ArticlePDF AvailableLiterature Review

Abstract and Figures

Weight loss is a major concern for the US population. Surveys consistently show that most adults are trying to lose or maintain weight (1). Nevertheless, the prevalence of overweight and obesity has increased steadily over the past 30 years. Currently, 50% of all adult Americans are con- sidered overweight or obese (2,3). These numbers have serious public health implications. Excess weight is associ- ated with increased mortality (4) and morbidity (5). It is associated with cardiovascular disease, type 2 diabetes, hypertension, stroke, gallbladder disease, osteoarthritis, sleep apnea and respiratory problems, and some types of cancer (6,7). Most people who are trying to lose weight are not using the recommended combination of reducing caloric intake and increasing physical activity (1). Although over 70% of persons reported using each of the following strategies at least once in 4 years, increased exercise (82.2%), decreased fat intake (78.7%), reduced food amount (78.2%,) and re- duced calories (73.2%), the duration of any one of these behaviors was brief. Even the most common behaviors were used only 20% of the time (8). Obesity-related conditions are significantly improved with modest weight loss of 5% to 10%, even when many patients remain considerably overweight (6). The Institute of Medicine (9) defined successful long-term weight loss as a 5% reduction in initial body weight (IBW) that is main- tained for at least 1 year. Yet data suggest that such losses are not consistent with patients’ goals and expectations. Foster (10) reported that in obese women (mean body mass index [BMI] of 36.3 􏰃 4.3) goal weights targeted, on average, a 32% reduction in IBW, implying expectations that are unrealistic for even the best available treatments. Interestingly, the most important factors that influenced the Address correspondence to Dr. Janet King, U.S. Department of Agriculture, Agricultural Research Service, Western Human Nutrition Research Center, University of California, 1 Shield Avenue, Building Surge IV, Room 213, Davis, CA 95616. E-mail: jking@ whnrc.usda.gov Copyright © 2001 NAASO selection of goal weights were appearance and physical comfort rather than change in medical condition or weight suggested by a doctor or health care professional. Is it any wonder that overweight individuals are willing to try any new diet that promises quick, dramatic results more in line with their desired goals and expectations than with what good science supports? The proliferation of diet books is nothing short of phe- nomenal. A search of books on Amazon.com using the key words “weight loss” revealed 1214 matches. Of the top 50 best-selling diet books, 58% were published in 1999 or 2000 and 88% were published since 1997. Many of the top 20 best sellers at Amazon.com promote some form of carbo- hydrate (CHO) restriction (e.g., Dr. Atkins’ New Diet Rev- olution, The Carbohydrate Addict’s Diet, Protein Power, Lauri’s Low-Carb Cookbook). This dietary advice is counter to that promulgated by governmental agencies (US Department of Agriculture [USDA]/Department of Health and Human Services, National Institutes of Health) and nongovernmental organizations (American Dietetic Associ- ation, American Heart Association, American Diabetes Association, American Cancer Society, and Shape Up America!). What is really known about popular diets? Is the in- formation scientifically sound? Are popular diets effec- tive for weight loss and/or weight maintenance? What is the effect, if any, on composition of weight loss (fat vs. lean body mass), micronutrient (vitamin and mineral) status, metabolic parameters (e.g., blood glucose, insulin sensitivity, blood pressure, lipid levels, uric acid, and ketone bodies)? Do they affect hunger and appetite, psy- chological well-being, and reduction of risk for chronic disease (e.g., coronary heart disease, diabetes, and osteo- porosis)? What are the effects of these diets on insulin and leptin, long-term hormonal regulators of energy in- take and expenditure? The objective of this article is to review the scientific literature on various types of popular diets based on their macronutrient composition in an attempt to answer these questions (see Appendix for diet summaries).
Content may be subject to copyright.
Popular Diets: A Scientific Review
Marjorie R. Freedman, Janet King, and Eileen Kennedy
EXECUTIVE SUMMARY
Introduction
Weight loss is a major concern for the US population.
Surveys consistently show that most adults are trying to lose
or maintain weight (1). Nevertheless, the prevalence of
overweight and obesity has increased steadily over the past
30 years. Currently, 50% of all adult Americans are con-
sidered overweight or obese (2,3). These numbers have
serious public health implications. Excess weight is associ-
ated with increased mortality (4) and morbidity (5). It is
associated with cardiovascular disease, type 2 diabetes,
hypertension, stroke, gallbladder disease, osteoarthritis,
sleep apnea and respiratory problems, and some types of
cancer (6,7).
Most people who are trying to lose weight are not using
the recommended combination of reducing caloric intake
and increasing physical activity (1). Although over 70% of
persons reported using each of the following strategies at
least once in 4 years, increased exercise (82.2%), decreased
fat intake (78.7%), reduced food amount (78.2%,) and re-
duced calories (73.2%), the duration of any one of these
behaviors was brief. Even the most common behaviors were
used only 20% of the time (8).
Obesity-related conditions are significantly improved
with modest weight loss of 5% to 10%, even when many
patients remain considerably overweight (6). The Institute
of Medicine (9) defined successful long-term weight loss as
a 5% reduction in initial body weight (IBW) that is main-
tained for at least 1 year. Yet data suggest that such losses
are not consistent with patients’ goals and expectations.
Foster (10) reported that in obese women (mean body mass
index [BMI] of 36.3 4.3) goal weights targeted, on
average, a 32% reduction in IBW, implying expectations
that are unrealistic for even the best available treatments.
Interestingly, the most important factors that influenced the
selection of goal weights were appearance and physical
comfort rather than change in medical condition or weight
suggested by a doctor or health care professional. Is it any
wonder that overweight individuals are willing to try any
new diet that promises quick, dramatic results more in line
with their desired goals and expectations than with what
good science supports?
The proliferation of diet books is nothing short of phe-
nomenal. A search of books on Amazon.com using the key
words “weight loss” revealed 1214 matches. Of the top 50
best-selling diet books, 58% were published in 1999 or 2000
and 88% were published since 1997. Many of the top 20
best sellers at Amazon.com promote some form of carbo-
hydrate (CHO) restriction (e.g., Dr. Atkins’ New Diet Rev-
olution, The Carbohydrate Addict’s Diet, Protein Power,
Lauri’s Low-Carb Cookbook). This dietary advice is
counter to that promulgated by governmental agencies (US
Department of Agriculture [USDA]/Department of Health
and Human Services, National Institutes of Health) and
nongovernmental organizations (American Dietetic Associ-
ation, American Heart Association, American Diabetes
Association, American Cancer Society, and Shape Up
America!).
What is really known about popular diets? Is the in-
formation scientifically sound? Are popular diets effec-
tive for weight loss and/or weight maintenance? What is
the effect, if any, on composition of weight loss (fat vs.
lean body mass), micronutrient (vitamin and mineral)
status, metabolic parameters (e.g., blood glucose, insulin
sensitivity, blood pressure, lipid levels, uric acid, and
ketone bodies)? Do they affect hunger and appetite, psy-
chological well-being, and reduction of risk for chronic
disease (e.g., coronary heart disease, diabetes, and osteo-
porosis)? What are the effects of these diets on insulin
and leptin, long-term hormonal regulators of energy in-
take and expenditure?
The objective of this article is to review the scientific
literature on various types of popular diets based on their
macronutrient composition in an attempt to answer these
questions (see Appendix for diet summaries).
Address correspondence to Dr. Janet King, U.S. Department of Agriculture, Agricultural
Research Service, Western Human Nutrition Research Center, University of California, 1
Shield Avenue, Building Surge IV, Room 213, Davis, CA 95616. E-mail: jking@
whnrc.usda.gov
Copyright © 2001 NAASO
OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001 1S
Evidence-Based Guidelines
This article is limited to the effects of popular diets in
overweight and obese adults; there are no good data on
children and adolescents. Dietary claims are scrutinized,
diets are analyzed, and information is compared with sci-
entific data published in peer-reviewed journals. No pub-
lished studies are excluded, despite inherent methodological
problems (e.g., small or inadequate sample size, limited
duration, lack of adequate controls and randomization, poor
or minimal dietary collection and/or description of diets,
and potential biases). However, the strength of the evidence
supporting various conclusions made throughout the paper
is based on the following grading system used by National
Heart, Lung, and Blood Institute (NHLBI) (6) (Table 1).
Characterization of Diets
Diets are characterized below and in Tables 2 and 3.
High-fat (55% to 65%), low-CHO (100 g of CHO per
day), high-protein diets (e.g., Dr. Atkins’ New Diet Rev-
olution, Protein Power, Life Without Bread).
Moderate-fat (20% to 30%), balanced nutrient reduction
diets, high in CHO and moderate in protein (e.g., USDA
Food Guide Pyramid, DASH diet, Weight Watchers).
Low-fat (11% to 19%), and very-low-fat (VLF) (10%),
very-high-CHO, moderate-protein diets (e.g., Dr. Dean
Ornish’s Program for Reversing Heart Disease, Eat
More, Weigh Less, The New Pritikin Program).
Summary of Findings
Weight Loss
Diets that reduce caloric intake result in weight loss. In
the absence of physical activity, a diet that contains
1400 to 1500 kcal/d, regardless of macronutrient com-
position, results in weight loss. Individuals consuming
high-fat, low-CHO diets may lose weight because the
intake of protein and fat is self-limiting and overall
caloric intake is decreased (11,12). Low-fat and VLF
diets contain a high proportion of complex CHOs, fruits,
and vegetables. They are naturally high in fiber and low
in caloric density. Individuals consuming these types of
diets consume fewer calories and lose weight (13–17).
Balanced nutrient reduction diets contain moderate
amounts of fat, CHO, and protein. When overall caloric
intake is reduced, these diets result in loss of body weight
and body fat (6,18). Importantly, moderate-fat, balanced
nutrient reduction diets produce weight loss even when
they are consumed ad libitum.
In sum, all popular diets, as well as diets recommended
by governmental and nongovernmental organizations, result
in weight loss. However, it is important to note that weight
loss is not the same as weight maintenance.
Evidence Statement: Caloric balance is the major
determinant of weight loss. Diets that reduce caloric
intake result in weight loss. In the absence of physical
activity, the optimal diet for weight loss contains
1400 to 1500 kcal/d, regardless of macronutrient
composition. Evidence Category A.
Evidence Statement: Free-living overweight indi-
viduals who self-select high-fat, low-CHO diets con-
sume fewer calories and lose weight. Evidence Cat-
egory C.
Table 1. Grading system and evidence categories
Evidence category Sources of evidence Definition
A RCTs (rich body of data) Evidence is from endpoints of well-designed RCTs (or
trials that depart only minimally from
randomization) that provide a consistent pattern of
findings in the population for which the
recommendation is made.
B RCTs (limited body of data) Evidence is from endpoints of intervention studies that
include only a limited number of RCTs, post hoc or
subgroup analysis of RCTs, or meta-analysis of
RCTs. In general, Category B pertains when few
randomized trials exist, they are small in size, and
the trial results are somewhat inconsistent.
C Nonrandomized trials observational studies Evidence is from outcomes of uncontrolled or
nonrandomized trials or from observational studies.
RCT, randomized controlled trial.
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
2S OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001
Evidence Statement: Overweight individuals con-
suming high-fat, low-CHO, low-calorie diets under
experimental conditions lose weight. Evidence Cate-
gory C.
Evidence Statement: Overweight individuals con-
suming moderate-fat, balanced nutrient reduction di-
ets lose weight because they consume fewer calories.
These diets can produce weight loss when consumed
ad libitum. Evidence Category A.
Evidence Statement: Overweight individuals con-
suming low-fat and VLF diets lose weight because
they consume fewer calories. Evidence Category B.
Evidence Statement: Weight loss on VLF diets
may be the result of lifestyle modification, which may
include decreased fat and energy intake, increased
energy expenditure, or both. Evidence Category B.
Body Composition
As body weight decreases, so does body fat and lean body
mass. The optimal diet for weight loss is one that maximizes
loss of body fat and minimizes loss of lean body mass. All
low-calorie diets result in loss of body weight and body fat
(6). Macronutrient composition does not seem to play a
major role (19–22). In the short-term, however, high-fat,
low-CHO ketogenic diets cause a greater loss of body water
than body fat (23). When these diets end, water weight is
regained (24). Eventually, however, all reduced calorie diets
result in loss of body fat if sustained long term (25).
Physical activity, an important factor with respect to lean
body mass, should be promoted to enhance the effects of
diet on body composition.
Evidence Statement: All low-calorie diets result
in loss of body weight and body fat. Macronutrient
composition does not seem to play a major role.
Evidence Category A
Evidence Statement: In the short term, low-CHO
ketogenic diets cause a greater loss of body water
than body fat. Water weight is regained when the diet
ends. If the diet is maintained long term, it results in
loss of body fat. Evidence Category C.
Nutritional Adequacy
Proper food choices are always important when consid-
ering the nutritional quality of a diet. When individuals
consume foods from all food groups, it is more likely that
their diet will be nutritionally adequate. The moderate-fat,
balanced nutrient reduction diet is optimal for ensuring
adequate nutritional intake. However, poor food choices
may result in inadequate levels of nutrients (e.g., calcium,
iron, zinc), regardless of overall macronutrient composition.
High-fat, low-CHO diets are nutritionally inadequate. They
are low in vitamins E, A, thiamin, B
6
, folate, calcium,
magnesium, iron, potassium, and dietary fiber, and require
supplementation. These diets are high in saturated fat and
cholesterol. VLF diets are low in vitamins E, B
12
, and zinc
because meat and fat intake is low.
Evidence Statement: With proper food choices,
the moderate-fat, balanced nutrient reduction diet is
nutritionally adequate. Evidence Category B.
