Review

Do Psychosocial Variables Predict Weight Loss

or Mental Health after Obesity Surgery?
A Systematic Review
S. Herpertz,* R. Kielmann,* A.M. Wolf, J. Hebebrand,‡ and W. Senf†

Abstract seem to be positive predictors of weight loss postsurgery.

HERPERTZ, S., R. KIELMANN, A.M. WOLF, J. The severity of the symptoms or the disorder is more
HEBEBRAND, AND W. SENF. Do psychosocial variables relevant for the outcome of obesity surgery than the speci-
predict weight loss or mental health after obesity surgery? A ficity of the symptoms. It is also not solely the consumption
systematic review. Obes Res. 2004;12:1554 –1569. of distinct “forbidden” foods, such as sweets or soft drinks,
Objective: The objective of this study was to present a but rather a general hypercaloric eating behavior, either as
systematic review of psychological and psychosocial pre- an expression of the patient’s inadequate compliance or a
dictors of weight loss and mental health after bariatric dysregulation in energy balance, which is associated with a
surgery. This systematic review included all controlled and poor weight loss postsurgery.
noncontrolled trials of the last 2 decades with either a
retrospective or prospective design and a follow-up period Key words: obesity surgery, psychosocial predictors,
of at least 1 year. weight loss, mental health, systematic review
Research Methods and Procedures: The relevant literature
was identified by a search of computerized databases. All
articles published in English and German between 1980 and Introduction
2002 were reviewed. Prevalence rate of obesity grade III (morbid obesity),
Results: Using the above inclusion/exclusion criteria, 29 defined as BMI ⱖ40 kg/m2 (1), is estimated to be 0.4% and
articles were identified focusing on psychosocial predictors 0.2% in men and 2.2% and 1.2% in women in the U.S. (2)
of weight loss and mental health after obesity surgery. and German population (3), respectively, with a consider-
Discussion: Personality traits have no predictive value for able increase in the last 2 decades (2). Nonsurgical treat-
ment such as the use of various diets, including very low
the postoperative course of weight or mental state. Apart
caloric diets, behavior therapy, or drugs is, at best, effective
from serious psychiatric disorders including personality dis-
for mild to moderate obesity (4). Furthermore, long-term
orders, psychiatric comorbidity seems to be of more predic-
follow-up of patients undergoing a combination of reduced
tive value for mental and physical well-being as two essen-
caloric diet and behavior therapy, for example, shows a
tial aspects of quality of life than for weight loss
return to baseline weight in the vast majority of subjects in
postsurgery. However, depressive and anxiety symptoms as the absence of continued intervention (5). For severe obe-
correlates of psychological stress with regard to obesity sity, all these weight reduction measures are, as a rule,
ineffective (6). Therefore, bariatric surgery has been recom-
mended for the treatment of severe obesity. Obesity surgery
Received for review July 9, 2003.
Accepted in final form July 21, 2004. pursues three goals: reduction of weight and an improve-
The costs of publication of this article were defrayed, in part, by the payment of page ment in medical risk factors and quality of life (QoL).1
charges. This article must, therefore, be hereby marked “advertisement” in accordance with
18 U.S.C. Section 1734 solely to indicate this fact.
Numerous studies have been performed in the past, es-
*Department of Psychosomatic Medicine and Psychotherapy, Westfaelische Klinik Dort- pecially with regard to the amount of weight reduction, the
mund, Ruhr-University Bochum, Germany; †Clinic of Psychosomatic Medicine and Psy-
chotherapy, Rheinische Kliniken, University of Duisburg, Essen, Germany; and ‡Depart-
ment of Child and Adolescent Psychiatry of the University of Marburg, Marburg, Germany.
1
Address correspondence to Stephan Herpertz, Department of Psychosomatic Medicine and Nonstandard abbreviations: QoL, quality of life; BE, binge eating; BED, binge eating
Psychotherapy, Westfälische Klinik Dortmund, Ruhr-University Bochum, Marsbruchstrasse disorder; MMPI, Minnesota Multiphasic Personality Inventory; VBG, vertical banded gas-
179, 44287 Dortmund, Germany. troplasty; RYGB, Roux-en-Y gastric bypass; SOS, Swedish Obese Subjects; TFEQ, Three-
E-mail: [email protected] Factor Eating Questionnaire; LAGB, laparoscopic adjustable gastric banding; EWL, excess
Copyright © 2004 NAASO weight loss.

1554 OBESITY RESEARCH Vol. 12 No. 10 October 2004

Psychosocial Predictors of Obesity Surgery, Herpertz et al.

