ArticlePDF Available

Confronting the Failure of Behavioral and Dietary Treatments for Obesity

  • Eating Attitudes LLC

Abstract and Figures

This review questions the appropriateness of behavioral and dietary treatments of obesity in light of overwhelming evidence that they are ineffective in producing lasting weight loss. The stigmatization of obesity, the overstatement of health risks, and the pervasive influence of the lucrative diet industry have maintained public demand for dietary treatment. However, decades of research on the biology of weight regulation make clear the unlikelihood of success with dietary treatment, information which the health professions have been slow to integrate. Recommendations are made for improving lifestyle, health risk factors, body image, and the self-esteem of the obese without requiring weight loss.
Content may be subject to copyright.
Clinical Psychology Rmew, Vol. 11, pp. 729-780, 1991
Printed in the USA. All rights reserved. 0272-7358/91 $3.00 + .oo
Copyright 0 1991 Pergamon Press plc
David M. Garner
Michigan State University
Susan C. Wooley
University of Cincinnati
ABSTRACT. This review questions the appropriateness of behavioral and dietar)r treat-
ments of obesity in light of overwhelming evidence that they are ineffective in producing
lasting weight loss. The stigmatization of obesity, the overstatement of health risks, and the
pervasive influence of the lucrative diet industry have maintained public demand for di-
etary treatment. However, decades of research on the biology of weight regulation make
clear the unlikelihood of success with dietary treatment, information which the health
professions have been slow to integrate. Recommendations are made for improving life-
style, health risk factors, body image, and the self-esteem of the obese without requim’ng
weight loss.
It has been over a decade since two major reviews questioned the effectiveness
and social appropriateness of behavioral treatments for obesity (Stunkard & Pen-
ick, 1979; Wooley, Wooley, & Dyrenforth, 1979b). Other papers and books have
since appeared challenging the basic precepts which underlie dietary treatments
for obesity (Bennett, 1984, 1987; Bennett 8c Gurin, 1982; Ernsberger 8c Haskew,
1987; Fitzgerald, 198 1; Foreyt, Goodrick, & Gotto, 1981; Krieshok & Karpowitz,
1988; Wooley & Wooley, 1982, 1984; Wooley, Wooley, & Dyrenforth, 1979a).
Their arguments, however, have not been embraced, accepted, or in many cases,
even addressed by the mainstream of behavioral scientists and health care pro-
Correspondence should be addressed to David M. Garner, Ph.D., Department of Psychia-
try, Michigan State University, B107 West Fee Hall, East Lansing, MI 48824.
730 D. M. Garner and S. C. Wooley
fessionals who treat obesity. It is still widely held within the health care profes-
sions that obesity confers significant health risks warranting weight reduction.
Weight loss as a means for achieving health and happiness is vigorously pro-
moted by the commercial weight loss industry, now a major economic force in
North America.
Behavioral or dietary treatments for mild and moderate obesity’ continue to
be advocated despite weak and often conflicting epidemiological data suggesting
that these levels of obesity are linked to significant health risks (cf. Ernsberger &
Haskew, 1987; Mann, 1974a, 1974b), and despite overwhelming evidence from
controlled studies that weight loss programs are ineffective in producing lasting
weight change (cf. Bennett, 1987; Stunkard & Penick, 1979). One can point to
behavioral programs for weight control recommended in the same publications
which document physiological resistances to weight change, seemingly without
recognition of the contradictions or problems involved in trying to override the
body’s biological regulatory mechanisms. Our failure to fully confront these is-
sues has meant that, despite new knowledge, there has been no fundamental
change in our practices. Old diets with new names, such as the “set point diet,”
seem almost to parody the efforts of scientists to understand the causes of obe-
sity and provide treatments consistent with this understanding.
Another expository review of obesity treatment studies is unnecessary since the
topic has been comprehensively reviewed elsewhere (Bennett, 1986; Brownell,
1982; Brownell & Wadden, 1986; Foreyt, 1977; Foreyt et al., 1981; Jeffery,
1987; Stunkard & Mahoney, 1976; Wilson & Bronwell, 1980). Instead, the pri-
mary aim of this paper is to provide an evaluative and integrative appraisal of
weight loss treatments within the context of what is known about (a) long-term
treatment efficacy, (b) the biology of weight regulation, (c) eating patterns of the
obese, (d) the genetic determinants of obesity, and (e) the health risks associated
with obesity. In view of this evidence, it will be argued that mental health pro-
fessionals should, under most circumstances, be advised uguinst the delivery of
dietary or behavioral treatments for mild or moderate obesity rather than pro-
posing more aggressive dietary approaches (Brownell & Jeffery, 1987). When
weight reduction is offered, consumers should be given complete information
about risks and probable outcome. Rather than expending further resources on
traditional treatments of obesity, health professionals should be encouraged to
further develop alternative approaches that more adequately address the physi-
‘There have been many attempts to define and classify obesity using either weight rela-
tive to norms, height to weight ratios, measures of body fat, or regional distribution of fat
(cf. Bjorntorp, 1987; Bray, 1978; NRC, 1989). Stunkard (1984) has proposed a simple
scheme of defining obesity relative to norms as mild (20%-40% overweight), moderate
(41%-100% overweight), or severe (more than 100% overweight). According to this deli-
nition, about 35% of women in this country are obese and of this group, 90.5% are
mildly, QLTO are moderately, and 0.5% severely obese (Stunkard, 1984). These definitions
of obesity are all arbitrary in the sense that they define the condition using cutoff points
along a normal distribution of body weight or fat, without reference to etiology or dis-
ease. Most of the behavioral and dietary research that will be summarized in this review
has been directed toward the mild and moderate groups. In the interest of avoiding te-
dium, the current review will not define the way that the term was used in each study
cited unless it has specific relevance to the interpretation of the results.