Table 2. Characterization of diets as percentage of
calories
Type of diet
Fat
(% kcals)
CHO
(% kcals)
Protein
(% kcals)
High-fat, low-CHO 55–65 20% (100 g) 25–30
Moderate-fat,
balanced nutrient
reduction
20–30 55–60 15–20
Low- and very-
low-fat
10–19 65 10–20
Table 3. Characterization of diets in absolute amount (grams)
Type of diet Total kcals Fat g (%) CHO g (%) Protein g (%)
Typical American 2200 85 (35) 275 (50) 82.5 (15)
High-fat, low-CHO 1414* 94 (60) 35 (10) 105 (30)
Moderate-fat, balanced nutrient reduction 1450 40 (25) 218 (60) 54 (15)
Low- and very-low-fat 1450 16–24 (10–15) 235–271 (65–75) 54–72 (15–20)
* Based on average intake of subjects who self-selected low-CHO diets (see Table 4).
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001 3S
Evidence Statement: High-fat, low-CHO diets are
nutritionally inadequate, and require supplementa-
tion. Evidence Category C.
Evidence Statement: VLF diets are low in vita-
mins E, B
12
, and zinc. Evidence Category B.
Metabolic Parameters
Low-CHO diets result in ketosis, and may cause a sig-
nificant increase in blood uric acid concentrations.
Blood lipid levels (e.g., total cholesterol [TC], low-den-
sity lipoprotein [LDL], high-density lipoprotein [HDL] and
triglycerides [TGs]) decrease as body weight decreases
(6,26,27). However, the macronutrient and fatty acid com-
position of energy-restricted diets can exert substantial ef-
fects on blood lipids. There are significantly greater de-
creases in LDL cholesterol during active weight loss when
diets are low in saturated fatty acids. Changes in HDL
cholesterol depend on dietary fat content and duration of
energy restriction (28). Moderate-fat, balanced nutrient re-
duction diets reduce LDL-cholesterol and normalize the
ratio of HDL/TC.
Plasma TG levels also decrease with weight loss (6).
Although they increase in response to short-term consump-
tion of a VLF, high-CHO diet (29), the type of CHO
consumed must be considered. High-fiber foods, including
vegetables and legumes, do not lead to hypertriglyceridemia
(30), and may easily be incorporated into moderate-fat,
balanced nutrient reduction diets to help normalize plasma
TG levels.
Energy restriction, independent of diet composition, im-
proves glycemic control (21,22,31–33). As body weight
decreases, so does blood insulin and plasma leptin levels
(21,34).
Blood pressure decreases with weight loss, independent
of diet composition (6,22,26). However, the DASH diet,
high in fruits, vegetables, and low-fat dairy products effec-
tively lowers blood pressure (35).
Evidence Statement: High-fat, low-CHO diets re-
sult in ketosis. Evidence Category B.
Evidence Statement: Metabolic profiles are im-
proved with energy restriction and weight loss. Evi-
dence Category A.
Evidence Statement: Low-CHO diets that result
in weight loss may also result in decreased blood lipid
levels, decreased blood glucose and insulin levels,
and decreased blood pressure. Evidence Category C.
Evidence Statement: Low-fat and very low-fat
diets reduce LDL-cholesterol, and may also decrease
plasma TG levels, depending on diet composition. Ev-
idence Category B.
Evidence Statement: Moderate-fat, balanced nu-
trient reduction diets reduce LDL-cholesterol, nor-
malize the ratio of HDL/TC, and normalize plasma
TGs. Evidence Category A.
Hunger and Compliance
Many factors influence hunger, appetite, and subsequent
food intake. Macronutrient content of the diet is one, and it
may not be the most important. Neurochemical factors (e.g.,
serotonin, endorphins, dopamine, hypothalamic neuropep-
tide transmitters), gastric signals (e.g., peptides, stomach
distention), hedonistic qualities of food (e.g., taste, texture,
smell), genetic, environmental (e.g., food availability, cost,
cultural norms) and emotional factors (e.g., eating when
bored, depressed, stressed, happy) must be considered.
These parameters influence appetite primarily on a meal-to-
meal basis. However, long-term body weight regulation
seems to be controlled by hormonal signals from the endo-
crine pancreas and adipose tissue, i.e., insulin and leptin.
Because insulin secretion and leptin production are influ-
enced by the macronutrient content of the diet (36,37),
effects of different diets on these long-term regulators of
energy balance also need to be considered when investigat-
ing hunger and appetite.
All fat-restricted diets provide a high degree of satiety.
Subjects who consume fat-restricted diets do not complain of
hunger, but of having “too much food” (38,39). These diets,
high in fiber and water content are low in caloric density.
Subjects who consume these diets develop a distaste for fat
(40), which may be useful in long-term adherence to reduced
fat, low-calorie diets. However, it is not clear that restricting fat
provides any advantage over restricting CHOs. Ogden (41)
reports weight loss maintainers used healthy eating habits and
adhered to calorie-controlled diets.
Long-term compliance to any diet means that short-term
weight-loss has a chance to become long-term weight main-
tenance (42–44). Dietary compliance is likely a function of
psychological issues (e.g., frequency of dietary counseling,
coping with emotional eating, group support) rather than
macronutrient composition, per se (42,45). Being conscious
of one’s behaviors, using social support, confronting prob-
lems directly, and using personally developed strategies
may enhance long-term success (46). Ogden (41) notes that
successful weight loss and maintenance may be predicted
by an individuals’ belief system (e.g., that obesity is per-
ceived as a problem that can be modified and if modifica-
tions bring changes in the short-term that are valued by the
individual concerned).
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
4S OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001
Evidence Statement: Many factors influence hun-
ger, appetite, and subsequent food intake. There does
not seem to be an optimal diet for reducing hunger.
Evidence Category B.
Evidence Statement: Long-term compliance is
likely a function of psychological issues rather than
macronutrient composition. Evidence Category B.
Summary and Recommendations
Caloric balance (calories in vs. calories out), rather than
macronutrient composition is the major determinant of
weight loss. However, what is not clear is the effect of
macronutrient content on long-term weight maintenance
and adherence. Furthermore, it is not known whether main-
tenance of weight loss and dietary adherence is related to
psychological issues (and brain neurochemistry), physiolog-
ical parameters (e.g., hormones involved in body weight
regulation such as insulin and leptin), physical activity,
energy density, or some other factor(s).
Controlled clinical trials of high-fat, low-CHO, and low-
fat and VLF diets are needed to answer questions regarding
long-term effectiveness (e.g., weight maintenance rather
than weight loss) and potential long-term health benefits
and/or detriments.
Prevention of weight gain and weight maintenance are
important goals. Scientifically validated, yet understandable
information is clearly needed by millions of overweight and
obese Americans who often find weight loss attainable, but
maintaining weight loss nearly impossible.
I. High-Fat, Low-Carbohydrate Diets
(55% to 65% fat, <100 g of CHO per day)
Despite controversy surrounding their use, high-fat, low-
CHO diets are among the most popular types of diets today.
The most famous is Dr. Atkins’ Diet Revolution (47) first
published in 1972, and updated 20 years later as Dr. Atkins’
New Diet Revolution (48). Promoting a “lifetime nutritional
philosophy,” Atkins claims that his diet has been embraced
by an estimated 20-plus million people worldwide since the
release of his first book (www.atkinscenter.com). His pro-
gram was one of the first to popularize low-CHO, high-
protein, ketogenic diets that individuals could use on their
own, rather than in a medical setting (e.g., a protein-sparing
modified fast). Other low-CHO diets with similar themes
include Protein Power (49), The Carbohydrate Addict’s
Diet (50), Dr. Bernstein’s Diabetes Solution (51), and Life
Without Bread (52).
A. Underlying Philosophy, Claims, and Proposed
Solutions
Proponents of high-fat, low-CHO diets dismiss the notion
that caloric intake is important to either weight gain or
weight loss. They claim that “most overweight individuals
do not overeat” (48, p. 7; 50, p. 21), even as they suggest
that high-CHO meals leave individuals less satisfied than
meals that contain adequate fat, resulting in increased hun-
ger and increased food intake (48, p. 55; 50, p. 43). They
suggest that those who do overeat do so “because of a
metabolic component driving them on, most often a truly
addictive craving for CHOs” (48, p. 7; 52, p. 142). Because
“carbohydrates are addictive,” the carbohydrate “addict”
continues to eat carbohydrates, producing more and more
insulin, which inhibits brain serotonin release. Reductions
in this “satiety” neurotransmitter result in a decreased sense
of satisfaction (50, pp. 26, 43; 51, p. 41). With respect to
weight loss, Atkins (48) claims that on a low-CHO diet
there are “metabolic advantages that will allow overweight
individuals to eat as many or more calories as they were
eating before starting the diet yet still lose pounds and
inches” (p. 10).
Furthermore, proponents contend overproduction of in-
sulin, driven by high CHO intake, is the cause of the
metabolic imbalance that underlies obesity (48,50,53). Eat-
ing too much CHO results in increased blood glucose,
increased blood insulin, and increased TGs (48, pp. 50–51).
An already overweight person who continues to overeat
CHOs develops hyperinsulinemia and insulin resistance,
“resulting in insulin’s lack of effectiveness in converting
glucose into energy, but enabling glucose (e.g., dietary
CHO) to be stored as fat” (48, p. 52).
Advocates of low-CHO diets propose a simple solution to
this “vicious cycle” of CHO addiction, CHO overeating,
hyperinsulinemia, decreased glucose use and increased fat
storage. It involves restricting CHOs severely enough to
produce ketosis. The ketosis is a reliable indicator of fat
mobilization. In this condition, the key benefit is that blood
glucose and blood insulin levels are reduced, and appetite is
suppressed. In short, authors contend that a high-fat, low-
CHO, high-protein, ketogenic diet results in weight loss,
body fat loss, preservation of lean body mass, and correc-
tion of serious medical complications of diabetes (51), heart
disease, and high blood pressure (48, pp. 6, 63). The con-
tention is that the high-fat, low-CHO diet supports long-
term health, controls weight without hunger, and should be
followed for the rest of one’s life (48, p. 27).
B. Scientific Evaluation of Claims
1. Caloric Intake, Body Weight, and Body Composition
Is caloric intake relevant when looking at weight gain
and weight loss?
What is the effect of diet composition on weight loss,
e.g., will consuming a high-fat, low-CHO diet, regard-
less of caloric intake, result in weight loss, body fat
loss, and preservation of lean body mass?
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001 5S
Energy intake and energy expenditure are relevant when
looking at weight gain and weight loss. Overweight and
obesity results from an energy imbalance (e.g., excess ca-
loric intake, decreased energy expenditure, or both) (54).
Reduction of body weight and body fat can be achieved by
creating an energy deficit (e.g., restricting energy intake,
increasing energy expenditure, or a combination of the two)
(6,18,54). Atkins (48) calls these basic thermodynamic prin-
ciples “a millstone around the neck of dieters and a miser-
able and malign influence on their efforts to lose” (p. 6). Do
followers of high-fat, low-CHO diets have a metabolic
advantage that enables them to eat a greater number of
calories, and still lose body weight and body fat?
No scientific evidence exists to suggest that low-CHO
ketogenic diets have a metabolic advantage over more con-
ventional diets for weight reduction (55). Studies consis-
tently show that under conditions of negative energy bal-
ance, weight loss is a function of caloric intake, not diet
composition (54). Table 4 indicates diet composition of
individuals who self-select high-fat, low-CHO diets, and
Tables 5a and 5b show weight change in obese individuals
consuming high-fat, low-CHO diets. In all cases, individu-
als on high-fat, low-CHO diets lose weight because they
consume fewer calories.
Evidence Statement: Free-living overweight indi-
viduals who self-select high-fat, low-CHO diets con-
sume fewer calories and lose weight. Evidence Cat-
egory C.
Evidence Statement: Overweight individuals consum-
ing high-fat, low-CHO, low-calorie diets under experi-
mental conditions lose weight. Evidence Category C.
Caloric Intake and Weight Change
Studies cited by Atkins (pp. 67–74) to support his
contentions were of limited duration, conducted on a
small number of people, lacked adequate controls, and used ill-
defined diets (24,58,61,63–65,67,68,71). Some of these, as
well as other studies, actually refute the contention that low-
CHO diets, in the absence of energy restriction, provide a
metabolic advantage (11,12,21,22,45,56,57,59,6064,6668,
70,72). These studies are reviewed below.
Early Studies (Pre-1960)
Early studies on a limited number of obese men and
women indicate individuals consuming low-CHO diets
Table 4. Diet composition of subjects who self-
selected low-CHO diets
Study
Total
kcals
CHO Fat Protein
g%g%g%
Evans (11) 1490 86 24 94 56 75 20
Yudkin (12) 1383 43 12 96 62 80 23
Rickman (56) 1325 7 1 73 50 160 48
Larosa (57) 1461 6 1.6 108 66 107 29
Table 5a. Effect of low-carbohydrate intake on body weight in obese subjects in studies without a control group(s)
Study n Duration CHO (g) kcal/day
Weight change
(kg)
Weight change
(g/day)
Kekwick (24) 14 5–9 days 10 1000 N/A N/A
Rickman (56) 12 7 days 7 1325 3.1 442
Benoit (58) 7 10 days 10 1000 6.6 660
Yudkin (12) 6 14 days 43 1383 2.8 200
Fletcher (59) 6 14 days 36 800 3.125 223
Lewis (60) 10 14 days 27 1115 5.2 371
Kasper (61) 16 16 days 56 1707 4.8 300
Bortz (62) 9 21 days 0 800 N/A N/A
Krehl (63) 2 30 days 12 1200 N/A N/A
Evans (11) 8 6 wk 80 1490 3.2 to 5.0 76 to 119
Golay (22) 22 6 wk 37.5 1000 8.0 111
Young (64) 3 6 wk 30 1800 16.18 385
Larosa (57) 24 12 wk 6 1461 6.8 0.91 81
Golay (21) 31 12 wk 75 1200 10.2 0.7 121
Cedarquist (65) 7 16 wk 85 1500 8.8 to 16.8 78 to 150
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
6S OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001
reduce overall caloric intake and lose weight
(12,65,72,73). Pennington’s (73) was one of the earliest
low-CHO diets, and contained less than 60 g CHO per
day, an amount “calculated to not interfere with ketogen-
esis.” The diet allows 24 ounces of meat with fat daily,
and one ordinary portion of any of the following: white
potatoes, sweet potatoes, boiled rice, half of a grapefruit,
grapes, melon, banana, pear, raspberries, or blueberries;
it allows no bread, flour, salt, sugar, or alcohol. The
Pennington diet resulted in an unspecified amount of
weight loss but critics were suspicious that the unpalat-
ability, or high satiety value of the diet, resulted in food
intake well below the minimum recommended 2870
kcal/d. However, Pennington concluded, “there is noth-
ing remarkable in the observation that some obese must,
of necessity, lose weight on an intake of 3000 kcal or
more per day,” considering their normal intake to be up
to 4500 kcal/d (72,73).