complications associated with bariatric surgery, and the Inclusion Criteria

impact of weight loss on obesity-related comorbidity. A We included in our review all articles on presurgery
recent metaanalysis (7) of six longitudinal studies reporting predictors of weight loss, mental health, and QoL published
weight loss at 1 year and four studies with a mean follow-up in English and German between 1980 and 2002. Both pro-
of 9 months to 7 years has demonstrated BMI reductions of spective studies with pre- and postoperative assessment and
16.4 and 13.3 kg/m2, respectively. Moreover, a decisive retrospective studies evaluating possible psychosocial pre-
improvement in almost all medical risk factors has been dictor variables for postsurgery weight loss and mental
observed (8 –12), together with a decrease in surgical risk health were reviewed. Six further studies investigating post-
factors in specialized centers with a current 1% mortality surgery predictors of weight loss (28 –32), QoL (33), and
rate (13). For a majority of the patients, obesity surgery also medical complications (30) were also included in the re-
leads to a considerable improvement in QoL which includes view. From the initial pool of articles identified in the
such psychosocial variables as psychiatric comorbidity, so- search, we excluded all reviews, letters to the editor, opinion
cial functioning, self-esteem, and economic variables (14). pieces, and other articles not reporting original research on
There is general agreement that loss of excess body weight psychosocial predictor variables of obesity surgery. More-
is not a sufficient outcome measure regardless of the nu- over, we calculated the drop-out rate of each study to
merous definitions of successful outcome. However, the estimate the representativeness of the results. With the ex-
majority of outcome variables such as comorbidity reduc- ception of one study (34), we excluded all studies with a
tion, economic changes, psychosocial adaptation, and ob- drop-out rate of ⬎50%. The study of Powers et al. (34) with
jective and subjective QoL changes are related to loss of a drop-out rate of 57% was not excluded because its main
excess body weight. Therefore, weight loss will always be goal was a comparison of bariatric patients requiring psy-
one of the most important outcome variables by which the chiatric hospitalization after operation and a matched group
success of obesity surgery will be measured, especially with of nonhospitalized bariatric patients to identify factors as-
regard to the costs of such an operation. sociated with postsurgical psychiatric hospitalization. For
Approximately 20 to 30% of patients undergoing obesity this purpose, the drop-out rate was of minor importance.
surgery regain weight at 18 months to 2 years after surgery Another inclusion criterion was the time frame from oper-
(15–17). Some researchers attribute the weight regain to ation to follow-up of at least 1 year. Short-term weight loss
physiological factors (18); others claim that inadequate cop- is common but for an intervention to be deemed as effective,
ing strategies, the psychological inability of the patient to weight loss must be sustained in the long term. Still, this
adjust to the new anatomy and its consequences (19), are follow-up period seems to be rather short when taking into
usually at the source of patients’ inability to maintain account the fact that weight loss after surgical treatment
weight loss (20). Multiple studies have been undertaken to reaches its nadir 18 to 24 months after the operation (28,35).
identify psychosocial factors, anthropometric data, eating If studies had multiple follow-up assessments, the longest
behavior, or metabolic parameters that predict successful follow-up period was chosen. Studies that did not exactly
weight loss after bariatric surgery. For instance, psychiatric indicate the diagnostic procedure or that had methodologic
comorbidities, especially eating disorders, are considered to flaws such as high drop-out rates (36,37) or the risk of
be factors that might adversely affect the outcome. Substan- considerable selection bias (38) were excluded. Studies
tial psychopathology is reported in morbidly obese individ- using projective techniques that require a high level of
uals requesting obesity surgery, with a lifetime prevalence inference to interpret were also not reviewed (39,40). Stud-
of 84% reporting major mental disorders and 39.5% to 72% ies on intestinal bypass surgery were not reviewed because
reporting personality disorders (21,22) according to the intestinal bypass surgery is considered obsolete nowadays
DSM (23). However, evidence is accumulating that severe and has been abandoned. Articles reviewed were based on
obesity cannot generally be ascribed to psychopathological both exclusively restrictive surgery procedures such as gas-
determinants; at most, such determinants can be ascribed to tric banding or gastroplasty and gastric bypass. No study on
certain subgroups of obese patients, for example, those with biliopancreatic diversion (BPD) evaluating predictive out-
a binge eating disorder (BED) (24 –26) or subtypes of come variables could be found. The remaining articles were
affective disorders (27). retrieved and reviewed in detail. Table 1 illustrates the
The objective of this paper was to perform a systematic inclusion and exclusion criteria of this review.
review of the psychosocial predictors of weight loss after
bariatric surgery as well as postsurgery mental health.
Results
On the basis of the described literature search procedure,
Research Methods and Procedures 29 studies that assessed possible predictor variables of
The relevant literature was identified by a search of weight loss and postsurgery mental health and that met all
computerized databases (including MEDLINE and PSYCH- criteria for inclusion in our review were identified. Seven
LIT). studies were from Australia, nine from Europe, one from

OBESITY RESEARCH Vol. 12 No. 10 October 2004 1555

Psychosocial Predictors of Obesity Surgery, Herpertz et al.

based on exclusively restrictive obesity surgery procedures

Table 1. Inclusion and exclusion criteria such as gastroplasty, either horizontal-banded or vertical-
Inclusion criteria banded, or unspecified gastroplasty. Ten studies
(20,24,29,31,41,42,44 – 47) involved gastric bypass proce-
Identified by Medline and Psychlit
dures or compared gastric bypass with other obesity surgery
English or German language
procedures. Twenty-one studies were interview-based [one
Published between 1980 and 2002 study used a standarized interview (44), eight studies used a
Minimum follow-up of 1 year (semi-) structured interview (20,24,46,48 –52), and 12 stud-
Report of any of the following outcome measures: ies used a clinical interview (28,30,33,34,41,42,53–58)].
Measures of weight change, absolute weight loss Eight studies employed only questionnaires (25,29,31,32,
(pounds, kilograms, kilograms per meter squared), 43,45,47,58).
percentage of weight loss relative to baseline
values, percentage of excess weight Predictor Variables
Evaluation of psychological/psychiatric symptoms or In Table 2, the results of the studies with different surgery
disorders by means of an expert judgment (e.g., procedures, mean sample sizes, drop-out rates, and fol-
psychiatrist), standardized instruments low-up periods investigating the relationship among pre-
(questionnaire, interview), or any standardized and postsurgical measures, weight loss, and mental health
measure of psychosocial variables are presented.
Exclusion criteria Twenty-two studies identified different psychological
and psychosocial predictors of the course of weight and/or
High probability of selection bias
mental health after bariatric surgery, whereas seven failed to
Small sample sizes (n ⬍ 10)
find any predictor (43,46,50,51,55,57,58). Of the 17 studies
Selection criteria not indicated identifying psychosocial variables with an effect on weight
Drop-out rates exceeding 50% loss, 12 had a prospective (28,29,32,41,44,45,47– 49,52–
Studies using projective techniques 54), and five had a retrospective design (20,24,25,31,42).
Intestinal bypass surgery Five prospective studies (30,33,34,48,56) and one study
comparing different subsamples pre- and postsurgery (24)
identified predictors of mental health postsurgery (Table 3).
Canada, and 12 from the U.S. One study compared different
surgery procedures prospectively on a randomized basis Personality Traits
(41). In 23 studies, surgery patients were assessed before Personality traits of obese patients undergoing bariatric
and after the operation. Six studies (20,25,31,42,43) evalu- surgery have frequently been studied (59). When not con-
ated data retrospectively or compared different subsamples sidering studies on jejunoileal bypass or those with a fol-
pre- and postsurgery (24). One longitudinal study (34) com- low-up period of less than 1 year, the number of studies to
pared bariatric patients requiring psychiatric hospitalization be reviewed finally included five studies (20,44,51,56,58).
after operation with a matched group of nonhospitalized Gentry et al. (44) studied 33 gastric bypass patients over a
bariatric patients to identify factors associated with postsur- period of 2 years using a modified version of the Minnesota
gical psychiatric hospitalization. Nineteen studies were Multiphasic Personality Inventory (MMPI). None of the