Failure of Obesity Treatment 731
cal, psychological, and social hazards associated with obesity without requiring
dieting or weight loss.
The enormity of the efforts directed toward eradicating obesity is perhaps best
illustrated in economic terms. According to a recent market analysis summarized
by Obesity and Health (“What’s Ahead?,” 1989), the 1988 cost of weight loss prod-
ucts and services in the United States alone was estimated at over $29 billion. To
put this figure in perspective, the Federal Government of the United States
spent a total of $31.9 billion in 1988 on education, training, employment, and
social services (Office of Management and Budget [OMB], 1990). According to
the weight loss industry report in 1988, $4.5 billion were spent in 1988 in hospi-
tal based clinics and weight loss programs, another $1.5 billion in the almost
13,000 nonhospital programs, and about the same amount on residential spas
offering weight loss programs. Americans additionally spent $8 billion on health
spas and exercise clubs, $382 million on 54 million diet books, $10 billion on
diet soft drinks, and billions more on low calorie foods and artificial sweeteners.
Surgery is not included in this analysis, but there are an estimated 30,000 stom-
ach redution and jaw wiring operations per year. Additionally, about 100,000
liposuction procedures were performed at a cost of $3,500 apiece. The weight
loss industry is a major economic force in this country, projected to reach over
$50 billion by 1995.
For several decades the fashion, entertainment, and publishing industries have
bombarded women with role models for physical attractiveness so gaunt as to
represent virtually no women in the actual population (Garner & Garfinkel,
1980; Garner, Rockert, Olmsted, Johnson, & Coscina, 1985). These representa-
tions imply that not only beauty but also success, personal happiness, and self-
worth can be achieved through slenderness. There seems little doubt that
cultural pressures for slimness have intensified in the past 30 years (Garner,
Garfinkel, Schwartz, & Thompson, 1980; Silverstein, Perdue, Peterson, Vogel, &
Fantini, 1986). Although the reasons for this are unclear, one argument holds
that as women have moved into previously male-dominated activities, the “tradi-
tional” female body shape has developed negative connotations while the mascu-
line shape has come to symbolize self-discipline and competency (Bennett &
Gurin, 1982; Wooley & Wooley, 1985). Indeed, Silverstein et al. (1986) found
that curvaceousness went out of style both in the 1920s and in the 1970s and
198Os, both periods of peak activity on the part of women to gain male preroga-
tives. Marketing geniuses, if not responsible for these trends, have certainly capi-
talized on them by promoting self-consciousness and personal discontent so that
consumers will believe that they require the remedies the advertisers offer
(Ewen, 1976; Wachtel, 1989).
The power of these influences is evidenced in studies showing that the major-
ity of adolescent and young adult women feel fat even when their weight falls
within the normal range (e.g., Dwyer, Feldman, & Mayer, 1967; Huenemann,
Shapiro, Hampton, 8c Mitchell, 1966; Klesges, 1983; Moses, Banilivy, & Lifshitz,
1989; Nylander, 1971; Wardle & Beales, 1986). This has led Rodin, Silberstein,
and Striegel-Moore (1985) to characterize the shape dissatisfaction endemic to
young women as “normative discontent.” Given these attitudes, it is not surpris-
73.2 D. M. Garner and S. C. Wooley
ing that about two out of every three high school girls report that they are try-
ing to lose weight (Rosen & Gross, 1987). Increasingly, there are reports
indicating that substantial numbers of young women use harmful weight control
behaviors, such as self-induced vomiting, with a subgroup of these manifesting
serious eating disorders (Crowther, Post, SC Zaynor, 1985; Killen et al., 1987;
Pyle, Halvorson, Neuman, & Mitchell, 1986). Recent studies indicate that feeling
fat and restricting food intake are not uncommon in schoolchildren 8-13 years
of age (Davies & Furnham, 1986; Maloney, McGuire, Daniels, & Specker, 1989;
Wardle 8c Beales, 1986). One of the more alarming manifestations of the fear of
obesity is a subgroup of children with short-stature syndrome and delayed pu-
berty caused by rigorous dieting due to fear of fatness (Puglifse, Lifshitz, Grad,
Fort, & Marks-Katz, 1983). The desperation connoted by these behaviors is eas-
ily overlooked in a cultural climate which champions thinness as an indisputable
virtue. Many of the behaviors used for diagnoses of anorexia nervosa and bu-
limia nervosa are culturally syntonic, considered neither uncommon nor abnor-
mal by the lay public (Huon, Brown, 8c Morris, 1988). Even anorexia nervosa
itself has developed a not altogether unfavorable connotation in our culture
(Branch & Eurman, 1980; Bruch, 1985; Garner et al., 1985; Wooley & Wooley,
The intensity of the prevailing cultural pressures for thinness and the psycho-
logical risks that they impart have particularly serious implications for those who
suffer from moderate to extreme obesity (Wooley et al., 1979a). Shape dissatis-
faction and attempts to restrict food intake are evident in all but a small minor-
ity of women who are statistically overweight (Davies & Furnham, 1986; Wardle
& Beales, 1986), and extreme body image disparagement has been recognized
for years as a grim consequence of obesity (Stunkard & Burt, 1967; Stunkard &
Mendelson, 1967; Wadden & Stunkard, 1985).