To substantiate weight loss could occur on 2870 kcal/d,
regardless of diet composition, Werner (72) studied 6 obese
subjects confined to a metabolic ward for 35 to 49 days. He
fed them Pennington’s low-CHO, high-fat diet (2874 kcal,
52 g CHO, 242 g fat) or an isocaloric, high-CHO, lower-fat
diet (2878 kcal, 287 g CHO, 146 g fat). Apart from transient
changes in water balance, the rate of weight loss in obese
subjects was the same on both diets, showing diet compo-
sition did not matter. Atkins (48) called Pennington’s study
“exciting” (p. 67) yet he dismisses Werner’s study as too
high in CHOs to promote ketosis (p. 70), despite the fact
that Werner received the diet from Dr. Pennington.
To support the concept of total caloric intake over diet
composition, Yudkin and Carey (12), studied six adult over-
weight subjects and found that when they followed a low-
CHO diet (30 to 55 g/d) for 2 weeks, caloric intake was
reduced 13% to 55% (180 to 1920 fewer daily calories).
Caloric intake averaged 1383 per day. Although all subjects
were allowed to consume an “unlimited” amount of fat,
none consumed significantly more fat than before, and three
showed a significant reduction of fat intake. Only one
showed a slight increase in protein intake.
Studies by Kekwick and Pawan (24,71) are cited by
Atkins to support his contention that diet composition,
rather than caloric intake, is the key variable for weight loss.
Yet, despite this contention, these studies support the notion
that calories do count. Obese individuals confined to a
metabolic ward were given diets with the same ratio of fat,
protein and CHO, but different caloric values. Individuals
lost more weight when they consumed lower calorie diets
(e.g., 500 and 1000 kcal/d) compared with when they con-
sumed higher calorie diets (e.g., 1500 and 2000 kcal/d). In
another study, 14 obese patients were fed 1000-kcal diets
containing either 90% protein (5 g of CHO), 90% fat (10 g
of CHO), or 90% CHO (225 g of CHO). Food available in
each of the diets was unspecified. Each subject consumed
the high-fat, high-protein, or high-CHO diet for 5 to 9 days
before being switched to another diet. Twenty-one days
later, all patients had lost weight, regardless of the order
they had consumed the different diets. However, patients
consuming 90% fat lost the most weight over 5 to 9 days,
whereas those eating 90% CHO lost little or none; some
even gained back some weight lost earlier on the 90% fat or
90% protein diets. These results led Kekwick and Pawan to
suggest, “obese patients must alter their metabolism in
response to the contents of the diet.” In another study, they
fed five obese individuals 2000 kcal balanced diets for 7
days, followed by a low-CHO, high-fat, high-protein diet
providing 2600 kcal/d for 4 to 14 days. Although patients
could maintain or gain weight on 2000 kcal/d, all, except
one, lost weight on 2600 kcal/d. Weight loss was reported to
be partly from body water (30% to 50%) and partly from
body fat (50% to 70%). Unfortunately, none of these studies
reported actual food intake, despite the author’s remarks,
“the main hazard was that many of these patients had
inadequate personalities. At worst they would cheat and lie,
obtaining food from visitors, from trolleys touring the
wards, and from neighboring patients.”
Convinced that fluid balance, not diet composition, was
the cause of the weight loss reported by Kekwick and
Pawan, Pilkington et al. (74) repeated their studies for
longer periods of time (18 or 24 days). His results were
Table 5b. Effect of low-CHO intake on body weight in obese subjects in studies with a control group(s)
Study n Duration
CHO
(g) kcal/day
Weight change
(kg)
Weight change
(g/day)
Worthington (66) 20 21 days 17 1182 12.0 3.7 571
Rabast (67) 13 25 days 48 1871 8.76 0.74 350
Rabast (68) 25 30 days 25 1000 11.77 0.77 392
Wing (69) 11 4 wk 10 800 8.1 270
Alford (45) 11 10 wk 75 1200 6.4 7.59 91
Baron (70) 66 3 months 50 1000 5.0 55
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001 7S
comparable with Kekwick and Pawan’s during the first few
days on each of the diets. However, there was a steady rate
of weight loss with each of the 1000-kcal diets thereafter,
regardless of whether the calories came from fat, protein, or
CHO. Although he did not measure fluid balance, Pilking-
ton (74) concluded that temporary differences in weight loss
were due to such changes. He stated “if the periods of study
are long enough to achieve a ‘steady state’ the rate of weight
loss on a diet consisting mainly of fat does not differ
significantly from the rate of weight loss on an isocaloric
diet consisting mainly of CHO.” Oleson and Quaade’s
(75) experiment, which lasted for 3 weeks, had a similar
conclusion.
Studies from 1961 to 1979
Fletcher et al. (59) gave six obese women who were
confined to a metabolic ward 800 kcal/d diets containing
mostly CHO, protein, or fat. They received each diet for 14
days. The high-fat and high-protein diets each contained
36 g of CHO. Statistical analysis showed no significant
difference in the rate of weight loss on the different diets.
Kinsell et al. (19) maintained obese subjects on a fixed
caloric intake and varied the macronutrient composition of
the diet (e.g., fat intake varied from 12% to 80%, protein
from 14% to 26%, and CHO from 3% to 61%). In any given
subject, the rate of weight loss after the initial depletion of
fluid was essentially constant throughout the entire study,
irrespective of diet composition. Bortz (62) fed an 800-kcal
liquid formula diet containing 80 g of protein, and either
54 g of fat (no CHO), or 120 g of CHO (no fat) to nine obese
subjects who were confined to a metabolic ward. Each diet
was given for 24 days, before switching to the other. No
difference in rate of weight loss was noted, apart from that
attributable to alterations in sodium and fluid balance. Krehl
et al. (63) studied four healthy, normal weight male prison
volunteers, and seven obese females (five were from 15 to
21 years old, and two were 36 and 53 years old) on a
metabolic ward. The obese females were given 1200-kcal,
12-g CHO diets, comprised of fat and protein in different
ratios (50/50; 60/40; 40/60; 70/30; 30/70). They received
each diet for 1 month. They also had three 1-hour periods
of supervised physical activity daily. Although it is difficult
to draw any conclusions from this small study, Krehl et al.
(63) reported that all patients lost weight at a rate commen-
surate with caloric restriction and physical activity, regard-
less of diet composition.
In another short-term study, Worthington and Taylor (66)
fed isocaloric diets (1182 kcal/d) for 2 weeks to 20 obese
women who were confined to a state correctional institution.
One diet was a low-CHO, ketogenic diet (17 g/d) with a
6:48:44 ratio of CHO to protein to fat calories. The other
was a “balanced low-calorie diet” and contained 96 g of
CHO and a 32:20:47 ratio of CHO to protein to fat. Al-
though this diet was not meant to be ketogenic, two subjects
tested positive for urinary ketones on Day 7 and four tested
positive on Day 14. The 10 women on the low-CHO diet
lost significantly more weight at the end of 14 days com-
pared with the 10 women on the balanced diet (12.0 3.7
vs. 8.7 3.5, low-CHO vs. balanced). The difference in
total weight loss was established primarily during the first
week, when the average weight loss in the low-CHO group
was 8.2 pounds, and that of the control group was 6 pounds.
During the second week, weight loss was similar for the
two groups.
In 1971, Young et al. (64), at Cornell University, looked
at the effect of diet composition on weight loss and body
composition. Eight moderately obese young male college
students were fed isocaloric diets for 9 weeks (interrupted
after 3 weeks for 1 week of spring vacation). Each diet
contained 1800 kcal and 112 g of protein, but different
amounts of CHO: either 104, 60, or 30 g/d. Physical activity
was not controlled. Only those in the 30 g/d group tested
positive for ketones throughout the 9-week study. As CHO
in the diet decreased, weight and fat loss slightly, but not
significantly, increased. Using underwater weighing to de-
termine body composition, Young et al. reported that the
weight lost by the lowest CHO group (30 g/d) was close to
100% fat. However, no difference between the groups with
respect to nitrogen, sodium, or potassium balances was
reported. Young et al. (64) concluded, “it would seem that
of the low CHO diets used, the one at the 104-g level would
be most suitable for long-term use.” Although their study
lasted 9 weeks, Atkins extrapolated data to 30 weeks, im-
plying even greater benefit (p. 73).
Rickman (56) monitored weight changes in 12 healthy
volunteers (hospital employees) who were no more than
10% above ideal body weight (based on Metropolitan Life
Insurance tables). Subjects were instructed to follow the
Stillman diet, which allowed unlimited quantities of protein
and fat, but no CHO. Average caloric intake was 1325 per
day, with 50% of calories from fat (73 g), 48% from protein
(160 g), and less than 1% from CHO (7 g). Subjects fol-
lowed the diet for 3 to 17 days (average 7.6 days). During
the first 3 to 5 days, each subject lost 1.3 to 2.2 kg. At the
end, mean weight loss was 3.1 kg. In 8 of 10 subjects for
whom there was follow-up within 7 days of the diet, average
weight regain was 2 kg (range, 1 to 4.5 kg).
Studies using low-CHO, liquid formula diets conducted
in Germany had small sample sizes, short duration (1
month), and poor design (61,67,68). Kasper et al. (61)
compared the weight loss of 16 obese subjects on low-CHO
diets (56g/d) with 4 obese subjects on isocaloric (1707
kcal), high-CHO diets (156 g/d). The average duration on
the low-CHO diet was 16 days (range, 6 to 30 days); mean
weight loss was 0.3 kg/d. The average duration on the
high-CHO diet was 10 days (range, 6 to 14 days); mean
weight loss was 0.05 kg/d. The small sample size, difference
in study duration, and fact that 3 of the 4 subjects on the
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
8S OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001
high-CHO diet also received the low-CHO diet (before or
after?) prevents adequate interpretation. Seventeen subjects
(some of whom had received other diets) were fed a high-fat
formula diet containing 2150 kcal and 112 g of CHO per
day. The average length of time on this diet was 18 days
(range, 6 to 40 days) and the mean weight loss was 0.32
kg/d, an amount comparable with the low-CHO, lower
calorie (1707) diet. Body composition was not measured
during any of these studies.
A similar, but better controlled study was conducted by
Lewis et al. (60). They compared the responses to two
cholesterol-free, isocaloric (10 kcal/kg per day; 1115
kcal), liquid formula diets of differing composition (70%
CHO, 20% protein, 10% fat vs. 70% fat, 20% protein,
10% CHO) in 10 obese men who were confined to a
metabolic ward. Diets were administered for 14 days in
random order and each diet was preceded by a 7-day
control, weight-maintenance diet (30 kcal/kg per day,
40% CHO, 20% protein, 40% fat). Although the low-
CHO diet was clearly ketogenic, Lewis et al. (60) con-
cluded that both low-calorie diets effected similar losses
of nonaqueous body weight. Their conclusions regarding
body composition changes were not based on actual body
composition measurements. Instead, they were based on
the significant rebound in body weight and the significant
urinary sodium retention observed when the weight
maintenance diet followed the ketogenic diet, along with
the significant increase in serum albumin concentration
noted during the period in which the low-CHO diet was
ingested. These changes were not seen when the mainte-
nance diet followed the high-CHO diet.
However, to support that low-CHO diets result in loss
of body fat, Atkins cites Benoit et al. (58), who compared
the effects of 10 days of fasting with a 1000-calorie, 10 g
of CHO ketogenic diet in seven active-duty Naval per-
sonnel (mean weight, 115.6 kg). Over the 10-day period,
the mean weight loss for the fasting and ketogenic groups
were 9.6 kg and 6.6 kg, respectively. The ketogenic diet
resembled fasting in terms of ketosis, acidosis, and mild
anorexia (which the authors speculated may influence
caloric restriction by the patient). However, the ketogenic
diet resulted in greater fat loss (97% vs. 35%) and de-
creased loss of lean body mass (3% vs. 65%) relative to
fasting. Although all patients on both diets were in neg-
ative N balance, potassium balance seemed unaffected by
the ketogenic diet, an impossibility according to Grande
(76), who seriously questioned the scientific validity of
Benoit’s entire study.
Atkins cites Rabast et al. (67,68) to support his contention
that low-CHO diets result in greater weight loss than high-
CHO diets. Rabast et al. (66,68) fed 45 obese German men
and women 1000-calorie, isonitrogenous, low-CHO (25
g/d) or high-CHO (170 g/d) formula diets. The duration of
the treatment period differed between the two groups. On
the low-CHO diet, it averaged 38 19 days (range, 15 to 78
days). On the high-CHO diet, it averaged 32 13 days
(range, 18 to 59 days). Due to significant drop out in both
groups, data were analyzed only up to Day 30. Results
indicate by Day 15, the 25 subjects following the low-CHO
diet lost significantly more weight than the 20 subjects
following the high-CHO diet (6.81 0.30 kg vs. 5.49
0.37 kg). There was no significant difference in weight loss
between the groups at Day 20 or 25. By Day 30, the weight
loss between the two groups again reached statistical sig-
nificance (11.77 0.77 kg vs. 9.81 0.43 kg, low-CHO
vs. high-CHO, respectively), even though by day 30, almost
40% of subjects in each group had dropped out (no reasons
given). Body composition data were not presented, and the
authors did not report any increased water or electrolyte
excretions during either of the diets. In another article,
Rabast et al. (67) presented the exact same data found in the
article just described (68). In addition, it included new data
from 28 additional subjects who received low-CHO (48
g/d, n 13) or high-CHO (355 g/d, n 15) liquid
formula diets containing 1900 kcal/d for 25 days. In this
study, all subjects lost weight, regardless of caloric intake or
diet composition.