Table 2. Study characteristics

Follow-up rate Drop-out rate Sample size Weight loss§ Weight loss§
(months) (28 studies*) (%) (26 studies†) (29 studies) (BMI) (10 studies) (kg) (19 studies)
Mean 37.2 20.1 130 11.5 37.5
(SD) (33) (15.2) (147.4) (1.7) (6)
Range 12 to 156 0 to 57‡ 20 to 730 9.8 to 14 30.3 to 49.7

* One study (43) does not indicate the mean follow-up period.
† Three studies (20,42,43) using a postdesign do not indicate drop-out rates.
‡ The study of Powers et al. (34) had a drop-out rate of 57% (see text).
§ Weight loss was indicated in BMI units on kilograms.

1556 OBESITY RESEARCH Vol. 12 No. 10 October 2004

 Table 3. Psychological and psychosocial predictors of weight loss following bariatric surgery
Study, authors, Mean
type of surgical follow-up Assessment methods
procedure Sample period and measures Weight loss Predictors

Barrash et al. (58), 138 (151); women, 138; ma, 12 Standardized questionnaire 37.5 ⫾ 15.5 kg Severe psychological problems presurgery
VBG 35.6 ⫾ 10 years; dor, 8.6% predicted lower absolute weight loss postsurgery
Brolin et al. (52), 42 (56); women, 30; men, 12; 18 Semistructured interview, 30.3 kg Positive predictor of weight loss: lower initial
horizontal ma, 36.6 to 47.0 years, dor, standardized questionnaires, percentage above ideal weight, younger age, low
gastroplasty 25% clinical interview number of preoperative food contacts, higher
levels of psychological stress; no predictor:
socioeconomic status
Busetto et al. (30), 80; women, 57; men, 23; ma, 12 Clinical interview, 24 hour BMI, 12.4 BED was a predictor of higher vomiting frequency
ASGB 36 ⫾ 11 years, dor, 0% dietary recall, vomiting and higher rate of neostoma stenosis; vomiting
diary frequency determined the rate of medical
complications; no predictor of weight loss: BED
Busetto et al. (53), 250 (women, 260; men, 188; 36 Clinical interview BMI, 9.8 Positive predictor of weight loss: age ⬍ 40 years,
LAGB ma, 72; 37.6 ⫹ 10.8 BMI ⬍50 kg/m2; negative predictor: male sex,
Psychosocial Predictors of Obesity Surgery, Herpertz et al.

years); dor, 4% nonsweet eating behavior

Davidson et al. (57), 18 (22); women, 16; men, 2; 33, range, Clinical interview 40.0 kg No predictor of weight loss: psychiatric
VBG ma, 35.0 years; dor, 8.9% 6 to 90 abnormality
Dixon et al. (33), 320; women, 459; men, 398; 12 Clinical interview, BMI, ⬃10 Predictor of improved QoL: age, comorbidity
LAGB ma, 71; 41 ⫾ 9.8 years; standardized questionnaire, (especially arthritis/joint pain, depression);
dor, 30.3% self-made questionnaire predictor of low QoL: younger age, developing
a weight problem at a later stage in life;
percentage EWL was not a major predictor of
QoL
Dixon et al. (54), 440 (730); women, 383; men, 12 Clinical interview, BMI, ⬃10 Positive predictor of weight loss: moderate alcohol
LAGB 57; ma, 40.0 ⫾ 9.5 years; standardized questionnaire, consumption; negative predictor: increasing age,
dor, 29.7% self-made questionnaire higher initial BMI, poor physical ability; no
predictor: gender, family history of obesity, age
at onset of obesity, history of medical, mental
illness
Dubovsky et al. (49), 36 (42 of 52 were operated; 26 Semistructured interview 42.5 kg Significant relationship between preoperative
VBG women, 37; men, 15; ma, depression and weight loss (percentage):
39 years); dor, 14.3% psychiatric “disturbances” (acute depression,
anxiety, psychosocial crisis, suicidal ideation)
were associated with less weight loss; no
relationship: age of onset of obesity,

OBESITY RESEARCH Vol. 12 No. 10 October 2004

socioeconomic class

1557
1558
Table 3. continued
Study, authors, Mean
type of surgical follow-up Assessment methods
procedure Sample period and measures Weight loss Predictors

Gentry et al. (44), 33 (33); women, 32; men, 1; 24 Standardized interview, 83% lost ⬎50% of Positive predictor of weight loss: eating in
gastric bypass ma, 32 years; dor, 0% standardized questionnaires excess weight response to pleasant feelings or a sense of
accomplishment, realizing preoperatively that
overeating was the cause of obesity; high levels
of optimism were not associated with weight
loss
Hafner et al. (45), 71 (118); (women, 83; men, 12 Standardized questionnaires 35.4 kg; BMI, 13.1 Positive predictor of weight loss: marital
gastric bypass 35); ma, not indicated; dor, dissatisfaction, increase in phobia scores;
39.8% negative predictor: generalized anxiety, increase
in extrapunitiveness
Kopec-Schrader et al. 45 (60); women, 41; men, 4; 108 Structured interview, self-made BMI, 9.4 No predictor of weight: demographic features, age
(50), VBG ma, 37 years; dor, 25% questionnaire at surgery, family and patient history of obesity,
patient weight at the time of marriage, patient
history of significant life events, emotional

OBESITY RESEARCH Vol. 12 No. 10 October 2004

Psychosocial Predictors of Obesity Surgery, Herpertz et al.