The social stigma against obesity is extraordinary in its magnitude and perva-
siveness. As Fitzgerald (1981) notes: “Public derision and condemnation of fat
people is one of the few remaining social prejudices . . . allowed against any
group based solely on appearance” (p. 223). It is well documented that obese
people are denied educational opportunities, jobs, promotions, and housing be-
cause of their weight (Bray, 1976; Canning & Mayer, 1966; Karris, 1977). The
affront to those who are obese goes beyond the almost uniform judgment that
they are unattractive (Furnham & Radley, 1989) and includes negative stereo-
types that begin early in childhood. The aversion to obesity is evident by kinder-
garten (Lerner, 1969; Lerner & Gelbert, 1969), and grade-school children
consistently associate larger body shapes with adjectives such as “stupid,” “dirty,”
“lazy, “sloppy,” “mean,” “ugly,” and “sad” (Maddox, Black, & Liederman, 1968;
Staffieri, 1967, 1972; see Wooley et al., 1979a for a review).
Harsh attitudes toward the obese depend on the assumption that they bring
their condition on themselves through lack of willpower and self-control (cf.
Harris & Smith, 1982; Maddox et al., 1968). In the face of the o~~erwhelming
social rejection, obese individuals may seek professional support; however, re-
search has shown that health professionals share in the culture’s perjorative view
of obesity and that this prejudice may influence their clinical judgement (Brot-
man, Stern, & Herzog, 1984; Maiman, Wang, Becker, Finlay, & Simonson, 1979;
Richardson, Goodman, Hastorf, & Dornbusch, 1961; Young & Powell, 1985). A
negative attitude toward obesity is suf~ciently ingrained in our social fabric that,
Failure of Obesity Treatment 733
in addition to social rewards for success, obesity treatment programs have at
times included social shaming procedures as a treatment principle. Many au-
thors have advocated a critical attitude toward obese patients (cf. Wooley &
Wooley, 1980). Confronted with evidence of consistent failure of behavioral
treatment programs - a failure seemingly understandable only as a conse-
quence of widespread patient noncompliance - several respected researchers
concluded that social pressure on obesity needed to be increased:
The ultimate social pressure treatment would be to increase the social sanctions
against obesity, so that being overweight would be a tremendously shameful thing.
In this manner, obesity would be under external control, as are other social behav-
iors for which society has learned that internal control is not enough for some, as in
the case of sexual or criminal acts. Obesity control would then be inculcated into
the normal socialization process for children. (Foreyt et al., 1981, p. 170)
This statement no doubt expressed for a generation of researchers the frustra-
tion of reported failure and needs to be understood not as the position of an
individual but as an illustration of the way in which ineffective treatment strate-
gies lead, as an almost inevitable consequence, to blaming the victim.
The Paradox of Dietary Treatment of Obesity
There are two indisputable facts regarding dietary treatment of obesity. The first
is that virtually all programs appear to be able to demonstrate moderate success
in promoting at least some short-term weight loss. The second is that there is
virtually no evidence that clinically significant weight loss can be maintained over
the long-term by the vast majority of people. Since health professionals and pro-
fessional societies recommend weight loss as the treatment of choice for the 23
million or so Americans judged to be overweight, the apparent inconsistency be-
tween the short- and long-term treatment findings needs to be illuminated.
Short-Term Follow-Up Studies
Behavioral weight control programs for obesity are often described as demon-
strating promising treatment effects that are well maintained during the first
year of follow-up (cf. Bennett, 1986; Brownell & Jeffery, 1987; Wadden 8~
Stunkard, 1989). In an analysis of the results from 105 behavioral studies, Ben-
nett (1986) reported a mean weight loss for the 6,121 participants of 5.38 kg
after an average of 13.37 weeks of treatment. The modest treatment effects were
maintained in the short-term with a mean weight loss of 5.11 kg recorded for
5,453 participants followed for an average of 35.5 weeks.
Although most of the behavioral treatment studies have involved oniy short-
term evaluation of effectiveness, follow-up duration has increased in recent
years. Brownell and Jeffery (1987) have reported that the average length of fol-
low-up for a sample of studies appearing in several major psychology journals
was 15.5 weeks in 1974, 30.3 weeks in 1978, 58.4 weeks in 1984, and 44 weeks
in 1986. The duration of treatments provided and the amount of weight loss
achieved has increased over the same period, although the number of pounds
lost per week has not changed dramatically (Brownell & Jeffery, 1987).
734 D. M. Garner and S. C. Wooley
There has been a move away from short-term, strictly “behavioral” programs
to greater emphasis on multicomponent interventions that incorporate exercise,
social influence, longer treatment duration, and continued therapeutic contact
after the end of formal treatment. These strategies have been found to promote
greater weight losses and improved maintenance during the first 18 months af-
ter treatment (Perri, McAdoo, McAllister, Lauer, & Yancey, 1986; Perri et al.,
1988; Perri, Nezu, Patti, & McCann, 1989), leading some authors to the conclu-
sion that obesity might be reconceptualized as a chronic condition “requiring
some form of continuous-care program long after completing an initial period
of treatment” (Perri et al., 1988, p. 533). However, over and over again the ini-
tial encouraging findings are eroded with time. Following up their initially im-
pressive treatment results, Perri et al. (1989), for example, state “After clients
conclude their active involvement in obesity treatment, they typically abandon
weight loss strategies and begin to regain weight” (p. 452).