The Rabast study that Atkins cites (p. 74) in support of
his position actually refutes it. This study confirms
weight loss on low-calorie diets, independent of CHO
content after Day 10 on 1900 kcal, and after Day 15 on
1000 kcal. Atkins cites the difference of 4.2 kg (9.24
pounds) in total weight loss between the 1000-calorie
low-CHO and 1000-calorie high-CHO groups as proof
that the low-CHO diet works better. The problem with
this is that these data (e.g., the 4.2-kg weight difference)
represent the final weight loss between the two groups at
the end of the study (59 to 78 days). However, we have
no idea how many subjects actually completed the study.
We do know that of 45 persons who started the study,
only 28 remained by Day 30.
Studies after 1980
Larosa (57) studied 24 obese free-living men and women
for 12 weeks. For the first 2 weeks, they followed their
current diet. For the next 4 weeks, they were instructed to
follow Stage I of the study diet, taken from the book, Dr.
Atkins’ Diet Revolution (47). Stage I is devoid of CHOs but
places no caloric limits on protein or fat. Based on urinary
ketone measurement all but 3 were confirmed as restricting
CHOs. After 4 weeks on Stage I, patients advanced to Stage
II, which allows 5 to8gofCHOperdayforanadditional
4 weeks, bringing the total time on the low-CHO diet to 8
weeks. The final 2 weeks (off the diet) allowed ad libitum
intake. No prescription for changes in exercise was given
and subjects were asked not to alter their exercise habits
from prestudy levels.
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001 9S
Results indicated that all but 2 of 24 subjects lost weight
the first 2 weeks of the study while eating ad libitum. After
8 weeks on the low-CHO diet, all subjects (except for 1
male) lost weight. Mean weight loss was 4.1 0.64 kg from
the ad libitum period, and 7.7 0.73 kg from the pre-diet
period (10 weeks before). Almost half of the total weight
loss occurred in the first 2 weeks on the low-CHO diet.
When subjects resumed ad libitum food intake at the end of
the 8-week diet period, some weight was gained back
(1.5 0.45 kg). However, data from 21 subjects showed
an overall significant loss of body weight (6.8 0.91 kg)
over the course of the 12-week study. One year later, weight
data were available from 62% of subjects. Although almost
all had gained back some of the weight they had previously
lost while on the low-CHO diet, only 2 subjects weighed
more than they had at the start of the study, whereas 13
weighed less (mean weight loss 5.9 1.7 kg). Body com-
position was not determined.
Results of this uncontrolled study support that low-
CHO diets lead to weight loss. Closer examination re-
veals weight loss results from caloric restriction. Diet
analysis (assessed using food intake records) revealed a
500-kcal decrease in total caloric intake from the start of
the study to the end of Stage II, 8 weeks later, when the
average intake was 1461 kcal/d. Just as Yudkin and
Carey (12) reported 20 years earlier, when protein and fat
were permitted in unlimited quantities, subjects did not
greatly increase their intake of these nutrients. In fact, fat
intake decreased (5 g) and protein intake only slightly
increased (11 g). The greatest caloric effect was the near
total elimination of CHO (165 g).
Alford et al. (45) manipulated CHO content of low-
calorie diets (1200 kcal/d) to determine possible effects
on body weight and body fat reduction over 10 weeks. At
least 11 women in each diet group consumed either a
low-, medium-, or high-CHO diet. The low-CHO diet
was 15% to 25% CHO (75 g/d) (30% protein, 45% fat),
the moderate-CHO diet was 45% CHO (10% protein,
35% fat), and the high-CHO diet was 75% CHO (15%
protein, 10% fat). The women were free-living, but at-
tended weekly classes on nutrition and behavior modifi-
cation. All were sedentary and agreed to remain so for the
duration of the study. Weight loss occurred in all groups,
but there was no significant difference in weight loss
among the groups. Percent body fat loss, based on un-
derwater weighing was similar among the groups. Alford
et al. (45) concluded, “there is no statistically significant
effect derived in an overweight adult female population
from manipulation of percentage of CHO in a 1200-kcal
diet. Weight loss is the result of reduction in caloric
intake in proportion to caloric requirements.”
Baron et al. (70) conducted a three-month randomized
controlled trial to determine acceptability of different sets of
dietary advice (e.g., low-CHO vs. low-fat) among free-
living subjects. Participants included 135 men and women
ranging from barely overweight to frankly obese, recruited
with the help of six diet clubs in Oxford, England. Within
each participating diet club, subjects were randomly given a
low-CHO diet (50 g/d) or a low-fat/high-fiber diet (30 g
fat/d). All diets contained 1000 kcal/d. Each subject planned
his/her own menus, with the assistance of group leaders and
study investigators, and received appropriate dietary in-
struction. Moderate weight loss occurred in both groups
during the 3-month period, although at 1 year, much of this
was regained. Body weight changes at 3 months indicated
that those following the low-CHO diet, especially women,
lost more weight than those following the low-fat/high-
CHO diet (5.0 vs. 3.7 kg, low-CHO vs. high-CHO). How-
ever, further analysis consistently showed club membership
(e.g., nature of participants in each club, or effectiveness of
leaders) to be a better predictor of weight loss than compo-
sition of diet.
Golay et al. (21,22) studied the effect of varying levels
of CHO intake (15%, 25%, and 45%) on weight loss in
obese subjects. In one study, 68 outpatients followed for
12 weeks received a low-calorie (1200 kcal), 25% CHO
(75 g), or 45% CHO diet (21). Protein content of the diets
was comparable (30%); fat made up the difference.
After 12 weeks, the mean weight loss was similar be-
tween the two groups (10.2 0.7 kg vs. 8.6 0.8 kg;
25% vs. 45% CHO, respectively). Loss of adipose tissue
was similar. Despite a high protein intake (1.4 g/kg IBW)
there was a loss of lean body mass in both groups. The
waist-to-hip ratio diminished significantly and identically
in both groups. In another study (22), 43 obese inpatients
followed for 6 weeks received a low-calorie diet (1000
kcal), and participated in a structured, multidisciplinary
program that included physical activity (2 h/d), nutri-
tional education, and behavioral modification. The natu-
ral food diet contained either 15% CHO (37.5 g), or 45%
CHO. Protein content of the diets was comparable
(30%); fat made up the difference. After 6 weeks, there
was no significant difference in weight loss in response
to either diet (8.9 0.6 kg vs. 7.5 0.5 kg; 15% vs. 45%
CHO, respectively). Significant and comparable de-
creases in total body fat and waist-to-hip ratios were seen
in both groups. Both studies show that energy intake, not
diet composition determines weight loss and fat loss in
response to low-energy diets over a short time period.
Wing et al. (69) confined 21 severely obese women to a
metabolic ward for 31 days. They were randomly assigned
to ketogenic (10 g of CHO) or nonketogenic liquid formula
diets containing 600 kcal/d for 28 days. Weight losses
were comparable between the two diets (mean, 8.1 kg).
Because the objective was to determine whether ketogenic
weight reducing diets have adverse effects on cognitive
performance, no data on body composition were obtained.
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
10S OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001
One might argue that because low-CHO diets result in
decreased caloric intake, these diets offer an advantage. If
subjects lose weight on these diets, or even gain some
weight back when the diet ends (57), these diets might still
be of long-term benefit. Astrup and Ro¨ssner (77) concludes
that a greater initial weight loss improves long-term main-
tenance, so long as the weight loss is followed by 1 to 2
years of an integrated weight maintenance program consist-
ing of dietary change, behavior modification, and increased
physical activity.
Body Composition Changes
During the early days of a ketogenic diet, weight loss
is partly due to water loss (25,55,78). In contrast, during
the early days on a mixed diet, weight loss is primarily
due to loss of body fat (23). After several weeks, subjects
who stay on a ketogenic diet regain water equilibrium
(25). Because they restrict calories, low-CHO diets result
in loss of body fat if the diets are maintained for a longer
period of time. A 4.5% reduction in body fat was reported
in individuals consuming low-CHO diets for 10 weeks
(45). Golay et al. (21,22) reported significant body fat
reduction (16.8% to 21.6%) in obese subjects consuming
15%, 25%, or 45% CHO isocaloric diets for 6 and 12
weeks. Losses of protein and fat are about the same
during a ketogenic diet as during an isocaloric, nonketo-
genic diet (21,22,25).
Table 6. Dr. Atkins’ New Diet Revolution: diet analysis compared with the USDA Food Guide Pyramid
Nutrient
Atkins’
induction
Atkins’
ongoing
Atkins’
maintenance
Food guide
pyramid
RDAs, DRVs,
DRIs*
Total energy (calories) 1152 1627 1990 1972 2000–2200
Moisture (H
2
O), g 682 736 1132 1879 none
Total fat, g (% total kcal) 75 (59) 105 (58) 114 (52) 54 (24) 65 (30)
Saturated fat, g 29 49 44 17 20
Monounsaturated fat, g 31 36 41 19 20
Polyunsaturated fat, g 6 11 19 15 20
Cholesterol (mg) 753 1115 955 154 300
Total protein, g (% total kcal) 102 (35) 134 (33) 125 (25) 90 (18) 75 (15)
Total CHO, g (% total kcal) 13 (5) 35 (8.6) 95 (19) 292 (59) 55%–60%
Alcohol, g 0 0 14 0 moderation
Dietary fiber (g) 38 13 22 20–35
Vitamin E (mg) 37 10 40 15
Vitamin A (RE) 669 2183 2231 4140 700
Thiamin (mg) .5 1.4 .7 3.8 1.1
Riboflavin (mg) 1.3 2.5 2.0 4.3 1.1
Niacin (mg) 18 20 25 51 14
Vitamin B
6
(mg) 1.2 1.8 2.2 5.5 1.3
Folate (
g) 135 391 282 1010 400
Vitamin B
12
(
g) 8 8 4.3 17 2.4
Vitamin C (mg) 67 95 226 288 75
Calcium (mg) 294 1701 889 1749 1000
Phosphorus (mg) 1096 1993 1418 1800 700
Magnesium (mg) 126 294 233 425 320
Iron (mg) 10.4 12.6 8.7 39 18
Zinc (mg) 15 14 11.7 31 8
Sodium (mg) 2934 4046 3604 2757 2400
Potassium (mg) 1734 2562 3339 4718 3500
RDAs, Recommended Dietary Allowances; DRVs, Dietary Reference Values; DRIs, Dietary Reference Intakes.
Note: Items in bold indicate values different from RDAs, DRVs, and DRIs.
* RDAs and DRIs used are those of a female, 31–50 years old. Calculated values (DRV) are based on a 2000-kcal diet: 30% total calories
from fat, 10% of total calories from saturated, monounsaturated, and polyunsaturated fat, and 15% total calories from protein.
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001 11S
Evidence Statement: In the short-term, low-CHO
ketogenic diets cause a greater loss of body water
than body fat. Water weight is regained when the diet
ends. If the diet is maintained long-term, it results in
loss of body fat. Evidence Category C.
In conclusion, calories count, and low-CHO diets fail to
confer a metabolic advantage with respect to body weight or
body composition.
2. Nutritional Analysis
What is the nutritional profile of high-fat, low-CHO
diets?
Do these diets provide adequate levels of nutrients,
based on current dietary recommendations?
Tables 6 and 7 are nutritional analyses of 1-day menus of
popular diets (presented in Tables 8 and 9), and a diet based
on the USDA Food Guide Pyramid. Menus came from
books (48, pp. 338–340; 49, pp. 147–164; 50, pp. 209–217)
and a representative diet based on the Food Guide Pyramid.
Table 7. Nutrition analysis of various diets: Carbohydrate Addict’s, Sugar Busters!, Weight-Watchers, and Ornish
Diets
Nutrient
Carbohydrate
Addict’s diet
Sugar
Busters!
Weight Watchers
diet
Ornish
diet
RDAs, DRVs,
DRIs*
Total calories 1476 1521 1462 1273 2000–2200
Moisture (H
2
O), g 746 1696 1200 1993 none
Total fat, g (% total kcal) 89 (54) 44 (26) 42 (25) 13 (9) 65 (30)
Saturated fat, g 35 11 9 2 20
Monounsaturated fat, g 31 20 18 3 20
Polyunsaturated fat, g 15 9 9 5 20
Cholesterol (mg) 853 128 116 4 300
Total protein, g (% total kcal) 84 (23) 89 (23) 73 (20) 48 (15) 75 (15)
Total CHO, g (% total kcal) 87 (24) 176 (46) 207 (56) 258 (81) 55–60%
Alcohol, g 0 14 0 2 moderation
Dietary fiber (g) 8 25 26 38 20–35
Vitamin E (mg) 77 29 7 15
Vitamin A (RE) 3039 948 5638 2318 700
Thiamin (mg) .8 2.4 3.0 1.8 1.1
Riboflavin (mg) 1.8 1.7 3.6 1.5 1.1
Niacin (mg) 16.4 32 37 17 14
Vitamin B
6
(mg) 1.8 2.6 4.0 2.5 1.3
Folate (
g) 176 377 636 615 400
Vitamin B
12
(
g) 6.5 3.4 11.6 1.0 2.4
Vitamin C (mg) 53 109 207 380 75
Calcium (mg) 640 712 1147 1053 1000
Phosphorus (mg) 1150 1510 1432 1181 700
Magnesium (mg) 173 400 325 477 320
Iron (mg) 8.2 20 28 24 18
Zinc (mg) 11 11 23 8 8
Sodium (mg) 3192 4012 2243 3358 2400
Potassium (mg) 2479 3020 3773 4026 3500
RDAs, Recommended Dietary Allowances; DRVs, Dietary Reference Values; DRIs, Dietary Reference Intakes.
Note: Items in bold indicate values different from RDAs, DRVs, and DRIs.
* RDAs and DRIs used are those of a female, 31–50 years old. Calculated values (DRV) are based on a 2000 kcal diet based on 30% total
calories from fat, 10% of total calories from saturated, monounsaturated, and polyunsaturated fat, and 15% total calories from protein.
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
12S OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001
Diets are compared with current Recommended Dietary
Allowances (RDAs) and Dietary Reference Intakes (DRIs).