problems, psychiatric treatment, alcohol/drug

abuse
Lang et al. (32), 97 (66, gender not indicated; 12 Standardized questionnaire BMI, 8.4 No predictor of weight loss: presurgical eating
LAGB ma, 38.1 years; dor, 32%) behavior
Larsen (48), 90 (95); women, 66; men, 24; 36 Structured interview, 31.5 kg Positive predictor of weight loss: higher initial
horizontal gastric ma, 35.4 years; dor, 7% standardized questionnaires weight and younger age; negative predictor:
banding presurgery personality disorder; no predictor of
weight loss: overweight in parents, degree of
juvenile obesity and effect of preop. dieting;
preoperative psychiatric help-seeking was
associated with negative psychosocial reaction
despite sufficient weight loss
Larsen and Torgerson 89 (97); women, 65; men, 24; 36 Semistructured interview 31.4 kg No correlation between scores of the personality
(51), gastric ma, 35.4 ⫾ 9.4 years; dor, (DSM-III), standardized questionnaire and weight loss
banding 8.2% questionnaire
Lindroos et al. (29), Gastroplasty, 365 (375; 24 Self-made questionnaire Gastroplasty, ⬃26 kg; No association between intake of sugar and sweet
VBG, gastric women, 248; men, 127); gastric bypass, foods before surgery and weight loss 2 years
banding, gastric gastric bypass, 34 (men, ⬃38 kg after gastroplasty, inverse association between
bypass 12; women, 22; ma, 46 ⫾ 6 weight change and relative intake of sugar and
years; dor, 13%); sweet foods: subjects with a high proportion of
gastroplasty, 0% gastric their intake from sweets lost more weight. They
bypass reduced energy intake from sweets less than
from ordinary food
 Table 3. continued
Study, authors, Mean
type of surgical follow-up Assessment methods
procedure Sample period and measures Weight loss Predictors

Powers et al. (34), 43 (100; women, 85; men, 12 Clinical interview, 43.8 kg No difference in weight loss between hospitalized
VBG comparison of 15; two deaths); ma, 38.8 standardized questionnaires psychiatric group and matched controls;
the group requiring years; dor, 57% predictor of postoperative psychiatric
psychiatric hospitalization: presurgical psychiatric
hospitalization (17%) hospitalization, presence of multiple axis I
with 16 matched diagnosis, untreated axis I diagnosis
patients of the
sample
Powers et al. (17), 81 关131; women, 111; men, 68.4 Clinical interview, 41 kg; BMI, 14.0 No predictor of weight loss: age, gender, presence,
gastric restrictive 20; five deaths (one standardized questionnaires, or absence of a presurgical psychiatric
surgery suicide); ma, 39.4 years兴; self-made questionnaire diagnosis; no relationship between the presence
dor: 38.2% or absence of a presurgical psychiatric diagnosis
and various mental health parameters
Powers et al. (55), 77 (116; five deaths, one 66 Clinical interview, BMI, 13.8 kg/m2 No relationship between presurgical eating
gastric restrictive suicide); women, 64; men, standardized questionnaire, pathology and weight loss or presence of
Psychosocial Predictors of Obesity Surgery, Herpertz et al.

surgery 13; ma, 39.6 ⫾ 9.3 years, self-made questionnaire vomiting at follow-up
dor, 33.6%
Schrader et al. (46), 39 (72; women, 66; men, 6, 36 Semistructured interview 33 kg No predictor of weight loss: age, gender, marital
gastroplasty, ma, 41 years); dor, 45.8% status, socioeconomic status, past psychiatric
gastric, bypass, history, parental/spouse support for surgery,
gastrogastrostomy violent parents/spouse
Sugerman et al. (41), A, 16 (20; women, 18; men, 36 Clinical interview, 24-hour A, 41.3 ⫾ 12.7 kg; B, VBGP: “sweets eaters” lost significantly less
VBG (A) and 2); ma, 38 ⫾ 9 years (one recall, 2-day diary food 27.2 ⫾ 14.5 kg excess weight than “nonsweets eaters”; RYGBP:
RYGB (B) death); dor, 20%; B, 18 frequency check no significant difference between sweets and
randomized (20; women, 18; men, 2; nonsweets eaters
prospective trial ma, 38 ⫾ 11 years; two
deaths); dor, 10%
Sugerman et al. (64), 222 (age and gender not 36 Dietary history (24-hour recall, Weight loss, A, Sweets eating is a negative predictor for weight
sweet eaters (A) indicated); A, 182 (82%); 2-day food diary, food 35 ⫾ 11%; B, loss in the frame of restrictive surgery but not in
3 gastric bypass B, 40 (18%); dor, 15% frequency checklist) 27 ⫾ 13% the frame of bypass surgery procedures
nonsweet eaters
(B) 3 VBG
Valley and Grace 49 (57; women, 51; men, 6; 12.7 Clinical interview, 43 kg No predictor of weight loss but of significant
(56), horizontal ma, 38 years); dor, 14.3% standardized questionnaire, medical complications: prior inpatient (but not
gastroplasty self-made questionnaire outpatient) psychiatric history, elevated scores of
the personality questionnaire, negative life events,

OBESITY RESEARCH Vol. 12 No. 10 October 2004

social support; predictor of psychological
complications: prior inpatient psychiatric history;

1559
no predictor at all: sociodemographic factors
1560
Table 3. continued
Study, authors, Mean
type of surgical follow-up Assessment methods
procedure Sample period and measures Weight loss Predictors

van Gemert et al. 62 (68); women, 44; men, 18; 85.9 ⫾ 4.8 Standardized questionnaire 45 kg Positive predictor of weight loss: self-esteem; no
(47), gastric ma, 33.1 ⫹ 9.4 years; dor, predictor: intelligence, personality features,
bypass, VBG 10% interpersonal behavior
Yale and Weiler (28) 195 (200), F: 150, M: 45; A: 12 Clinical interview, self-made A: 34.0 ⫾ 13.61 Negative predictor of weight loss: consumption of
VBG A: 4.5 cm F: 79, M: 19; ma: questionnaire B: 34.9 ⫾ 10.89 caloried liquid and/or soft foods; no predictor:
circumference B: 38.6 ⫾ 9.1 year B: F: 71, age, gender, unemployment, preoperative mental
5.0 cm M: 26; ma: 35.6 ⫾ 9.2 health problems
circumference years, dor (total sample):
2.5%
Postsurgery assessment only (months after operation)
Cook and Edwards 100 (selection criteria, 0 84 (Phone) interview, self-made 42.2 kg Positive predictor of weight loss: eating three well-
(42), gastric indicated); women, 95; questionnaire balanced meals and two snacks daily, drinking
bypass men, 5; ma, not indicated water, avoiding carbonated beverages, taking