The problem is a subtle one that occurs repeatedly; results are presented as
positive when, in fact, all indications are that the long-term projections are quite
poor. This matter of interpretation is an especially questionable convention since
there are consistent data on the pattern of weight gain that follows weight loss
achieved through a wide range of treatment strategies. To give a fairly typical
example, the findings of Craighead, Stunkard, and O’Brien (1981) are often
cited as one of the more successful behavioral programs in maintaining weight
loss (Figure 1). It is evident from Figure 1 that the active treatments led to statis-
tically significant weight loss and that there was a trend for the drug and com-
bined treatments to produce weight loss superior to the behavioral therapy
alone. At one year follow-up, however, the behavior therapy group showed the
least regain. In discussing these results, Stunkard (1984) states:
This most recent study provides grounds for optimism as to the future of behav-
ioral treatment of obesity . over the long run, behavior therapy clearly outper-
formed the most potent alternative treatment with which it has yet been compared.
(p. 1’35)
This interpretation does not reflect an appreciation of the apparent trajectory
of the posttreatment weight rebound for all treatment groups. The most parsi-
monious interpretation of these findings is that the participants are returning to
their pretreatment weight, and the graph is simply a snapshot of this process
one year after the end of treatment. It also fails to acknowledge that the trend
toward weight gain, together with similar findings from other studies, probably
represents a basic underlying mechanism unlikely to show a reversal or levelling
off if data were to be collected at subsequent follow-up points.
Brownell and Jeffery (1987) have similarly avoided the most obvious implica-
tion of these and comparable results, arguing that the negative findings may ac-
tually be positive since those who have regained most or all of their weight at
follow-up might have weighed even more had they not received treatment. To
give this speculation more credence than the alternative possibilities that obese
individuals would be at the same or lower weights if they had not received any
treatment goes beyond data on long-term weight fluctuations in the obese
(Williamson & Levy, 1988). Findings from another study by Brownell and col-
leagues (Brownell, Greenwood, Stellar, & Shrager, 1986), as well as others to be
Failure of Obesity Treatment 735
0 2 4 6 8 IO I2 14 16 18
L \ I
Treatment Follow-up
FIGURE 1. Weight Changes During 6-mo Treatment and 12-mo Follow-
Up. (Vertical lines represent 1 SEM. From “Behavior Therapy and Phar-
macotherapy for Obesity” by L. W. Craighead, A. J. Stunkard, and R. M.
O’Brien, 1981, Archives of General Psychiatry, 38, pp. 763-768. Copyright
1981 by the American Medical Association. Reprinted by permission.)
discussed later, support the hypothesis that dieting itself may lead to enhanced
metabolic efficiency (i.e., greater energy storage at a fixed energy inake) and a
possible tendency to maintain a higher body weight.
Brownell and Jeffery (1987) also assert that the failure to demonstrate success
in clinical samples does not obviate the possibility that weight loss may be a real-
istic goal for those who do not seek treatment. It is reasoned that treatment stud-
ies draw participants from clinical settings; these may be the most recalcitrant
individuals from the potential pool of candidates for weight loss. While this is
obviously a theoretical possibility and is supported by one study on a highly se-
lected sample (Schachter, 1982), it seems to stretch the weight loss litany to its
limits and leads to the curious interpretation that treatment is most appropriate
for those who do not seek professional help.
Against a tradition in science which assumes that there are no treatment ef-
fects until they are demonstrated, there is an unchallenged convention by which
weight loss interventions are presumed effective until there is explicit evidence
736 D. M. Garner and S. C. Wooley
to the contrary. The reality is that we do not have effective treatment to offer,
and we should be candid about this until there is reliable evidence to the con-
trary. To avoid confronting the failures of obesity treatment is to mislead a pub-
lic desperately waiting with cash in hand for an effective remedy.
Another criticism of behavioral and certain other treatments for obesity is that
the amount of weight loss achieved, while statistically significant, is generally too
small to be of practical consequence (Wing & Jeffery, 1979). According to
Brownell and Jeffery (1987), the mean weight loss for controlled behavioral tri-
als reported in 1984 was 15.4 lb, with an average of just over 13 weeks of treat-
ment; however, the initial mean weight of participants was 197 lb.
In answer to the rather modest weight losses achieved by most dietary ap-
proaches, it has been recommended that more aggressive methods be imple-
mented in order to achieve greater initial losses (Brownell & Jeffery, 1987).
These have included very low calorie diets (VLCD); longer treatments; and the
addition of exercise, financial contracts, relapse prevention, and social support
to behavioral regimens (Brownell & Jeffery, 1987; Brownell, Marlatt, Lichten-
stein, & Wilson, 1986). These proposals all appear to presume that main-
tenance of weight loss is a realistic expectation and/or that failure to maintain
weight loss does no harm. As will be more evident later, both of these assump-
tions are inconsistent with much of what we know.