All food records were analyzed using the USDA 1994 to
1998 Continuing Survey Nutrient Database.
Analyses reveal high-fat, low-CHO diets are also low in
calories (e.g., 1152 to 1627 kcal/d). The Atkins’ Maintenance
Diet, to be followed after weight loss, provides 1990 kcal/d.
Low-CHO diets are high in fat, especially saturated
fat, and cholesterol. They are also high in protein (mainly
animal), and provide lower than recommended intakes of
vitamin E, vitamin A, thiamin, vitamin B
6
, folate, cal-
cium, magnesium, iron, potassium and dietary fiber.
Evidence Statement: High-fat, low-CHO diets are
nutritionally inadequate, and require supplementa-
tion. Evidence Category C.
Low-CHO diets are often referred to as high-protein or
high-fat diets because of the high percentage of calories from
protein (25% to 30%) and fat (55% to 60%). Because overall
caloric intake decreases on low-CHO diets, and consumption
of protein and fat is self-limiting (11), the absolute amount of
protein and fat is not as high as these percentages imply.
However, the absolute amount of these nutrients are higher in
low-CHO as compared with the typical American diet (105 g
vs. 82.5 g of protein and 94 g vs. 85 g of fat, low-CHO vs.
American diet, respectively) (Table 3). When low-CHO diets
are compared with moderate-fat, balanced nutrient reduction
diets, they provide twice as much protein and 2.4 times more
fat at the same caloric level.
3. Metabolic and Adverse Effects
What are the metabolic effects of high-fat, low-CHO
diets?
Will these diets correct the complications of diabetes,
heart disease, and high blood pressure?
What effects, if any, do these diets have on bone health,
cancer risk, and renal function?
Are there any adverse effects when consuming these
diets?
Table 8. Dr. Atkins’ New Diet Revolution: menu items compared to the USDA Food Guide Pyramid
Meal Atkins’ induction Atkins’ ongoing Atkins’ maintenance Food Guide Pyramid
Breakfast 2 scrambled eggs
2 strips bacon
Decaffeinated coffee
3 egg Western omelet
(with milk, butter,
peppers, onions, ham)
3 oz. tomato juice
2 CHO g bran crisp
bread
2 egg spinach & cheese
omelet
2 CHO g bran crisp
bread, 1 T butter
1/2 cantaloupe
1 C orange juice
1 C Total cereal with 3/4
C skim milk
Coffee with 1 oz. 1%
milk
Snack 6 oz. apple juice
Lunch Bacon (1 slice)
cheeseburger (4 oz; 1
oz cheese)
Small salad (no dressing)
Seltzer water
Chef’s salad with 1 hard-
boiled egg, 2 oz. ham, 1
oz. cheese, 2 oz. chicken
Iced tea
4 oz. roast chicken
2/3 C broccoli
Green salad with creamy
Italian dressing
Turkey sandwich (3 oz.
meat, 1 T mayonnaise,
tomato)
10 baby carrots
1 C milk (1%)
Snack 1 C deep-fried pork
rinds
10 saltine crackers
(low-salt)
6 oz. V-8 (no-salt added)
Dinner Clear consomme´
1.5 C shrimp salad
Steak (4 oz)
Salad with dressing
1 C Sugarless Jell-O
with 1 T whipped
sugar-free cream
3 oz. poached salmon
3/4 C spinach
1/2 C strawberries with
1 T heavy whipping
cream
Salad w/tomatoes,
onions, carrots
1 C green beans
1/2 small baked potato
w/sour cream, chives
5 oz. loin of veal
1 C fresh fruit salad
5 oz. white wine
3 oz. Atlantic salmon
1/2 C rice
1/2 C zucchini
w/parmesan cheese
1 slice whole wheat bread
with 1 T canola margarine
Snack 4 oz. Swiss cheese, 3
slices of bacon, fried
6 gingersnaps, 1 banana
1/2 C chocolate ice cream
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001 13S
A number of different metabolic effects have been re-
ported for high-fat, low-CHO diets. The most common is
ketosis, as measured by increased urinary ketones
(24,57,58,60,63,69,79). Ketogenic diets usually have less
than 20% calories from CHOs (80). Because many of these
are also low calorie, average CHO intake is 50 to 100 g/d.
All popular low-CHO diets recommend 100 g of CHO per
day. Ketogenic diets may cause a significant increase in
blood uric acid concentration (57,60,63,67,78).
Other metabolic effects range from decreased blood
glucose and insulin levels, to altered blood lipid levels
(Table 10). Many of these effects (e.g., decreased LDL
and HDL cholesterol) may be the consequence of weight
loss, rather than diet composition, especially considering
that the absolute amount of fat consumed on the low-
CHO diet may be similar to that consumed before the diet
(Table 3).
Evidence Statement: High-fat, low-CHO diets re-
sult in ketosis. Evidence Category B.
Evidence Statement: Low-CHO diets that result
in weight loss may also result in decreased blood lipid
levels, decreased blood glucose and insulin levels,
and decreased blood pressure. Evidence Category C.
Possible effects of such high saturated fat diets on endo-
thelial dysfunction need to be assessed. It has been proposed
that a single high-fat meal transiently impairs conduit vessel
endothelial function (81). However, this hypothesis has
been recently challenged (82).
If excess weight causes complications of diabetes, heart
disease, and high blood pressure, then individuals who lose
weight on low-CHO diets, and maintain weight loss, may
see health benefits. However, no data support long-term
adherence to such diets, and high-fat, low-CHO diets con-
tradict all governmental and nongovernmental dietary rec-
ommendations with respect to reducing risk, or treating
such conditions.
Bone Health, Cancer Risk, and Renal Function
The potential effect of low-CHO diets on bone health
is an important consideration. In a study of diet and
osteoporosis, Wachman and Bernstein (83) hypothesized
the role of the skeleton in acid-base homeostasis in
adults, observing a reservoir of alkaline salts of calcium
as key to the regulation of pH and plasma bicarbonate
concentrations (see also 8486). New (87), after review-
ing other studies showing that acidification increases the
activity of osteoclasts and inhibits that of osteoblasts,
concluded that a diet high in meat but low in fruits and
vegetables could lead to bone loss. Barzel and Massey
(88) concluded that excessive dietary protein from foods
with high potential renal acid load leads to calciuria,
which adversely affects bone, unless buffered by the
consumption of alkali-rich foods (e.g., fruits and vegeta-
bles). Recently, New et al. (89) found that potassium,
magnesium, fiber,
-carotene and vitamin C, and a high
intake of fruit was important to bone health. Low-CHO
diets, often providing inadequate amounts of these nutri-
ents (and foods) may pose long-term risks to the skeleton.
The effect of high protein intake on renal function during
weight loss induced by high- (25%) vs. low-protein (12%),
moderate-fat (30%) diets in overweight subjects over 6
months was assessed by Skov et al. (90). Protein intake in
the low-protein group decreased from 91 to 70 g/d, and
increased from 91 to 108 g/d in the high protein group.
Results indicate moderate changes in dietary protein intake
caused adaptive alterations in renal size and function with-
out indications of adverse effects. However, CHO content
of diet was not restricted (e.g., 45% or 58%) so this study
did not directly speak to the issue of a high-protein, high-fat,
and low-CHO diet. For further information on metabolic
consequences of high-protein intake see Metges and
Barth (91).
Finally, low-CHO diets are often low in fruits, vegeta-
bles, and dietary fiber. This raises the specter of increased
cancer risk if such diets are consumed long-term (92–95).
However, because no long-term consumption data exist, it is
currently impossible to assess cancer risk in individuals
consuming low-CHO diets.
Adverse Effects
Few clinically significant adverse effects have been re-
ported in subjects consuming high-fat, low-CHO diets.
Some reported side effects include bad taste in mouth
(57), constipation (70), diarrhea (49,56,72), dizziness (66),
halitosis (57), headache (66), insomnia (49), nausea
(56,66,74), thirst (57), and tiredness, weakness, or fatigue
(49,56,57,64,74).
Only one study assessed cognitive effects of low-CHO
diets (69). Performance on attention tasks did not differ as
a function of diet. Performance on the trail making task, a
neuropsychological test that requires higher order mental
processing and flexibility, was adversely affected by the
ketogenic diet. Worsening of performance was observed
primarily between baseline and Week 1 of the diet.
4. Hunger and Appetite: Compliance, CHO Cravings,
and Addiction
Do low-CHO ketogenic diets decrease hunger?
What data support compliance to a low-CHO diet?
Are CHOs addictive?
Dietary adherence is one of the most difficult challenges
faced by obese dieters (54). The stronger the feeling of
hunger, the greater the urge to break the diet. If diet com-
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
14S OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001
position affects feelings of hunger, it may influence the
ability of patients to adhere to the weight-loss regimen.
Atkins claims the low-CHO diet is a revolution because no
hunger is experienced (48, pp. 112–113). Individuals are
allowed to eat as much protein and fat as they desire as long
as they avoid CHOs. Atkins believes this combination of
nutrients has a high satiety value and results in individuals
eating less (and losing weight). In studies lasting up to 16
weeks, data indicate subjects consuming low-CHO diets
decrease food intake and lose weight (Tables 5a,b). Young
et al. (64) found each of the low-CHO levels (30 g to 104
g/d) effective in controlling hunger, and that hunger was not
a problem after the first week. Cedarquist et al. (65) wrote
“subjects had a feeling of well-being and satisfaction. Hun-
ger between meals was not a problem.” Krehl et al. (63)
reported the highest level of satiation on a 12-g CHO diet
with a 70:30 ratio of fat to protein compared with diets
having 60:40, 50:50; 40:60; or 30:70 ratios. (Note: this
70:30 ratio is close to the Atkins’ ongoing weight loss
phase, which has a ratio of fat to protein of 60:30.).
Not all studies support these findings. Baron et al. (70)
found that low-CHO dieters complained of hunger with the
Table 9. Menu items of various diets: Carbohydrate Addict’s Diet, Sugar Busters!, Weight Watchers, and Ornish
Diets
Meal Carbohydrate Addict’s Sugar Busters! Weight Watchers diet Ornish diet
Breakfast 3 egg onion-cheese (1
oz. cheese) omelet
(made with whole milk
and margarine)
2 sausage links
Coffee or tea
3/4 C grapefruit juice
1 pkg instant oatmeal,
2/3 C skim milk
Coffee
1 oz. Total Corn flakes,
1/2 C non-fat milk
1 slice whole wheat
bread, 1 pat margarine
orange
1/2 grapefruit
1 package oatmeal, 1 oz.
raisins
1 C skim milk
Brewed tea
Snack None allowed 3 rye crispbread with 1
T peanut butter
1 apple apple
Lunch 1/2 C water-packed tuna
salad (mayonnaise,
scallion and eggs)
1 C salad
Turkey (3 oz.) sandwich
on whole wheat bread
with mustard, lettuce,
tomato
Diet cola
2 oz. roast beef on rye
bread
2 raw carrots
tossed green salad, low-
calorie French dressing
1 cup non-fat milk
10 grapes
1 corn tortilla, 2 T salsa,
1/2 C black beans,
1/2 C canned tomatoes,
onions, 1/4 C green
peas
1 C salad, 1/4 cantaloupe
Snack None allowed Apple 1 ounce almonds
1 fig bar
Dinner 3 oz. steak
Baked potato with sour
cream and chives
1.5 C salad with 1 T
buttermilk dressing, 1
raw carrot
1/2 butter pecan ice
cream
4 oz. pork tenderloin
broiled with chopped
onion
1/2 C brown rice made
with fat-free chicken
broth
1/2 C green beans
1 C salad
5 oz. red wine
1 C beef bouillon soup;
2 saltines
2.5 oz. salmon, broiled
3/4 C zucchini
1/2 baked potato
1 C brown rice with
1/4 C tofu, stir-fry
vegetables (1/2 C
broccoli, 1/8 C cabbage,
3 scallions, 1/8 C bean
sprouts, 1/8 C peppers,
1/4 C snow peas, 1/8 C
carrots); teriyaki sauce,
2 oz. cooking wine, 1/4
t sesame seeds, 1/4 C
pineapple
1 C salad with no-oil
salad dressing
1 orange
Snack None allowed 12 nuts (mixed) 1/2 C chocolate ice
cream
1/2 C non-fat milk
1/2 C strawberries
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001 15S
same frequency as low-fat dieters. Worthington (66) re-
ported no difference in acceptability, appetite, or satiety
after 2 weeks on either low-CHO or balanced diets, and
ketosis did not suppress appetite. Rosen et al. (96,97) found
no support for the idea that a minimal-CHO, protein-sup-
plemented fast (800 kcal; 58% protein, and 42% fat) de-
creases appetite and elevates mood in comparison with an
isocaloric CHO-containing diet that minimized ketosis.
Thus, the effect of low-CHO diets on hunger and satiety
remains controversial.
Compliance
Although compliance was not directly assessed, some
data indirectly apply to this issue. Kekwick and Pawan (24)
fed patients low-calorie diets containing either 90% calories
from fat, protein, or CHOs. They noted, “Many of these
patients had inadequate personalities. At worst they would
cheat and lie, obtaining food from visitors, from trolleys
touring the wards, and from neighboring patients. (Some
required almost complete isolation). A few found the diet so
trying they could not eat the whole of their meals. When this
happened, the rejected part was weighed, and the equivalent
calories and foodstuffs were added to a meal later in the day.
A considerable number of failures in discipline were dis-
carded.” Rabast et al. (67,68), who studied subjects on a
metabolic ward receiving low-calorie, low-CHO liquid for-
mula diets, reported that after 30 days, “conditions for
comparative investigations were no longer met because the
two groups were declining rapidly.” No explanation for
dropouts was given.
Most studies on low-CHO diets (or of subjects receiving
advice to consume low-CHO diets) were of short duration
and had small sample sizes (Tables 5a,b). Of studies pub-
lished over the last 44 years, those that lasted 9 weeks or
longer included a total of 76 subjects (21,45,57,64,65).
Are CHOs Addictive?
Some authors state that “CHOs are addictive” (50,51).