OBESITY RESEARCH Vol. 12 No. 10 October 2004

Psychosocial Predictors of Obesity Surgery, Herpertz et al.

multiple vitamins, iron, calcium, sleeping 7 hours

per night, regular exercise, feeling personally
responsible for weight control; negative predictor:
lack of exercise, poorly balanced meals, grazing
and snacking, drinking carbonated beverages
Delin et al. (20), 20; women, 16; men, 4; ma, 24 Semistructured interview, Excess weight loss, Positive predictor of weight loss: self-esteem;
gastric bypass 42.5 ⫾ 9.4 years; dor, not 42 ⫾ 9.13 standardized questionnaires 89.57 ⫾ 23.88% valuing of food and eating, personality features;
indicated negative predictor: depression, disinhibition (EI
subscale), personality features; no predictor: age,
anxiety, social support, employment status
Hudson et al. (43), 200; women, 163; men, 37; At least 12 Standardized questionnaire, 0 Indicated No predictor of weight loss: preoperative sweets
LAGB, normal 41.5 ⫾ 9.9 years; controls, self-made questionnaire eating
weight community 40; women, 26; men, 14;
controls 37.0 ⫾ 10
Mitchell et al. (31), 78 (100; eight deaths, two 156 to 195 Standardized questionnaires 49.7 kg; BMI, 10 Redeveloping problems with BED postsurgery led
RYGB died from a psychiatric- to tendency to regain weight
related condition; women,
65; men, 13); ma, 56.8
years; dor, 22%
Pekkarinen et al. 27 (33; one suicide; women, 64.8 Standardized questionnaires, 31 kg, BMI: 12 Negative predictor of weight loss: binge eating; no
(25), VBG 19; men, 8); ma, 36 years; daily food records predictor of weight loss: age, sex, age at onset
dor, 10% of obesity; frequency of vomiting did not
correlate to weight loss
Psychosocial Predictors of Obesity Surgery, Herpertz et al.

MMPI traits correlated with weight loss after surgery.

Negative predictor of weight loss: BED; negative

predictor of physical well-being: age and BMI

Larsen and Torgerson (51), using a personality question-
naire (Basic Character Inventory) (60), were not able to

predictor of mental well-being: BED; no

demonstrate any correlation between preoperative levels in

ASGB, adjustable silicone gastric banding; dor, drop out rate; LASGB, laparoscopic adjustable silicone gastric banding; ma, mean age; NES, night eating syndrome.
the score of personality traits and degree of weight change.
Valley and Grace (56) examined the predictive utility of
several classes of variables including psychiatric history and
Predictors

the presence of any clinically elevated MMPI scale. They

found that inpatient psychiatric admissions and elevated
MMPI scales each predicted postoperative medical compli-
cations and psychological disturbance, but neither could
predict weight loss. Barrash et al. (58) provided a cluster
at follow-up

analysis of 138 MMPI profiles of women with a subsequent

vertical banded gastroplasty (VBG) and a weight loss fol-
low-up at 12 months. MMPI types with preoperative profile
patterns indicative of the most severe psychological prob-
lems demonstrated poor weight loss. Delin et al. (20) used
the Adjective Checklist (61) in their study on 20 gastric
postoperative: 10
preoperative: 4.6
Weight loss

bypass patients with a mean follow-up period of 2 years.

Percentage of weight lost correlated significantly with sev-
eral of the Adjective Checklist subscales (significant corre-
lations, r ⬎ 0.45: affiliation, heterosexuality, military lead-
BMI:

ership, nurturing parent).

Personality variables were summarized, with the excep-
standardized questionnaires,

tion of one study (20), solely by means of the MMPI. With

Assessment methods

one exception, no connection could be demonstrated be-

tween the scales of the MMPI and the postsurgery course of
and measures

156 to 195 Structured interview,

weight.

Psychiatric Status
As was shown in our previous review (14), presurgery
prevalence of axis I psychiatric disorders according to the
criteria of the DSM (23), DSM-R (62), varies between
27.3% (44) and 41.8% (63) and of axis II between 22% and
follow-up
period
Mean

24% (17,34,48). Axis I diagnoses, when divided into the

single psychiatric disorders (44,48,63), were predominantly
affective disorders, anxiety disorders, and eating disorders.
Compared with presurgery status, prevalence rates of axis I
related condition); women,

psychiatric disorders were considerably less at follow-up,

died from a psychiatric-
de Zwaan et al. (24), 78 (100; eight deaths, two

preoperative group, 110

(164; women, 96; men,
65; men, 13; ma, 56.8

ranging from no diagnosis (44), one-half (48), or one-third

14); ma, 39.6 years

(63) of the presurgical rate. Personality disorders did not

Sample

years; dor, 22%;

change substantially after surgery (48,63). Because of their

considerable importance, eating disorders will be addressed
separately.
With the exception of one study (41), which compared
VBG with Roux-en-Y gastric bypass (RYGB) in a random-
ized prospective trial, all 18 studies assessing the effects of
Table 3. continued

psychiatric symptoms or comorbidity (including disordered

type of surgical
Study, authors,

eating behavior) on weight loss after obesity surgery were

group, RYGBP
(Mitchell et al.
procedure

postoperative
Cross-sectional

nonrandomized and noncontrolled clinical trials either with

assessment

a prospective (17,28,30,34,45,46,48 –50,52,54 –57) or retro-

sample)

spective (20,24,31) design. Psychiatric disorders or symp-

toms were diagnosed by standardized, (semi-)structured, or
clinical interviews. Within the group of prospective studies,

OBESITY RESEARCH Vol. 12 No. 10 October 2004 1561

Psychosocial Predictors of Obesity Surgery, Herpertz et al.