Long- Term Folio w- Up Studies
In contrast to the body of literature indicating that both traditional and behav-
ioral treatments for obesity produce weight loss that is reasonably well main-
tained after one year, it is evident that weight is gradually regained over time,
with many individuals eventually weighing more than they did at the start of
treatment (Adams, Grady, Lund, Mukaida, & Wolk, 1983; Dubbert & Wilson,
1984; Kirschenbaum, Stalonas, Zastowny, & Tomarken, 1985; Murphy, Bruce, &
Williamson, 1985; Rosenthal, Allen, & Winter, 1980). This effect becomes more
conspicuous the longer that the follow-up is extended, and it is apparent that
most individuals will regain most or all of their weight after four or five years
(Bjorvell 8c Rossner, 1985; Graham, Taylor, Hovell, & Siegel, 1983; Jordan,
Canavan, & Steer, 1985; Kramer, Jeffery, Forster, & Snell, 1989; Murphy et al.,
1985; Stalonas, Perri, & Kerzner, 1984; Stunkard & Penick, 1979). For example,
in a 5-year follow-up, Stalonas et al. reported that the average participant had
gained 11.9 lb since the end of a behavioral program, making him or her 1.49 lb
heavier than when treatment began. Murphy et al. found that the average self
reported weight loss after four years for several different treatment conditions was
7.12 lb, but the average measured weight loss was actually about one pound. Jor-
dan et al. (1985) studied selected patients who had achieved at least a 15-lb
weight loss during treatment. They found that patients showed distinct patterns
of weight fluctuation, with only 12.2% remaining consistently below their post-
treatment weights for 5 years. The odds against maintaining long-term weight
loss are dramatically illustrated by Kramer et al. in a recent study of 114 men
and 38 women who had successfully completed a 15-week behavioral weight loss
program. They reported that less than 3% of the patients maintained their post-
treatment weight throughout the four yearly follow-up assessments. Monotonic
weight regain was the predominant pattern for the men, whereas the women
tended to lose and regain weight more often during the follow-up period.
Failure of Obesity Treatment 737
In a study that is interesting because it did not rely on a sample of patients
seeking weight loss treatment, Binnie (1977) identified and treated 43 individu-
als in a rural general practice who were above “ideal” weight.’ After 10 years,
only one patient had achieved ideal weight, and only two had achieved 80% of
their goal. These results are similar to other long-term studies indicating that
only a small minority of patients are able to achieve and maintain ideal weight
goals. Stunkard and Penick (1979) found that after 5 years, the majority of pa-
tients receiving both behavioral and traditional treatment regained most of the
weight originally lost.
The very few long-term studies reporting somewhat less dismal findings are
problematic because of possible bias in the sample of subjects studied at fol-
low-up (Graham et al., 1983; Grinker, Most, Hirsch, Borsdorf, & Wayler, 1985).
The most impressive results were those of Grinker et al., who reported that 55%
of subjects followed after a residential treatment program had maintained at
least a 5-kg weight loss; however, there was only a 38% response rate. Similarly,
Graham et al. targeted 62 of 138 initial participants in a weight loss program on
the basis of having telephone numbers and addresses available; it is unclear from
the report why such a high percentage was lost to follow-up
The effects of treatment do not stop with an initial weight loss or maintenance
period. Weight rebound seems to be almost as reliable a consequence of treat-
ment as initial loss. The pattern of weight gain recorded in extended follow-up
indicates that short-term results are frankly misleading indicators of long-term
outcome. Our tradition of reporting and interpreting data provides a model,
used in the promotion of virtually all weight loss programs and products, in
‘The terms “ideal” or “desirable” weight are typically derived from insurance company
actuarial tables to indicate weights at which insurance policy holders can expect greatest
longevity. They are widely used by health professionals and the lay public to characterize
the goal weight that one should try to achieve for greatest health. They have been criti-
cized on a number of technical grounds such as sampling, variability of methodologies
used to gather weight and mortality data, and the fact that they do not take into consid-
eration individual differences in lean body mass, skeletal size, and regional fat distribu-
tion (Harrison, 1985; Knapp, 1983). “Frame sizes” reported in the 1983 Metropolitan Life
Insurance height-weight tables were not derived from anthropometric measurements but
were based on arbitrary groupings of the data (Weigley, 1984). Perhaps the greatest con-
cern regarding “ideal” and “desirable” weights relates to the interpretation of the weight
table figures. They fail to take into consideration that body weight, like other physical
attributes, could be expected to vary naturally in the population (measures of dispersion
are not included in the weight tables used to recommend desirable weights). Diverse ge-
netic propensities within the population may mean that what is a healthy weight for a
population may not have any relevance to the individual. Moreover, there is little attempt
to actually inform individuals of the magnitude of risk associated with particular deviations
from specified weights. Results from Sorlie et al. (1980) indicate that for both men and
women within the middle three quintiles of the weight distribution, there is no relation-
ship between weight and mortality. For women of average height (5 ft 3 in. to 5 ft 6 in.)
the risk of death was virtually identical for those between 115 lb and 194 lb. While women
in the highest and lowest 20% of the weight distribution had a higher risk of death, this
was considerably lower than the mortality rate for thin and average weight men. Finally,
even though certain weights may be related to longevity in the insurance company sam-
ples, as will be discussed more fully later, it is mere speculation that changing one’s weight
category will actually alter longevity.
738 D. M. Garner and S. C. Wooley
which it is implied that treatments can be adequately evaluated by their immedi-
ate effects. However, if treatment studies are to be relevant, the maintenance of
clinically meaningful weight loss must be the agreed-upon standard.
One would expect that existing data would lead to unanimity in the research
community about the need for long-term evaluations in order to draw conclu-
sions about treatment efficacy. It is not that failures of dietary treatment are un-
known to researchers. Indeed, a number of well respected researchers have
based their repeated calls for long-term follow-up upon these observations
(Brownell et al., 1986; Foreyt et al., 1981; Stunkard, 1975; Wilson, 1978). Nev-
ertheless, many others have remained undaunted by the failure to demonstrate
long-term success and have argued not for the abandonment of dietary treat-
ment but instead have called for more “agressive approaches” to treatment such
as the very low calorie diet (VLCD; e.g., Brownell & Jeffery, 1987).