Furthermore, they speculate that hyperinsulinemia prevents
a rise in brain serotonin, leading the CHO craver to feel
hungry and eat more CHOs. This vicious cycle of hunger,
CHO craving, CHO consumption, and hyperinsulinemia is
proposed to be the underlying cause of obesity (50,51).
Some confuse the matter further by stating, “certain people
have a natural, overwhelming desire for CHO that doesn’t
correlate to hunger. These people in all likelihood have a
genetic predisposition toward CHO craving . . . which can
be reduced for some by embarking on a low-CHO diet” (51,
p. 118). The latter suggests that a change in dietary com-
position will override a purported genetic defect. Research
has not substantiated any of these contentions.
Wurtman (98) characterized self-selected, obese, “CHO
cravers” by their powerful and frequent cravings for and
consumption of foods rich in CHO over those high in
protein, especially during the afternoon snacking period.
This snacking among obese CHO-cravers represents a vari-
able that contributes to excess caloric intake (and weight),
and became the basis for The Carbohydrate Addict’s Diet
(50). This diet limits daily food intake to two “Complemen-
tary” high-fiber, low-fat, low-CHO meals (how is that pos-
Table 10. Reported metabolic effects of low-CHO, ketogenic diets
Clinical measure Increased Decreased No change
Blood uric acid 57,60,63,67 11
Blood glucose 21,22,60,67 24
Blood insulin 21,22,60
Blood glucagon 60
Glucose tolerance 60,79
SGOT 67
SGPT 67
Serum albumin 60
Blood urea nitrogen 45 79
Sodium balance 62 63,64
Potassium balance 64 63
Blood cholesterol 56 11,57
HDL cholesterol 21,57 [women only], 60 22,57 [women only], 60
LDL cholesterol 57 [women only] 60
Blood TG 21,45,57 [men only], 60,61,63,67 11,56
Blood pressure 22,67
Blank cells indicate no published data; numbers refer to studies cited.
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
16S OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001
sible?) and one “Reward Meal” of unlimited quantity or
quality, consumed within 60 consecutive minutes. No
snacks are allowed. The authors claim that eating “Comple-
mentary Meals” fools the body into producing less insulin,
relative to what it would have produced if CHOs were
consumed at each meal. The claim that insulin output will
be low no matter what is consumed at the “Reward Meal” so
long as it is limited to 1 hour (50, p. 96) is unsubstantiated
(and if true, potentially dangerous). This diet works simply
because eliminating snacks and after-dinner eating results in
decreased caloric intake.
If CHO cravings were due to decreased serotonin, then
drugs that increase serotonergic output should alleviate
cravings and result in decreased food intake. Early studies
with the serotonergic drug fenfluramine showed effective-
ness in decreasing CHO intake (99). However, the effect
was not limited to CHO; it resulted in decreased intake of
protein as well (100). Toornlivet et al. (101,102) demon-
strated that obese CHO cravers and non-CHO cravers re-
sponded similarly to treatment with fenfluramine with re-
spect to eating behavior and weight loss. Although the
evidence may be interpreted to provide support for the
existence of a self-medication effect among a large segment
of obese individuals, the mechanism by which CHO medi-
ates this effect has not yet been identified. Furthermore, a
more likely interpretation is that some people simply have
an unusually large appetite (i.e., they are cravers).
Drewnowski (103) has pointed out that the so-called
“sweet-tooth” characterizing CHO cravers is just as much a
“fat-tooth” because the foods typically selected are high in
both CHO (often sugar) and fat. Thus, the effect of low-
CHO diets on hunger, appetite, and satiety need further
study.
4a. Role of Insulin in Obesity
Is overproduction of insulin, driven by high CHO in-
take, the cause of the metabolic imbalance that under-
lies obesity?
If so, can obese, hyperinsulinemic individuals lose
more weight on low-CHO diets as compared with
high-CHO diets?
Does leptin interact with insulin in regulation of appe-
tite and body weight?
Dietary CHO, as well as dietary protein, increases insulin
secretion. The hyperinsulinemia of obesity may be the result
of dietary factors, genetic factors (e.g., “thrifty genotype”)
or secondary adaptation to insulin resistance (31). Increased
appetite and consequent overconsumption may drive in-
creased insulin, but as body weight increases, and insulin
resistance develops, this too will drive increased
insulin secretion.
The relationship between insulin resistance and weight
gain yield conflicting results (104). Swinburn et al. (105)
and Schwartz et al. (106) indicated that insulin resistance
and hyperinsulinemia predicted decreased weight gain over
3 years in glucose-tolerant adult Pima Indians. In contrast,
Sigal et al. (107) reported hyperinsulinemia predicts in-
creased weight gain. However, this study was questioned on
the basis of subject sampling and methodology (33). Even if
hyperinsulinemia is the cause of the metabolic imbalance, is
there evidence to show that low-CHO diets are better for
weight loss than high-CHO diets?
Energy restriction, independent of diet composition (e.g.,
15% to 73% CHO) improves glycemic control (21,31–33).
The ability to lose weight on a calorically restricted diet
over a short-time period does not vary in obese healthy
women as a function of insulin resistance or hyperinsulin-
emia (104). Although diet composition may play a role in
absolute reduction in blood insulin levels, weight loss seems
to be independent of such changes. For example, Golay et
al. (21) reported subjects consuming isocaloric diets (1000
kcal) containing 15% CHO had significantly lower insulin
levels as compared with those consuming 45% CHO, yet
there was no difference in weight loss between the two
groups. In another study, isocaloric diets (1200 kcal) con-
taining 25% and 45% CHO resulted in similar reductions
in blood insulin levels as well as similar average weight
losses (22).
Grey and Kipnis (31) studied 10 obese patients who were
fed hypocaloric (1500 kcal/d) liquid-formula diets contain-
ing either 72% or 0% CHO for 4 weeks before switching to
the other diet. A significant reduction in basal plasma insu-
lin levels was noted when subjects ingested the hypocaloric
formula devoid of CHO. Refeeding the hypocaloric, high-
CHO formula resulted in a marked increase in the basal
plasma insulin. However, patients lost 0.75 to 2.0 kg/wk
irrespective of caloric distribution.
The effect of protein vs. CHO on blood insulin levels and
subsequent weight loss was assessed by Baba et al. (32),
who studied 13 male obese hyperinsulinemic subjects for 4
weeks. They were fed a hypoenergetic diet (comprised of
80% of the person’s resting energy expenditure) containing
either 25% CHO and 45% protein, or 58% CHO, and 12%
protein. Both diets contained 30% calories from fat. Despite
the significant, but not different degrees of reductions in
blood insulin levels that occurred on both diets, the insulin
levels were reduced to within the normal range only in the
high-protein group. Although individuals in both groups
lost weight, the mean weight loss was significantly higher
on the high-protein as compared with the high-CHO diet, a
consequence, perhaps, of the higher protein, lower CHO
content of the diet.
The optimal macronutrient composition of a weight-re-
ducing regimen in obese hyperinsulinemic patients is the
subject of research, but beyond the scope of this article (for
more information see Reaven et al. (108)).
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001 17S
Insulin and Leptin in the Endocrine Regulation of
Appetite and Body Weight
Insulin and leptin are hormones that act as medium- to
long-term regulators of body weight through their actions to
decrease food intake and increase energy expenditure (met-
abolic rate), ensuring that energy intake and energy expen-
diture is closely matched (109–111).
People who do not produce leptin due to a genetic defi-
ciency, or who have defects in the leptin receptor, have
dramatically increased appetites and overeat to the point of
becoming massively obese (112,113). The effects of leptin
deficiency are ameliorated by the administration of recom-
binant leptin (114).
Insulin, in addition to its effects in the central nervous
system to inhibit food intake, acts in the periphery to ensure
the efficient storage of incoming nutrients. The role for
insulin in the synthesis and storage of fat has obscured its
important effects in the central nervous system, where it acts
to prevent weight gain, and has led to the misconception that
insulin causes obesity (115). It has recently been shown that
selective genetic disruption of insulin signaling in the brain
leads to increased food intake and obesity in animals
(116) demonstrating that intact insulin signaling in the
central nervous system is required for normal body weight
regulation.
Insulin also has an indirect role in body weight regulation
through the stimulation of leptin (117). Both insulin and
leptin are transported into the central nervous system, where
they may interact with a number of hypothalamic neuropep-
tides known to affect food intake and body weight (118).
Insulin and leptin are released and circulate in the blood-
stream at levels that are proportionate to body fat content.
Secretion and circulating levels are also influenced by
amount and type of foods eaten, with decreased concentra-
tions noted during fasting or energy-restricted diets
(119,120). The decrease of leptin during a prolonged ener-
gy-restricted diet has been shown to be related to increased
sensations of hunger (120) suggesting a role for low leptin
levels to increase appetite during dieting in humans, and
therefore to the predisposition for weight regain after ini-
tially successful dieting.
Circulating concentrations of both insulin and leptin,
measured over a 24-hour period, are reduced in women
consuming high-fat meals (60% fat, 20% CHO) compared
with when equicaloric meals high in CHO and low in fat
(60% CHO, 20% fat) are consumed (36,37). Increased in-
sulin secretion has been suggested to protect against weight
gain in humans (106). Because insulin also stimulates leptin
production, which acts centrally to reduce energy intake and
increase energy expenditure, decreased insulin and leptin
production during the consumption of high-fat diets could
help contribute to the obesity promoting effects of dietary
fat (42,44,121).
Recent studies show consuming a high-fat diet induces
resistance to the actions of leptin to decrease food intake
(122,123), and that increased energy intake and weight
gain is related to an impairment of insulin transport into
the brain (124). Therefore, dietary macronutrient compo-
sition affects not only production of insulin and leptin but
also may influence their ability to gain access to the brain
to signal target neurons. In studies investigating the ef-
ficacy and long-term consequences of weight loss diets, it
is important to consider the effects of dietary macronu-
trient content and composition on the production of in-
sulin and leptin, and their actions to regulate energy
intake and expenditure.
5. Performance and Physical Activity
Does the low-CHO diet affect physical performance?
Although reference is made to physical activity and
exercise by proponents of low-CHO diets (48, pp. 260
267; 49, pp. 187–206; 50, pp.143–144), only one study
examined the capacity for moderate exercise in obese
subjects after adaptation to a hypocaloric, ketogenic diet.
This study was conducted in six slightly to moderately
overweight, untrained subjects on a protein-supple-
mented fast for 6 weeks (e.g., 500 to 750 kcal/d, 10 g
CHO, weight loss, 10.6 kg). Results indicate that subjects
adapt to prolonged ketosis and use lipid, rather than
CHO, as the major metabolic fuel during prolonged ex-
ercise at 60% of maximum oxygen concentration. This
shift was confirmed by an respiratory quotient of 0.66
during exercise (125).
Other studies were conducted in physically untrained,
but normal weight males who were fed eucaloric low-
CHO (20 g/d), high-fat (80%) ketogenic diets, or non-
ketogenic, low-, medium-, or high-fat diets (15%, 30%,
or 55% fat) (126,127). They report diet manipulation, per
se, did not effect maximal or submaximal aerobic per-
formance in untrained individuals. However, one cannot
extrapolate results from these studies to typically un-
trained, sedentary, overweight individual consuming
low-calorie, low-CHO diets.
II. Moderate-Fat, Balanced Nutrient
Reduction Diets
Moderate-fat, balanced nutrient reduction diets contain
20% to 30% fat, 15% to 20% protein, and 55% to 60%
CHO. Popular diets in this category include those promoted
by commercial weight loss centers (e.g., Weight Watchers,
Jenny Craig, Nutri-Systems). These diets have a long his-
tory of use, millions of followers worldwide, and are typi-
cally based on sound, scientific principles. The DASH diet,
diets based on the USDA Food Guide Pyramid, and the
National Cholesterol Education Program Step I and Step II
Diets also fit into this category if calories are reduced (128).
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
18S OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001
However, most consumers (and governmental agencies) do
not perceive the latter as “popular diets,” probably because
they are typically promoted for reasons other than weight
loss and because they are not promoted as commercial diets.
For example, the DASH diet (35) has been shown to reduce
hypertension, and the Food Guide Pyramid, in conjunction
with the Dietary Guidelines, provides recommendations for
healthful eating. Although the main focus of the National
Cholesterol Education Program Step I and Step II diets is
the reduction of saturated fat and cholesterol, these diets
also promote weight control because obesity contributes to
high blood cholesterol in many patients (128).
Balanced nutrient reduction diets (moderate-fat, high-
CHO) used for weight loss have been studied extensively
(6,18,42).
Table 11. Relation between dietary fat intake (20% to 30% kcal) and weight change in overweight subjects
Study n BMI
Duration
(weeks)
Fat g
(%)
Energy
start
(kcal)
Energy
end
(kcal)
Energy
change
(kcal)
Weight
change
(kg)
Weight
change
(g/day)
Buzzard (132) 17 28.6 13 35 (23) 1840 1365 475 2.8 31
Carmichael (133) 39 35 0–17 40 (22) 2177 1608 (569) 3.0 25
17–26 46 (25) 2177 1658 (519) 2.6 21
Golay (22) 21 38 6 30 (26) N/A 1027 N/A 7.0 167
Golay (21) 37 33 12 34 (26) N/A 1178 N/A 9.0 100
Hammer (134) 4 37 16 37 (23) 1934 1450 (484) 5.8 52
Harvey-Berino (40) 29 30 24 45 (27) 2171 1477 (694) 11.8 70
28 24 38 (21) 1929 1650 (279) 5.2 31
Insull (135) 184 68.8 26 31 (21) 1734 1316 (418) 3.16 17
173 52 31 (22) 1734 1299 (435) 2.93 8
Henderson (136) 163 104 34 (23) 1734 1355 (379) 1.91 3
Jeffery (137) 39 31 26 36 (23) 1506 1125 (391) 4.6 25
52 40 (26) 1506 1112 (394) 2.1 6
78 40 (26) 1506 1199 (307) 0.4 1
Knopp (131) 57 28 52 63 (28) 2395 2019 (356) 2 5
55 27 52 58 (26) 2294 1995 (299) 2 5
62 28 52 52 (25) 2281 1917 (364) 6 16
Pascale (138) 15 N/A 16 30 (22) 1658 1201 (457) 7.7 69
16 16 31 (22) 2024 1246 (778) 7.5 67
Powell (20) 8 20% 12 26 (20) 1642 1163 (479) 5.0 60
9 above IBW 12 37 (28) 2081 1190 (891) 4.6 55
Prewitt (139) 6 38 1–4 70 (37) 1894 2047 153 1.5 53
5–20 58 (21) 2047 2428 381 0.6 5
Puska (140) 35 28% body fat 6 51 (23) 2490 2001 (489) 0.7 17
Rumpler (141) 4 28% body fat 4 34 (20) 3095 1542 (1553) 5.0 178
Shah (142) 47 31 26 37 (21) 1893 1580 (313) 4.4 24
Skov (143) 23 30 0–13 80 (29) N/A 2533 N/A 5 54
13–26 86 (29) 2533 2676 143* 5 27
Swinburn (144) 49 30 52 52 (26) 2195 1832 (369) 3.1 8
N/A, Not available.