only four studies suggested an impact of psychiatric comor- psychiatric symptoms and required hospitalization (34). The
bidity and symptoms on postsurgery weight loss (45,48, hospitalized group was compared with a matched group
49,52), whereas the other 10 studies failed to find any drawn from the original 100 patients. The two groups were
association. The two retrospective studies (18,36) demon- similar in terms of presurgical weights and weight lost 3 and
strated an association between psychiatric variables and 12 months after surgery. However, the hospitalized group
course of weight. However, a considerable number of stud- was more likely to have had previous psychiatric hospital-
ies found a significant association between presurgery di- izations, multiple axis I diagnoses, and untreated axis I
agnoses of psychiatric symptoms and disorders and various diagnoses at the time of the presurgical assessment. Similar
mental health parameters at follow-up (20,24,30,31,34, findings were obtained by Davidson et al. (57), who were
48,56). also not able to demonstrate any effect of major psychiatric
Dubovsky et al. (49) observed a statistically significant symptoms on the postoperative course of weight; however,
positive correlation between the degree of clinically esti- the occurrence of postoperative psychiatric problems corre-
mated preoperative depression and the percentage of body lated closely with preoperative psychological assessment.
weight loss after surgery, speculating that patients who do Hafner et al. (45), assessing 71 patients undergoing gastric
not express as much distress before surgery may be less bypass, identified generalized anxiety and phobia to be
dissatisfied with their weights and less willing to extend negatively associated with weight loss after a follow-up
themselves in the service of postoperative weight loss than period of 1 year. In contrast, Delin et al. (20), in their
patients who are more upset by their obesity and more retrospective study of 20 gastric bypass patients with a
motivated to diet once barriers to weight loss are reduced by 2-year follow-up, could not find anxiety to be associated
surgery. However, psychiatric disturbances such as depres- with weight loss.
sion, anxiety, psychosocial crisis, and suicidal ideation that In a qualitative comparison of the criteria of the individ-
did not seem to be an imminent result of obesity were ual studies (prospective study design, adequate sample size,
associated with less weight loss; this was also confirmed by and low drop-out rate), the study of Larsen (48) must be
Hafner et al. (45). Brolin et al. (52) found a level of emphasized. Because of the large sample of 103 patients
psychological stress to be positively related to weight loss. undergoing horizontal gastric banding surgery and a small
Valley and Grace (56) tried to isolate psychological predic- drop-out rate of only 7% after 3 years, the Norwegian study
tors of outcome in 57 patients with horizontally reinforced is of major importance with regard to possible psychiatric
gastric surgery 1 year after surgery. Prior inpatient psychi- predictors of postsurgery weight loss. All 103 patients were
atric history (but not outpatient history), negative life interviewed and rated on DSM (23). Larsen (48) identified
events, and low social support related significantly to med- a group with insufficient weight loss that had a significantly
ical complications. Prior inpatient psychiatric history and higher frequency of personality disorder diagnoses with no
low social support also predicted psychological complica- predominance of any distinct personality disorder compared
tions postoperatively. However, there were no significant with the rest of the patients. Moreover, a group of patients
relationships between weight loss and psychopathology. with negative psychosocial outcome was found. This group
Dixon et al. (54), Kopec-Schrader et al. (50), Schrader et al. was characterized by a higher frequency of preoperative
(46), and Yale and Weiler (28) investigated prospectively psychiatric help-seeking than the rest of the patients. Inter-
samples of obesity surgery patients ranging from 60 to 440 estingly, this pattern was most evident in a subgroup of
patients for a period of at least 1 year; however, they were these patients who had a negative psychosocial reaction
not able to identify a history of mental illness or measures despite sufficient weight loss. Preoperative psychopathol-
of mental health to be predictive of weight loss. Davidson et ogy, defined as contact with health services for psychiatric
al. (57) reviewed patients undergoing VBG with regard to problems or manifest personality disorder, was associated
preoperative psychological assessment and postoperative with negative outcome results. However, the degree of
outcome, grouping them into three broad categories (no
contact with health services for psychiatric reasons before
psychiatric abnormality, minor psychiatric disorders, major
the operation was highly correlated to the score on the same
psychiatric disturbance). Psychiatric morbidity had no ad-
variable in the follow-up period and not associated with the
verse effect in terms of postoperative weight loss or post-
degree of weight loss.
operative complications such as vomiting. Similar results
were obtained by Powers et al. (17), who found neither a
relationship between the presence or absence of a presurgi- Eating Patterns and Eating Pathology
cal psychiatric diagnosis and weight loss nor an influence of Ten studies (28 –30,32,41,44,52,53,55,64) evaluating eat-
various mental health parameters on weight loss in a sample ing patterns and eating pathology as possible predictors of
of 131 patients after a follow-up period of more than 5 weight loss had a prospective design, whereas six studies
years. In another study, they assessed psychiatric status of (20,25,31,42,43) had a retrospective design or compared
100 patients before surgery; of these, 17% developed severe different subsamples pre- and postsurgery (24). Four studies

1562 OBESITY RESEARCH Vol. 12 No. 10 October 2004

Psychosocial Predictors of Obesity Surgery, Herpertz et al.

