenge, in my experience, is that weight loss
gets harder, not easier. But if you get that idea out there, it may prevent people from
ever starting weight loss – so what will that do? In my own experience, very small
behavior changes at the beginning were the easy ones. I could start off by parking my
car far away and taking the stairs. Those are things that I have now incorporated in a
sustainable way in my own behavior. But now, 10 years later, I am at that point where I
am at a steady state. I can’t find anything else to do that is easy enough and that I am
capable of doing to bring my bodyweight down further and to sustain that weight loss.
And yet, the social norm and my knowledge drive me nuts because I am still obese. So
I think you are right, our strategies are backwards. We need to focus a little less on our
initial weight loss interventions during the first 6 months and put more emphasis on
maintenance.
Dr. Oppert: Going back to prevention versus treatment, I think that obesity treatment
is more of an individual topic but that prevention should be more properly focused at
the population level. For example, obesity treatment in Europe is managed by national
health systems but prevention issues can be addressed at an international meeting at the
European level. So, I agree that there are similarities between prevention and treatment,
but there are also major differences. However, I think it is an interesting issue that
prevention efforts aimed at changing the environment will also affect the environment
of obese patients. So, prevention initiatives apply not only to the whole population but to
obese patients and may help with treatment.
Dr. Rosenbaum: Would you say that it’s harder to engage politicians in prevention
than in treatment? In my experience, working with the public schools in New York City,
if you say you are going to do something now that is going to decrease the prevalence of
obesity 10 years from now, they say that’s very nice but I am not going to be here in 10
years. They want something that before the next election is going to change this. It’s
much easier to get the powers that you alluded to on board with something that they can
see right now. That’s a big barrier to some of the changes that we need to make.
Dr. Finegood: That is a fundamental challenge in terms of getting government
regulation. The way I would think about it is that the things that we might do around
prevention tend to be at the higher levels. They are less about genes and proteins or even
individuals per se, and they are more about systems and structures. And the things that
we do at the higher levels can be helpful in both prevention and treatment. But it’s true
that treatment occurs at the individual level, particularly with the morbidly obese who
need individual attention. For example, there are some people for whom bariatric surgery
may be the best solution. So yes, treatment can be individualized, but the prevention step
is going to apply to everybody.
Concluding Remarks
147
obesity infancy interventions, see Infancy
mechanisms of action 71–74 interventions, obesity
profiles 68–71 metabolically healthy obesity 50, 51, 58
prevalence trends 1, 2, 81
Infancy interventions, obesity recommendations for treatment and
behavioral factors linked to weight prevention studies 143–145
outcomes 82–85 systems approach for intervention
complementary foods 83–85 causality and complexity 123–125
cultural considerations 93 competition for healthier food 136
design 92 influencing emergence 129, 130,
food preferences 91, 92 134
outcome studies 85–88 leverage points in complex
prenatal factors 82, 92 systems 127–129
prospects for study 88, 89 matching capacity and
sleep 83, 84 complexity 126, 127
Insulin, salsalate effects on sensitivity 54, Omega-3 fatty acids, anti-inflammatory
55 activity 53
Interleukin-6 (IL-6)
adipocyte production 50 Physical activity
dietary response 52 diet synergy in weight loss 30, 31
health benefits beyond weight loss 28,
Leptin 35
bodyweight regulation 2 weight loss effects
repletion effects in weight loss 10, exercise intensity 31, 32
11 guidelines 35, 36
Lipopolysaccharide (LPS), inflammatory overview 22–24
response 74 responders versus nonresponders
adherence factors 25–27, 32, 33
Macrophage, weight loss effects 52, 58, energy balance 27
59 genetics 25
Microbes, see Gut microbiome sex differences 34
Monocyte chemoattractant protein-1 Physical activity environment
(MCP-1) assessment 114, 115
adipocyte production 50 behavior and weight status
weight loss response 51 relationship 116, 117
Myosin heavy chain I, weight loss intervention strategies and prospects
response in muscle 9 for study 117, 118
Portion size, see Diet
Nuclear factor-κB (NF-κB), salsalate Poverty, see Socioeconomic status
inhibition 54, 55 Pre-portioned foods (PPFs) 41
Nutrition, see Diet
Resolvin D1, anti-inflammatory
Obesity activity 53
economics 106
gut microbiome Salsalate, anti-inflammatory activity 54, 55
mechanisms of action 71–74 SERCA, weight loss response in muscle 9,
profiles 68–71 10