Adapted from Bray G and Popkin BM (42).
* Weight loss occurred from 0–13 weeks, and no further weight was lost from 13–26 weeks. Caloric intake was not significantly different
during the two time periods. Actual caloric intake is suspect based on methodology.
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001 19S
A. Underlying Philosophy, Claims, and Proposed
Solutions
The underlying philosophy of moderate-fat, balanced nutri-
ent reduction diets is that weight loss occurs when the body is
in negative energy balance. Diets are calculated to provide a
deficit of between 500 to 1000 kcal/d, but a minimum number
of daily calories (e.g., 1000 to 1200 for women, 1200 to 1400
for men) are recommended. Increased energy expenditure
(e.g., physical activity) is promoted as well.
The goal of moderate-fat, balanced nutrient reduction diets
is to provide the greatest range of food choices to the con-
sumer, to allow for nutritional adequacy and compliance, while
still resulting in a slow but steady rate of weight loss (e.g., 1 to
2 lbs/wk). Programs are usually based on up-to-date, scientific
information. For example, recently Weight Watchers started a
new program called the 10% difference, based on scientific
findings that a 10% reduction of body weight improves health
(e.g., decreases blood pressure, improves lipid profile, etc). A
similar program is the Shape Up and Drop 10! program of
Shape Up America! (129).
B. Scientific Evaluation of Claims
1. Caloric Intake, Body Weight, and Body Composition
What is the effect of balanced nutrient reduction diets
on body weight and composition?
Which is better for weight loss: reducing calories or
reducing fat?
The NHLBI evidence report Clinical Guidelines on the Iden-
tification, Evaluation, and Treatment of Overweight and Obesity
in Adults reviewed the effects of dietary therapy on weight loss
(6). Key evidence statements from that report follow.
Evidence Statement: Low-calorie diets* can re-
duce total body weight by an average of 8% over 3 to
12 months. Evidence Category A.
Evidence Statement: Although lower-fat diets
without targeted caloric reduction help promote
weight loss by producing a reduced calorie intake,
lowerfat diets coupled with total caloric reduction
produce greater weight loss than lower-fat diets
alone. Evidence Category A.
The NHLBI concludes, “there is little evidence that low-
er-fat diets cause weight loss independent of caloric restric-
tion.” This is supported by Powell et al. (20), who studied
obese subjects consuming isocaloric reduced calorie diets
(1200 kcal/d) containing either 15%, 20%, 28%, or 34%
calories from fat for 12 weeks. All subjects lost body weight
and body fat. However, there were no significant differ-
ences in the rate or amount of body weight or percent body
fat lost across the four groups during the intervention.
Bray and Popkin (42) and Astrup et al. (43,121) reviewed
the relationship between dietary fat intake and weight
change extensively. Based on a meta-analysis of normal
weight and overweight subjects, Bray and Popkin con-
cluded a 10% reduction in the percentage of energy from fat
would reduce body weight by 16 g/d†. Table 11 summarizes
* Low-calorie diets contain 1000 to 1200 kcal/d.
† The meta-analyses by both Bray and Astrup included normal weight and overweight
individuals.
Table 12. Change in body weight, overall caloric intake, and diet composition in studies that evaluated the efficacy
of low-fat, ad libitum diets (LFAL) vs. low-energy (LE) diets for weight loss*
Shah (142) BW (kg) Calories Fat g (%) CHO g (%) Protein g (%)
LFAL 4.4 1580 37 (21) 245 (62) 60 (16)
LE 3.8 1550 34 (30) 206 (54) 60 (16)
Jeffery (137) BW (kg) Calories Fat g (%) CHO g (%) Protein g (%)
LFAL 1.9 1350 40 (26) N/A N/A
LE .5 1414 50 (32) N/A N/A
Schlundt (145) BW (kg) Calories Fat g (%) CHO g (%) Protein g (%)
LFAL 4.6 1425 30 (19) 210 (58) 64 (18)
LE 8.3 1265 28 (19) 179 (58) 61 (18)
Harvey-Berino (40) BW (kg) Calories Fat g (%) CHO g (%) Protein g (%)
LFAL 5.2 1650 38 (20) 251 (61) 69 (16)
LE 11.8 1477 45 (27) 200 (54) 60 (16)
N/A, Not available.
* Bold indicates the type of diet (in each study) that was more effective for weight loss. No difference between diets in the Shah paper.
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
20S OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001
the relationship between dietary fat intake and weight
change in overweight subjects consuming 20% to 30% fat
diets. Clearly, subjects who reduce fat intake reduce overall
caloric intake, and lose weight. Despite these data, Willett
(130) argues that the relationship between dietary fat and
obesity is unconvincing. In support, Knopp et al. (131) is
cited. In this study, moderately overweight hypercholester-
olemic men were randomly assigned to one of four levels of
dietary fat for 1 year. Despite differences in actual fat intake
in the seven diet groups (ranging from 22% to 28% fat), all
groups lost an equivalent amount of body weight (2 to 3 kg).
Willett (130) argues that if fat matters, then those consum-
ing 22% fat should lose more weight than those consuming
28% fat.
The retort may be “reduction in fat alone is not enough,
calories matter too!” (see Calories vs. Fat below). Second,
the reduction of fat to 27% may have accounted for the
majority of weight loss effects and that further reductions
provide minimal additional benefit (e.g., there may exist a
threshold of dietary fat below which changes in body weight
are minimal) (44). In reality, the study by Knopp et al. (131)
supports the relationship between fat intake and body
weight. When fat decreases from 34% to 36% to less than
30%, caloric intake significantly decreases and results in
significant body weight reduction. This occurs in all groups,
but is most pronounced in overweight subjects (BMI, 27)
who decreased fat intake from 34% to 25%, and who lost 6
kg over the course of a year (16 g/d), as Bray and Popkin
Table 13. Effect of low-fat ad libitum diets (LFAL) vs. low-energy diets (LE) on caloric intake, diet composition
and change in body weight
Diet
BW
(kg)
Calories Fat CHO Protein
Start End ⌬⌬g % g % g %
Shah (142): Data from 6 months.
Conclusion: No difference between groups at 6 months.
LFAL 4.4 1893 1580 313 39 13 1 11 13 0.7
LE 3.8 2119 1550 569 24 4 49 3.5 15 0.6
Diet
BW
(kg)
Calories Fat CHO Protein
Start End ⌬⌬g % g % g %
Jeffery* (137): Data from 12 months.
Conclusion: Low-fat better at 12 months; no difference at 18 months.
LFAL 2.1 1735 1350 385 31 11 N/A N/A N/A N/A
LE .5 1774 1414 360 20 4 N/A N/A N/A N/A
Diet
BW
(kg)
Calories Fat CHO Protein
Start End ⌬⌬g % g % g %
Schlundt (145): Data from 16–20 weeks.
Conclusion: Low-energy better at 16–20 weeks, no difference at 9–12 months
LFAL 4.6 2200 1425 775 64 19 19 17 20 3
LE 8.3 2000 1265 735 59 20 12 19 38 2
Diet
BW
(kg)
Calories Fat CHO Protein
Start End ⌬⌬g % g % g %
Harvey-Berino (40): Data from 24 weeks.
Conclusion: Low-energy better at 24 weeks. No long term follow-up available.
LFAL 5.2 1929 1650 279 26 9.2 3 8.3 9 4.2
LE 11.8 2170 1477 693 33 5.1 75 3.6 23 1.1
N/A, Not available.
* Data was taken as the average of intakes reported by food frequency and food recalls.
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001 21S
(42) predicted. That there was not a significant difference in
total caloric intake and weight loss at levels of fat between
22% to 28% does not mean that reducing fat below 30%
kcal does not make a difference in body weight; it does.
Calories vs. Fat
Which diet is better for weight loss: a low-fat ad libitum
(LFAL) or a low-energy (LE) diet? At present, data are insuf-
ficient to say that one is better: weight loss occurs on both.
Tables 12 and 13 are from studies in which subjects
consumed either LFAL or LE diets containing 19% to 30%
calories from fat (40,137,142,145). Both diets result in
reduction of total caloric intake from fat, CHO and protein,
and both result in weight loss independent of diet compo-
sition. Studies reveal percent fat does not always correlate
with calories, so that fat content of a diet, in and of itself,
may not determine weight loss (131). Just because a diet is
lower in fat does not mean it is lower in calories (40), and
two diets may be the same percent fat, but differ in caloric
content, resulting in differences in weight loss (145). Fi-
nally, studies show that what predicts short-term weight loss
may be different from what predicts long-term weight main-
tenance.
Body Composition
The NHLBI concludes that low-calorie diets resulting in
weight loss decrease body fat. Subjects who consume mod-
erate-fat, balanced nutrient reduction diets lose body fat,
and decrease waist and hip circumferences. The change in
lean body mass is inconsistent (Table 14).
2. Nutritional Analysis
What is the nutritional profile of balanced-nutrient
reduction diets?
Do diets provide adequate levels of nutrients, based on
current dietary recommendations?
The nutritional profile of a balanced-nutrient reduction
diet (e.g., “Weight Watchers”) is presented in Table 7.
When proper food choices are made, these diets are nutri-
tionally adequate (128, 146).
Evidence Statement: With proper food choices
based on the USDA Food Guide Pyramid the mod-
erate-fat, balanced nutrient reduction diet is nutrition-
ally adequate. Evidence Category B.
However, if appropriate food choices are not made, diets
may fall short with respect to calcium, zinc, magnesium,
iron, vitamin B
12
and dietary fiber (132,135,144). Swinburn
et al. (144) studied the effect of reduced fat ad libitum diets
over a 1-year period in 110 New Zealand adults. Nutrient
intakes were derived from 3-day food diaries at the begin-
ning and end of the study. Blood levels of retinol,
-tocoph-
erol, and
-carotene were determined at the end of the
study. Results show that reduced fat intake (from 35% to
25% of energy) led to reduced calorie intake and weight loss
compared with a usual diet (33% fat). There were no dif-
ferences between the diets in changes in micronutrient in-
takes, except for an energy-adjusted increase in
-carotene
intake in the reduced-fat group. Fiber intake remained at 20
g/d. However, the calcium intake in both groups was low
(600 mg/d).
Buzzard et al. (132) indicated that intake of zinc and
magnesium was significantly reduced in women with stage
II breast cancer consuming reduced fat (22%) diets. Cal-
cium intake (628 mg/d) was similar to that reported by
Swinburn et al. (144), but this amount is less than two-thirds
the recommended intake. Fiber intake (15 g/d) was also low.
Insull et al. (135) assessed dietary intake among women
consuming low-fat vs. control diets (22% vs. 37% fat,
respectively). Despite similarities in dietary intakes between
Table 14. Effect of moderate-fat diets on changes in body composition
Study Duration Body fat (% change) LBM (kg) Waist circumference Hip circumference
Golay (22) 6 weeks 2 17 N/A 22
Golay (21) 12 weeks 2 5 N/A 22
Siggaard (39) 12 weeks 2 4.2 .25 N/A N/A
Prewitt (139) 20 weeks 2 2.3 8 N/A N/A
Carmichael (133) 6 months N/A N/A 22
Shah (142) 6 months 2 2.2 1.2 N/A N/A
Skov (143) 6 months Yes (4 kg)* N/A Yes N/A
N/A, Not available.
* % change not available.
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
22S OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001
groups, calcium, iron, and vitamin B
12
intake was slightly
below daily requirements, necessitating the use of dietary
supplements.
3. Metabolic and Adverse Effects
What are the metabolic effects of moderate-fat, bal-
anced nutrient reduction diets?
Studies allowing ad libitum intake of reduced-fat, high-
CHO diets show the changes in blood lipids are dominated
by the slight weight loss induced by such diets (43). This
is confirmed by the systematic review of effects of the
National Cholesterol Education Step I and II dietary inter-
vention programs on cardiovascular disease risk factors.
Meta-analysis reveals these diets reduce LDL-chol-
esterol, normalize plasma TGs, and normalize the ratio of
HDL/TC (27).
In studies reviewed here, Golay et al. (21,22) reported
diets containing 26% fat (and either 1000 or 1200 kcal)
resulted in reduced TC, HDL cholesterol, and TGs. Skov et
al. (143) reported diets containing 29% fat (but 2600 kcal)
reduced TC and HDL but increased TGs. Theusen et al.
(147) reported subjects consuming 21% fat diets containing
1835 to 2026 kcal for 1 year had decreased levels of total
and LDL cholesterol, no change in HDL cholesterol, and
decreased TGs. Henderson et al. (136) measured changes in
plasma TC among intervention group women at 12 and 24
months, and reported TC decreased in the group consuming
low-fat diets (22% fat) but not in those consuming control
higher-fat diets (37% fat).
Moderate-fat, balanced nutrient reduction diets reduce
blood pressure (6,21,22,35).
Only Golay et al. (21,22) measured fasting insulin levels,
which were significantly reduced in subjects who lost
weight on balanced nutrient reduction diets. When moder-
ate-fat (20%) meals are consumed, postprandial insulin se-
cretion is enhanced and circulating leptin levels are in-
creased over a 9-hour (37) or 24-hour (36) period. These
changes of insulin secretion and leptin production could
contribute to the effects of balanced nutrient reduction
diets on energy intake, hunger and appetite, and energy
expenditure.