investigated postsurgery eating patterns as a possible pre- showed that no association was observed between intake of
dictor for postoperative course of weight (28,29,31,32). sugar and sweet foods before surgery and weight loss at 2
Not only patients suffering from eating disorders, espe- years after surgery. Similar findings were obtained by
cially BED, but also sweets eaters, and here especially those Busetto et al. (30) studying 260 patients prospectively over
who undergo gastric restrictive surgery, are believed to have a period of 36 months and Hudson et al. (43) studying 200
poor outcomes. Sugerman et al. carried out two studies patients within the scope of a retrospective evaluation. All
(41,64): VBG was compared with RYGB in a randomized these authors challenged the hypothesis that sweets eating is
prospective trial that included preoperative dietary separa- predictive of the outcome after gastric restrictive surgery.
tion of sweets eaters vs. nonsweets eaters (41). The patient They found that sweets eaters were just as likely to be
was classified as a sweets eater if he or she consumed ⬎300 successful as nonsweets eaters.
calories of sweet foods or beverages more than three times Apart from the evaluation of the foods preferred, e.g.,
per average week. Greater weight loss was demonstrated foods high in simple sugars and calorie-containing bever-
with gastric bypass, supposedly due to the development of ages, three studies (20,32,44) investigated eating behavior
dumping syndrome symptoms in sweets eaters after gastric as a possible predictor of postsurgery weight loss. Gentry et
bypass. It was noted that sweets eaters lost significantly less al. (44) showed that the patients who ate in response to
weight than nonsweets eaters with VBG 1 year after sur- pleasant feelings or a sense of accomplishment lost more
gery. In the second study (64), Sugerman classified “sweets weight after operation than did those who ate in response to
eaters” as subjects consuming ⬎15% of total caloric intake stress, suggesting the institution of new reward systems not
in the form of sweet foods such as candy, cake, ice cream, based on food. Delin et al. (20) found a negative relation-
or caloric beverages. Nonsweets eaters were assigned to ship between the Three-Factor Eating Questionnaire sub-
VBG and sweets eaters to gastric bypass. Weight loss after scale “disinhibition” and percentage weight lost, i.e., sub-
VBG was significantly improved 2 years postoperatively, jects who tend not to respond to emotional disinhibitors
with selective assignment of all sweets eaters to gastric such as anxiety or loneliness by eating lost a greater per-
bypass without any loss of efficacy in the gastric bypass centage of excess weight after surgery.
group. Although gastric bypass was statistically more effec- Examining pre- and postsurgery predictors, Lang et al.
tive with regard to percentage excess weight lost than VBG, (32) assessed the eating behavior of 66 obese patients before
it was associated with a greater incidence of considerable and 1 year after laparoscopic adjustable gastric banding
side effects such as stomal stenosis, marginal ulcer, and (LAGB) by means of standardized questionnaires such as
vitamin B12 deficiency. Yale and Weiler (28) assessed the German version of the Three-Factor Eating Question-
eating behavior by means of a four-part scale (volume of naire (66). Results indicated significant changes after sur-
solid food, amount of liquid calories consumed per day, gery, with dramatically increased levels of “cognitive re-
frequency of eating large amounts of soft food, and snack- straint” and “flexible control” and simultaneous decreases
ing between meals). Postoperative eating behavior dramat- of disinhibition and “hunger”; however, the change in the
ically affected weight loss, although the food consumed did eating behavior as a postoperative predictor did not seem to
not exclusively have a sweet taste but also consisted of be directly associated with the observed weight loss.
“chips and dips.” Among the patients who lost ⬍20% of Especially with regard to the controversial discussion as
their original weight, 25% drank large volumes of calories to whether sweets eating has a predictive function for the
as liquids, whereas 75% ate large amounts of soft foods. course of weight, three carefully designed studies must once
Cook and Edwards (42) studied the eating behavior of a more be emphasized (29,41,64): The results of the study of
sample of 100 patients 1 year after gastric bypass. Success- Lindroos et al. (29) may be contrasted with two studies by
ful patients, defined as those who maintained at least 74% of Sugerman et al. (41,64). One fundamental difference be-
the initial weight loss, ate three well-balanced meals and tween the Swedish study (29) and Sugerman’s studies
two snacks daily, drank water, and avoided carbonated (41,64) was the difference in design: Sugermans’s patients
beverage of any kind. Apart from the issue of sweets or were randomized (41) or selected (64) for type of operation
nonsweets eaters, Brolin et al. (52) demonstrated that the based on dietary preferences, whereas the surgical part of
number of preoperative food contacts in the sense of fre- the SOS study (29,65) was observational in the context of
quent snacking per day was likely to be associated with diet. Furthermore, Sugerman et al. (41,64) reported primar-
inadequate weight loss. In an extensive study, Lindroos et ily on intake of sweet foods before surgery, whereas the
al. (29) used data from the Swedish Obese Subjects (SOS) SOS study (29,65) also monitored changes in intake pat-
study (65) to test the hypothesis that a diet containing many terns after surgery. Another potential difference is of con-
sweet foods is associated with poor weight loss after gas- cern when considering the three studies: Sugerman et al.
troplasty. A total of 375 subjects were followed for 2 years (64) classified sweets eaters as subjects consuming ⬎15%
after VBG or gastric banding, and questionnaires were used of total caloric intake in the form of sweet foods. Lindroos
to evaluate diet both before and after surgery. Results et al. (29) pointed out that in the SOS study, 62% of the

OBESITY RESEARCH Vol. 12 No. 10 October 2004 1563

Psychosocial Predictors of Obesity Surgery, Herpertz et al.