Very Low Calorie Diets
The very low calorie diet of 300-500 calories per day was enthusiastically re-
ceived because of its ability to produce large and rapid weight loss. In a compre-
hensive review, Wadden, Stunkard, and Brownell (1983) described the VLCD as
“the most important recent development in the medical treatment of obesity” (p.
675). More recently, Brownell and Jeffery (1987) suggested that the VLCD “may
be the treatment of choice for persons who are severely obese and who have had
difficulty losing sufficient weight on more moderate dietary regimens” (p. 363).
Initial trials of VLCDs combined with behavior therapy seemed to support the
initial fervor by achieving significant weight loss, reasonably well maintained at
one year follow-up (Wadden & Stunkard, 1986; Wadden et al., 1983). However,
as with most approaches, the initial optimism has been dampened by a sobering
series of long-term follow-up studies indicating that the pattern with the VLCD
is the same as with other dietary treatments.
In their initial report, Wadden and Stunkard (1986) found that the VLCD
combined with behavior therapy produced significant weight loss (mean = 19.2
kg) and was more effective than either the behavioral or VLCD therapy alone.
At the end of one year, the weight loss for the combined treatment was reason-
ably well maintained (12.9 kg), but after three years virtually all of the treatment
effects were reversed, particularly when results were corrected for the effects of
intervening weight loss efforts (Wadden, Stunkard, 8c Liebschutz, 1988). Failure
of this approach is even more evident in a follow-up of 497 patients receiving a
combination of a VLCD (Optifast) and behavior modification through a large
health maintenance organization (Hovel1 et al., 1988). Hovel1 and colleagues
found that 55% of the patients who started treatment dropped out before com-
pletion. While the remaining 45% lost a mean of 83.9% of their “excess weight,”
the four follow-up groups, illustrated in Figure 2, regained between 59% and
82% of their initial excess weight by 30 months from the start of treatment. Pa-
tients began treatment, on average, about 50% over “ideal weight” defined by
weight norms.
Although there is an upward trajectory for all groups in Figure 2, it is inter-
esting to note that the group least consistently available for follow-up assess-
ments had the steepest slope. This may have implications for results of studies
that have selected subsamples for follow-up based upon availability. These re-
sults are consistent with those of another recent study that compared VLCD with
Failure of Obesity Treatment 739
5 I Entry
A group1
m group2
0 group3
* group 4
End TX
2 mo 18 mo 30 mo
Follow-up Measures
FIGURE 2. Mean Percent of Excess Weight and 95% Confidence Inter-
vals of Patient Groups Followed After Supplemented Fasting Regimen.
(Groups were defined by consistent membership across follow-up peri-
ods: Group 1 is an average of all patients available at a given follow-up,
Group 2 patients were available for two, Group 3 for three, and Group 4
for all follow-up periods. From “Long-Term Weight Loss Maintenance:
Assessment of a Behavioral and Supplemented Fasting Regimen” by M.
F. Hovell, A. Koch, C. R. Hofstetter, C. Sipan, P. Faucher, A. Dellinger,
G. Barok, A. Forsythe, and V. J. Felitti, 1988, American Journal of Public
Health, 78, pp. 663-666. Copyright 1988 by the American Public Health
Association. Reprinted by permission.)
gastroplasty and found that after five years only 3% of the VLCD group could
be considered successful compared to 16% of the surgery group (Andersen,
Stokholm, Backer, & Quaade, 1988). Moreover, there was no apparent leveling
off of the weight regain even after 5 years.
The failure of the VLCD to lead to sustained weight loss should not have been
a suprise since 15 years earlier Sohar and Sneh (1973) reported that most of the
small proportion of patients who were actually able to complete a low calorie
diet returned to within 10% of their pretreatment weight at a 14-year follow-up.
Even those who were below their pretreatment weight at follow-up remained
obese by most standards.
740 D. hf. Gamer and S. C. Wooley
Perhaps of even greater concern than the negative findings is the unjustified
optimism found in the initial reports of some of the more recent VLCD studies.
A year prior to publication of the findings in the Optifast VLCD study men-
tioned earlier, a news release from the manufacturer of the protein supplement
attributed the following summary of the results to one of the researchers:
Weight loss can now be easily, safely and routinely accomplished with good proba-
bility of maintaining the loss . . .[F]ive years ago this just wasn’t the case. The fron-
tier of dieting has moved from simply accomplishing a weight loss to maintaining
the loss long-term. (San Diego VLCD Study, 1989)
Although the rate and magnitude of weight loss have been the basis for rec-
ommending the VLCD, its most remarkable feature is the speed of weight re-
gain following treatment. The association between large weight losses during
treatment and rapid regain was observed by Stunkard and Penick (1979) in ref-
erence to behavioral and traditional treatments. They noted that this pattern was
similar to that found following fasting treatments. In recalculating the weight
losses reported by Swanson and Dine110 (1970), Stunkard and Penick (1979)
found that the more weight lost during treatment, the higher the follow-up
weight. Accordingly, Swanson and Dine110 (1970) reported that 78% of their 18
super obese patients had returned to their original weight within 3 years of end-
ing a fasting treatment.
These findings are consistent with those of Johnson and Drenick (1977), who
followed 121 moderately obese patients for an average of 7.3 years after losing
an average of 28.2 kg through a supervised fast. After 2-3 years, 50% of the
group had reverted to their original admission weight and 90% had done so af-
ter nine years. As Figure 3 illustrates, only 7 of the patients (5.7%) remained at
reduced weights during the follow-up period. According to the authors, this out-
come “required continuous conscious dieting [and] the rate of regain was quite
uniform, regardless of baseline weight, degree of weight loss, length of the fast,
or the duration of obesity” (p. 1382). In another report of these findings, it was
stated that L‘a sizable number of patients who had maintained a stable obese
weight for several years prior to treatment eventually ended up considerably
more obese than before weight reduction” (Drenick SC Johnson, 1980, p. 33). Al-
though fasting is now regarded as carrying an unacceptable mortality risk (Wad-
den et al., 1983), it is important to note that the poor maintenance found with
this approach is consistent with the more recent results with VLCDs.