Evidence Statement: Moderate-fat, balanced nu-
trient reduction diets reduce LDL-cholesterol, nor-
malize plasma TGs, and normalize the ratio of HDL/
TC. Evidence Category A.
Evidence Statement: Moderate-fat, balanced nu-
trient reduction diets reduce blood pressure. Evidence
Category A.
4. Hunger and Appetite: Compliance
What is the effect of moderate-fat, balanced nutrient
reduction diets on hunger and appetite?
What data support compliance to these diets?
A number of studies in which subjects consumed
LFAL diets report that individuals do not complain of
hunger, but rather, that there is too much food. Siggaard
et al. (39) reported a high degree of satisfaction with the
changes in food intake in Danish workers consuming a
LFAL diet for 12 weeks. Subjects stated, “I have never
been eating as much as I do now” and “I have not felt
hungry at any time.” Shah et al. (142) reported subjects
consuming LFAL diets rated these higher than LE diets
in palatability, satiety, and quality of life measures. Jef-
fery et al. (137) found that subjects asked to reduce fat
were more compliant with treatment instructions, re-
ported greater success in reaching their dietary goals, and
rated their diet as higher in palatability. They had greater
decreases in binge eating scores than those in the LE
group.
In support of LE diets (over LFAL diets), Harvey-Berino
(40) found that subjects consuming LE diets had more
positive changes in eating behaviors, and greater improve-
ments in feelings of physical wellness that were not corre-
lated to weight loss. However, these diets were rated more
inconvenient than LFAL diets.
Subjects consuming both LE and LFAL diets reported an
increased distaste of fat. Subjects in both groups reported
increased feelings of deprivation, but the changes in the
group were not significant (40). Bray and Popkin (42)
conclude adherence to an LFAL diet is a function of the
frequency of dietary counseling. This may be the case for
any reduced-calorie diet, regardless of macronutrient com-
position. Alford et al. (45) remarked, “For the women in our
group, the interaction and support were the most important
aspects. I think that is true for most women. We tried
to make the nutrition classes personal, so the women
wouldn’t tune them out.” Perhaps psychological issues
are more important than dietary factors in the discussion
on compliance.
Finally, The Women’s Health Trial Vanguard Study
(135,136) examined the feasibility of a nationwide, ran-
domized multicenter intervention trial to test the hypoth-
esis that a low-fat diet followed for 10 years reduces
breast cancer risk. Feasibility studies of women at in-
creased risk show dietary intervention targeted to lower
dietary fat below 25% can be implemented and main-
tained successfully over a 2-year period. However, be-
cause women in these studies were highly selected and
motivated, caution is urged in extrapolating compliance
data to the “dieting” public, whose motivation for reduc-
ing fat may be very different from those in this study
group.
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001 23S
III. Low-Fat and VLF Diets
(<10% to 19% Fat)
VLF diets are defined as containing 10% fat, and low-
fat diets contain between 11% and 19% fat. Both are very
high in CHO, and moderate in protein. Representative VLF
diets are those promoted by Dr. Dean Ornish, and Nathan
and Robert Pritikin. There are no commercial diets that fit in
the category of low-fat diets, although there is research on
these diets meriting their inclusion in this paper.
The American consumer has traditionally ignored the
VLF diet for weight loss. It seems they would rather restrict
CHO intake to less than 10% of calories than to restrict
intake of their favorite fat (or sweet-fat combination) to the
same degree. In addition, these diets were not historically
promoted as weight loss diets, but rather, diets to prevent or
reverse heart disease. VLF diets recommended reduction of
dietary fat and cholesterol based on the (now) well-known
association between saturated fat consumption and cardio-
vascular disease.
However, as Americans became fatter, and their hunger
for diet books remained unsatiated, proponents of VLF diets
(e.g., Ornish and Pritikin) capitalized on their program’s
apparent effect on body weight. They changed the focus
(and title) of their books from heart disease to weight loss‡.
Neither trial (e.g., The Lifestyle Heart Trial or The Pritikin
Program was originally designed to assess the effect of diet
‡ This theory is supported by changing book titles over the past 20 years. Ornish’s book
titles include Stress, Diet and Your Heart (1982), Dr. Dean Ornish’s Program for Reversing
Heart Disease (1990), and Eat More, Weigh Less (1993). The Pritikin plan was originally
popularized by Nathan Pritikin, whose books included The Pritikin Program for Diet and
Exercise (1979) and The Pritikin Promise (1983). His son, Robert, head of the Pritikin
Longevity Center wrote The New Pritikin Program (1990), The Pritikin Weight Loss
Breakthrough (1998), and The Pritikin Principle (2000).
Table 15. Relation between dietary fat intake (10–19% kcal) and weight change in overweight subjects
Study n BMI Duration
Fat g
(%)
Energy
(kcal)
Weight change
(kg)
Weight change
(g/day)
Agus (152) 10 30.6 9 days 30 (18) 1493 3.23 358
Alford (45) 12 28 12 wk 13 (10) 1200 4.8 57
Barnard (155) 2643 30 3 wk 13 (10) 1200 5.1 242
1897 30 3 wk 13 (10) 1200 3.3 157
Djuric (150) 113 27.8 12 wk 35 (17) 1843 .9 10
57 (32) 1559 2.3 27
28 (18) 1392 3.6 42
Havel (34) 17 35.1 6 mo N/A (12) N/A 4.0 21
8 mo N/A (11) N/A 5.0 21
Heilbronn (33) 35 N/A 12 wk 18 (10) 1600 6.6 78
Kasim-Karakas (151) 54 27.6 6 mo 20 (12) 1449 2.5 13
8 mo 19 (11) 1503 3.1 13
10 mo 19 (12) 1420 3.3 11
12 mo 19 (12) 1474 4.6 13
Lissner (149) 12 101% 2 wk 40 (18) 2087 .4 28
MLI standards*
Noakes (28) 22 31 12 wk 17 (10) 1553 7.9 94
Ornish (156,14) 25 28.4 52 wk 13 (6) 1821 10.7 29
20 28.4 260 wk 17 (8) 1846 5.8 8
Powell (20) 8 20% above IBW 12 wk 19 (15) 1113 4.5 53
Schaefer (38) 27 (LFAL) 26.3 10–12 wk N/A (15) N/A 3.63 43
Schlundt (145) 27 (LFAL) 31 16–20 wk 30 (19) 1425 4.6 32
29 (LE) 35 16–20 wk 28 (19) 1264 8.3 59
Surwit (153) 20 36 6 wk 13 (11) 1087 7 166
22 35 6 wk 14 (11) 1156 7.4 176
N/A, Not available.
* MLI is Metropolitan Life Insurance Tables.
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
24S OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001
on weight loss. Thus, the scientific information on the effect
of these diets on body weight and body composition is
limited. In addition, caution in the interpretation of results is
necessary because diet is but one component of these total
lifestyle-modification plans.
A. Underlying Philosophy, Claims, and Proposed Solutions
Proponents of VLF diets support reducing caloric intake
and increasing energy expenditure to achieve weight loss.
Rather than counting calories, per se, dietary recommenda-
tions focus on “type of calories” (13, p. 31) and “caloric
density” (17, p. 5). “Eat more, weigh less” (13) does not
mean, “eat more calories.” It means, “consume more high
complex CHO, and high-fiber foods whenever you feel
hungry and until you feel full, but not stuffed” to lose
weight (13, p. 32). The “calorie density solution” enables
individuals to eat as much as they want—six to seven times
daily—and lose weight safely, gradually, and without hun-
ger (17, p. 17). These VLF diets are based primarily on
vegetables, fruits, whole grains, and beans, with moderate
quantities of egg whites, nonfat dairy or soy products, and
small amounts of sugar and white flour. Ornish’s diet is
vegetarian; Pritikin allows a limited amount of low-fat
animal protein daily (no more than 3.5 ounces of lean beef,
fowl or fish).
Dr. Dean Ornish’s Program for Reversing Heart Dis-
ease (148) and The Pritikin Program (15,16) promotes
lifelong changes in diet, exercise, and lifestyle. Each plan
includes a nutrition and exercise component; Ornish’s
plan includes stress reduction and emotional support as
well. Current claims range from weight loss, to overcom-
ing or reversing heart disease, reducing symptoms of type
2 diabetes, high blood pressure, cancer, arthritis, stress,
and smoking, in addition to general wellness, sometimes
in as little as 2 weeks (13,15,16,148). Pritikin also claims
medications for heart disease, diabetes, and high blood
pressure may be reduced or completely eliminated by
following these plans.
B. Scientific Evaluation of Claims
1. Caloric Intake, Body Weight, and Body Composition
Do low-fat and VLF diets result in decreased caloric
intake?
What is the effect of these diets on body weight and
body composition?
Overweight subjects who consume low-fat, and VLF,
high-CHO diets eat fewer calories and lose weight
(20,34,38,145,149,150–152) (Table 15). Again, total ca-
loric intake is more important than diet composition, in this
case fat, for weight loss (28,33,45,154). In the context of
reduced caloric intake (1200 kcal/d) percent calories de-
rived from fat (15% to 35%) does not influence weight loss
(20). Havel et al. (34) reports a family history of diabetes is
predictive of weight loss (and fat loss) in women consuming
LFAL diets for 6 months.
Subjects who lose weight on low-fat diets lose body fat
(120,34,45,151) and lean body mass (145). However, in the
context of a reduced calorie diet, the amount of dietary fat
(10% to 40%) does not affect losses of body fat or lean body
mass over 12 weeks (20).
Evidence Statement: Overweight subjects con-
suming low-fat, high-CHO diets eat fewer calories,
lose weight, and lose body fat. Evidence Category A.
Alford et al. (45) fed adult, sedentary, overweight women
reduced calorie diets (1,200 kcal/d) containing 10%, 35%,
or 45% calories from fat for 12 weeks. The 10% fat diet was
70% to 80% CHO. Weight loss was the same on each diet.
Noakes and Clifton (28) fed 62 overweight subjects (mean
BMI 31) one of three test diets for 12 weeks. One was
VLF (10% fat), high-CHO (71.6%), whereas the other two
were moderate-fat (31.8%), and moderate-CHO (50%).
Caloric intake on all diets was limited to 1533 kcal/d.
Overall weight loss was 8.6 0.4 kg (9.7%) with a reduc-
Table 16. Energy intake from 0–5 years (The Lifestyle Heart Trial)
Experimental Control p*
Baseline (B) 1 year 5 years Baseline 1 year 5 years B–1 year B–5 years
Energy intake 1950 1821 1846 1711 1673 1572 0.64 0.86
Energy change 129 104 38 139
Body weight (kg) 91.4 80.64 85.64 75.74 77.18 77.09 0.001 0.001
* All p values are two-tailed and each is the result of a test of the null hypothesis that the change between two particular visits (e.g. baseline
and 1 year) does not differ between the experimental and control groups (14). Adapted from Ornish (14,156).
Popular Diets: A Scientific Review, Freedman, King, and Kennedy
OBESITY RESEARCH Vol. 9 Suppl. 1 March 2001 25S
tion in waist circumference of 8%. There were no signifi-
cant differences in weight loss between diet groups, al-
though weight loss was least on the VLF as compared with
the moderate-fat diet.
Surwit et al. (153) conducted a 6-week weight-loss trial that
compared the efficacy of two hypoenergetic (1100 kcal/d),
VLF (11%), high-CHO diets (71%) varying in sucrose con-
tent. The high-sucrose diet contained 43% total energy from
sucrose; the low-sucrose diet only 4%. Subjects in both groups
lost comparable amounts of body weight and body fat. CHO
source had no effect on weight loss as long as energy was
restricted. These data clearly show that high sucrose or com-
plex CHO consumption does not cause obesity, hyperglycemia
or insulin resistance in the absence of dietary fat. Although it
is quite possible that sucrose or complex CHOs may produce
different effects when total energy intake is greater, the use of
sucrose or other CHOs in a low-fat, weight-reduction program
seems both safe and effective (providing a good refutation to
proponents of low-CHO diets).
Heilbronn et al. (33) studied 35 obese patients with type 2
diabetes assigned to one of three 1600 kcal/d diets for 12
weeks. The diets were VLF (10%), high CHO (72%), or high-
monosaturated or high-saturated fat (32%), lower CHO (50%).
Diet composition did not affect the magnitude of weight loss,
with subjects losing an average of 6.6 0.9 kg.
VLF diets
Do VLF diets, when consumed ad libitum, decrease ca-
loric intake? The answer is a qualified yes because most
studies of individuals consuming these types of diets were
not designed to assess the effect of diet on weight loss, but
rather the effects of lifestyle change (e.g., low-fat diet,
exercise, stress reduction) on disease risk or reversal. One
exception is the study of Shintani et al. (157) who fed 20
obese native Hawaiians a pre-Western contact traditional
Hawaiian diet low in fat (7%), high in complex CHOs
(78%), and moderate in protein (15%) for 21 days. Partic-
ipants were encouraged to eat to satiety. Average energy
intake decreased from 2594 to 1569 kcal/d and average
weight loss was 7.8 kg. Although interesting, this study is
not relevant outside Hawaii.
Other studies allowing ad libitum intake of VLF diets
were published by Barnard (using the Pritikin diet) and
Ornish. In the Barnard articles (155,158–166), the subjects
were 3-week residents of the Pritikin Longevity Center who
engaged in medically supervised daily aerobic exercise,
primarily walking on a treadmill, and consumed the Pritikin
high-complex CHO, high-fiber, low-fat, low-cholesterol,
and low-salt diet. All meals and food were provided onsite.
Barnard (155) reports 2643 males and 1897 females con-
suming VLF, high-fiber diets for 3 weeks lost 5.1 and 3.3
kg, respectively, representing a 5.5% and 4.4% decrease in
body weight (men and women, respectively.) Although
BMI of patients is not provided, average weights of 91.9 kg
and 74.8 kg (men and women) indicate program participants
were overweight.
Another 3-week study (162) assessed the role of diet and
exercise in management of hyperinsulinemia. Seventy-two
patients were divided into three groups based on fasting
serum glucose and