Irrespective of the relative merits of the VLCDs and the behavioral or dietary
approaches to obesity, there is remarkable consistency in the pattern of weight
regain. It would seem that the correct interpretation of long-term findings is
quite straightforward. Most approaches lead to weight loss during active treat-
ment, and many individuals continue to lose in the interval directly following
treatment; however, most participants ultimately regain to levels that approxi-
mate their pretreatment weight. It is only the rate of weight regain, not the fact
of weight regain, that appears open to debate. While this may be discouraging to
the individual intent on weight loss, it should also provide some solace to the
many individuals who have failed at dieting and have attributed the failure to a
personal lack of will power.
Failure of Obesity Treatment 741
Months Post Weight Loss
(mean loss from 143 kg to 106 kg)
FIGURE 3. Percent of Patients Remaining at Reduced Weights at Vari-
ous Time Intervals After Achieving Weight Loss. (Solid line represents
mean for all patients. From “Therapeutic Fasting in Morbid Obesity:
Long-Term Follow-Up” by D. Johnson and E. J. Drenick, 1977, Archives
of Internal Medicine, 137, pp. 1381-1382. Copyright 1977 by the Ameri-
can Medical Association. Adapted by permission.)
What Really Accounts for the Failure to Maintain Weight loss?
There is remarkable resemblance between the pattern of weight regain seen in
humans following treatment, illustrated in Figures l-3, and the results from ex-
periments in which laboratory animals are exposed to caloric deprivation, made
to lose weight, then returned to unrestricted eating. Figure 4 presents the results
from one study in which laboratory rats lost weight in response to caloric restric-
tion (Mitchel & Keesey, 1977). When ad lib eating was restored, the animals re-
gained to body weight levels that were very close to their nondeprived littermates
- all of this without the benefit of bathroom scales or (presumably) social pres-
sure. As further illustrated in Figure 4, the same pattern of body weight defense
was evident in a group of animals with lesions to the lateral hypothalamus, ex-
cept that the absolute weight levels were lower.
The concept of “set point” has been proposed to account for these and other
data from human and animal studies showing that there is remarkable stability
in body weight over time. According to the set point concept, body weight is
regulated by physiological mechanisms that oppose the displacement of body
weight caused by either over or underfeeding (Nisbett, 1972; Powley & Keesey,
1970). There is some debate about the precise physiological mediators and the
factors that influence the absolute levels at which weight regulation occurs, lead-
ing some to question the utility of the set point concept (Mrosovsky & Powley,
1977). Nevertheless, there are impressive data indicating that:
742 D. M. Garner and S. C. Wooley
300- A
c; I
5 I
2 am-
(3 400-
z 350 - 0 CONTROL-CONTROL W-3)
300. I I
0 IO 20 30 40 SO 60 70
FIGURE 4. Recovery of Body Weight by Control and L-H-Lesioned Rats
Following Food Restriction and return to Ad Libitum Feeding Schedule.
(From “Defense of a Lowered Weight Maintenance Level by Lateral Hy-
pothalamically Lesioned Rats: Evidence From a Restriction-Refeeding
Regimen” by J. S. Mitchel and R. E. Keesey, 1977, Physiology and Behav-
ior, 28, p. 1123. Copyright 1977 by Pergamon Press plc. Reprinted by
Displacement of body weight usually results in “homeostatic” metabolic ad-
justments designed to return the organism to the body weight normally
This body weight “defense” occurs in obese as well as nonobese animals.
Genetic factors influence both regulated weight and the propensity toward
upward and downward regulation of body weight under certain circum-
stances in response to environmental challenges.
Certain environmental factors, such as diet palatability, exercise, smoking,
climate, and certain drugs, appear to influence the absolute levels (within
certain limits) at which body weight is regulated (Keesey, 1986), and thus
the term “set point” is probably less satisfactory than “regulated weight.”
Metabolic Adaptation to Changes in Body Weight
It is so commonly assumed that obesity reflects a failure of body weight regula-
tion that the idea it may represent an adaptive or “normal” state seems almost
inconceivable. However, there is consistent evidence from both human and ani-
mal studies that weight loss in the obese as well as nonobese leads to reductions
of 15-30s in energy requirements (Barrows & Snook, 198’7; Boyle, Storlien,
Failure of Obesity Treatment 743
Harper, & Keesey, 1981; Bray, 1969; Donahoe, Lin, Kirschenbaum, & Keesey,
1984; Dull00 & Girardier, 1990; Elliot, Goldberg, Kuehl 8c Bennett, 1989; Finer,
Swan, 8~ Mitchell, 1986; Geissler, Miller, & Shah, 1987; Keesey, 1988: Leibel &
Hirsch, 1984; Ravussin, Burnand, Schutz, & Jequier, 1982). Corbett, Stern, and
Keesey (1986) reported that reducing the body weight of rats by 14.9% through
caloric restriction led to a 24.6% decline in resting metabolic rate. Comparable
declines in metabolic rate have been demonstrated in humans exposed to semi-
starvation conditions (Keys, Brozek, Henschel, Mickelson, & Taylor, 1950).
Obese patients have been shown to make the same or greater metabolic adjust-
ments as their lean controls during reduced caloric intake and weight loss (Bar-
rows & Snook, 1987; Elliot et al., 1989; Finer et al., 1986; Leibel & Hirsch,
1984). Leibel and Hirsch reported that obese humans who had lost a significant
amount of weight (52 kg) and were still judged to be 60% overweight had ca-
loric requirements 28% below pre-weight-loss levels, which were themselves 25%
lower than would be predicted from body size alone. Even short-term restric-
tions involving small decrements in weight are accompanied by rather dramatic
metabolic adjustments. For example, Bray (1969) has reported a 17% decline in
resting metabolic rate in obese patients with only a 3% weight reduction! This
observation parallels Keesey’s (1988) finding that, while the resting metabolism
of genetically obese rats was initially 26% higher than that of lean littermates,
following only a 4.4% diet-induced weight loss, energy requirements were re-
duced to that of the lean controls weighing less than half as much. Studies of
metabolic changes during weight loss indicate that there is tremendous individ-
ual variability in the degree of resistance to displacement resulting from caloric
restriction. Absolute weight alone provides little information about caloric needs
and the speed of weight loss given a set caloric intake (Miller & Parsonage,
Importantly, when food intake is normalized after a period of food restriction,
there is a tendency for energy to be redeposited preferentially as body fat (Dul-
loo & Girardier, 1990; Ozelci, Romsos, & Leveille, 1978). In animal studies, Dul-
loo and Girardier (1990) have shown that this is true even if the fat content of
the diet is negligible during regain (i.e., 3%). Their research has also shown that
when the extra energy cost of fat synthesis during refeeding is considered it ac-
counts for even further metabolic efficiency.
Wadden, Foster, Letizia, and Mullen (1990) have recently challenged the in-
terpretation that Resting Metabolic Rate (RMR) is adversely affected by severe
caloric restriction by showing that the dramatic initial decline in RMR following
a VLCD is largely reversed when the diet is stopped. In this study, 18 obese
women were assigned to either a Balanced Deficit Diet (BDD), in which 1200
calories were prescribed for the 48 weeks of the study, or a VLCD in which 420
calories were consumed for 16 of the first 17 weeks and a conventional reducing
diet was followed for the remainder of 48 weeks of treatment. Resting Metabolic
Rate and body weight were assessed nine times over the course of the 48 weeks
(see Table 1). Although Table 1 illustrates that the dramatic short-term decline
in RMR following the VLCD is partially reversed when participants increased
their intake, several findings are particularly noteworthy. First, although the au-
thors concluded that “neither dietary regimen, combined with modest physical
activity, was associated with long-term reductions in metabolic rate that exceeded
decreases anticipated with the achievement of a lower body weight” (p. 707), the
caloric intake for both groups indicates that they were still actively restricting
744 R. M. Gamer md S. C. Wooley
TABLE 1. Changes in Weight and Resting Metabolic Rate Over 48 Weeks in
Patients in the BDD and VLCD Conditions
Changes in Weight, kg”
-4.1 i 0.7
-7.0 5 1.0
-8.7 & 1.8
-11.0 -+ 2.0
-12.9 * 2.1
- 12.6 t 2.5
- 14.3 t 3.0
- 18.2 * 3.5
-7.5 i_ 0.6”
- 14.2 -+ 0.7%
- 18.7 % 1.0”
-23.1 rfr 1.1”
-25.2 r 1.6”
-26.0 2 1.6*
-24.7 I 2.5”
-21.6 z!z 2.9
Changes in Resting
Metabolic Rate, kJ/d”
-808 rt 396
-895 rt 267
-1114 It 295
- 1053 rf: 356
-1035 ~fr 234
-936 -t- 264
-786 1?7 341
-851 i: 358
- 1622 -e 397**
-1629 t 3%3**
- 1677 2 33a***
- 1865 * 552*“”
- 1237 f 321
- 1095 ?? 397
- 948 t 330
- 697 t 200
Note. BDD indicates balanced-deficit diet; VLCD, very-low-calorie diet. Values are mean
* SEM. Through Week 21, n = 9 for both conditions; n = 8 and n = 7 in both condi-
tions at weeks 25 and 48, respectively. From “Long-Term Effects of Dieting on Resting
Metabolic Rate in Obese Outpatients” by T. A. Wadden, G. D. Foster, K. A. Letizia, and
J. L. Mullen, 1990, Journal qf the A~e~c~n clerical Associatiorl, 244, pp. 707-71 I. Copyright
1990 by the American MedicaI Association. Reprinted by permission.
“Changes in weight and resting metabolic rate were significantly different from baseline
for both conditions at all assessment periods (p<.OOl).
*Difference between conditions was significant at p<.OOl. **Difference between
conditions was significant at pc.01. ***Difference between conditions was significant at
their food intake. The VLCD and the BDD subjects reported a mean caloric in-
take at Week 25 of 1118 calories and 1198 calories, respectively, and at Week 48
of 1392 and 1285 calories, respectively. Although RMR for both conditions was
less than 10% below baseline (adjusted for changes in weight), it should be
pointed out that levels of caloric intake at the end of treatment were approxi-
mately 500 calories a day below the average caloric requirements for the average
woman of the same age (National Research Council [NRC], 1968). Secondly, by
Week 25 it is evident that the weight of the women in the VLCD was beginning
to rebound, indicating that metabolic homeostasis had not vet been achieved de-
spite continued caloric restriction. Finally, the obese condition was not cured in
these women. The women began treatment at an average of more