women consumed ⬎15% of their calories in form of sweet which was consistent with the recurring theme during the
foods. By this definition, differences between sweets eating interviews that patients purposely overate knowing that it
patterns in obese and nonobese women were fairly small, might result in vomiting. The study of Mitchell et al. (31),
and, more important, nonsweets eaters were a minority in although reporting retrospectively, has several strengths in-
both groups. Therefore, other important selection criteria for cluding a standardized questionnaire, a sample size of 100
the Sugerman studies must be discussed; for example, pa- patients undergoing gastric bypass surgery, and a long-term
tients who were unusually health-conscious or cooperative (13- to 15-year) follow-up with a relatively small drop-out
might have been selected for VGB by applying the criteria rate of 12%. With regard to BED, the majority of individ-
of ⬍15% sweet foods as a cut-off point. uals who met criteria for BED did not meet such criteria at
long-term follow-up, even when the criterion for eating a
Binge Eating large amount of food was excluded. With regard to post-
A substantial percentage of bariatric surgery patients suf- surgery predictors, it is relevant that those who developed
fer from BED as a newly defined provisional eating disorder BED symptoms after the procedure tended to regain more
category included in the DSM (67) or BE symptoms in the weight. Furthermore, de Zwaan et al. (24), using a cross-
sense of overeating episodes that are accompanied by both sectional design, compared the Mitchell et al. (31) sample of
a subjective loss of control and significant emotional dis- 78 patients with a preoperative control group of 110 pa-
tress. Prevalence rates range from 7.3% (68) to 49% tients. They showed that patients suffering from BED had a
(31,69), respectively. Of the six studies explicitly evaluating significantly higher BMI at follow-up, and their weight loss
BED as a possible predictor of weight loss and mental since surgery was significantly less than that of non-BED
health, three studies used a prospective design (30,53,55), patients. Similar results were obtained by Pekkarinen et al.
two a retrospective design (25,31), and one study compared (25) in their retrospective but questionnaire-based study
different subsamples pre- and postsurgery (24). Busetto et assessing a smaller sample of 25 patients who had under-
al. (30,53) performed two studies analyzing the relation- gone VBG more than 5 years previously. BE behavior
ships among BED, eating pattern, vomiting frequency, emerged as the main predictive factor for poor weight loss.
weight loss, and the rate of band-related complications in It is striking that with regard to BED and the postopera-
patients undergoing laparoscopic adjustable gastric band- tive course of weight, the three prospective studies
ing. In the earlier study (30), patients with BED had a (30,53,55) were not able to demonstrate any connection;
significantly higher vomiting frequency and a 5-fold higher however, the two retrospective studies (25,31) demonstrated
frequency of neostoma stenosis than patients without BED. BED to be a negative predictor. Only Mitchell et al. (31)
However, the percentage of overweight lost did not differ investigated patients with gastric bypass, whereas the four
between patients with or without BE. Vomiting was a major other studies (25,30,53,55) were concerned exclusively with
determinant of global outcome after adjustable silicone gas- restrictive surgery procedures. Concerning the question of
tric banding. Interestingly, in the later study (53), prior BED whether BE represents a negative predictor for the postop-
did not significantly affect the 3-year outcome. The authors erative course of weight, the follow-up period and the
assumed that the preoperative psychological treatment of differentiation between pre- or postsurgery predictors are
patients in the second study with postoperative reinforce- relevant. Thus, Pekkarinen’s patients (25) did not differ in
ment affected the prognosis favorably. Powers et al. (55) body weight 1 year after surgery; this corresponds to the
assessed the relationship of presurgical eating pathology follow-up period of the Busseto (30,53) studies. However,
and weight outcome in 116 patients before bariatric surgery more than 5 years later, weight regain was significantly
(gastric restrictive procedures). The prevalence of BE was higher in the BE compared with the non-BE patients. Fi-
52% presurgery. After an average follow-up period of 5 nally, it is important to point out that in the Mitchell et al.
years, there was no reported BE among any of the patients, study (31), only those who redeveloped BE symptoms post-
probably related to the fact that the gastric restrictive pro- surgery regained weight. However, in the Powers et al. (55)
cedures make it physiologically very difficult to binge eat. study, not a single case of binging could be detected at
Psychopathology was greater in the BE group compared follow-up.
with the non-BE group. There was evidence of more cog-
nitive distortions, greater body image disturbance, and an Other Psychosocial Variables
increased level of depression for this group. Although no The predictive value of other psychosocial variables has
relationship between presurgical eating pathology and been examined. Marital dissatisfaction (45), high levels of
weight outcome or presence of vomiting could be observed life stress (52), and low self-esteem (20,47) have been
at follow-up, the authors described a new eating pathology. shown to be positive predictors of weight loss at follow-up.
Of the patients, 79% were occasionally vomiting, and one- Valley and Grace (56) and Delin et al. (20) could not
third were vomiting at least weekly. Powers et al. (55) identify social support to be of predictive value for weight
postulated that some patients would binge eat if they could, loss postsurgery.

1564 OBESITY RESEARCH Vol. 12 No. 10 October 2004

Psychosocial Predictors of Obesity Surgery, Herpertz et al.

Age and Presurgery Body Weight reduction defined as ⬍20% to 30% (15–17,19,20) com-
Ten studies (33,46,48,50,52–54,56 –58), all with a pro- pared with these weight loss results, the question as to
spective design, have evaluated age as a possible predictor possible predictors of postoperative weight loss is of sub-
of weight loss after bariatric surgery, yielding inconsistent stantial interest. Besides weight reduction and the associ-
data. Six studies (33,48,52–54,58) reported age to be sig- ated decrease in medical risk factors, one further goal of
nificantly correlated with absolute weight loss. Younger obesity surgery is the improvement of QoL where medical,
patients tended to lose more weight postsurgery. Four stud- psychological, and psychosocial variables merge.
ies (46,50,56,57), however, found age not to be predictive of All reviewed studies reported substantial weight loss after
postsurgery weight loss. With regard to the predictive value, surgery. In terms of predicting the course of weight, post-
the studies did not differ considerably in the mean age of the surgery data are inconsistent. However, some conclusions
single samples. Presurgery body weight is highly correlated can be drawn.
with weight loss; however, the absolute weight at follow-up Either personality traits do not predict postoperative
is of major importance. Therefore, only studies indicating course of weight (20,44,51,56,58) or the definition of per-
the absolute weight at follow-up were considered (45,52– sonality used in the respective studies was too broad (20) for
54,57,58). All six studies measuring absolute weight at application in daily clinical routine. Evidence of the rela-
follow-up found that patients who were more overweight tionship of psychiatric disorders or psychopathologic symp-
preoperatively had less successful weight loss. toms and the course of postsurgery weight is meager. Con-
trary to clinical assumptions, the majority of studies could
Predictors of QoL not identify psychiatric comorbidity as a negative predictor
Dixon et al. performed two studies (33,54) assessing QoL for weight loss. In single studies, increased psychological
as measured by the SF-36 questionnaire in 218 and 383 distress as assessed through higher levels of depression
patients, respectively, before and 1 year after Lap-Band (20,49) and anxiety (20,45,49), elevated SCL-90 profiles
placement. Poor physical ability as a major aspect of QoL (52), elevated phobia scores (45), and low self-esteem
before surgery was a clear predictor for a lower rate of (20,47) before surgery even appeared to be positively asso-
weight loss on the postoperative course of weight (54). ciated with weight loss after surgery. If at all, poor weight
Conversely, the percentage excess weight loss (EWL) at 1 loss after surgery can be observed in patients suffering from
year follow-up was not a major (postsurgery) predictor in serious and chronic psychiatric disorders requiring inpatient
improved QoL (33). In contrast, de Zwaan et al. (24) com- psychiatric treatment (56) and personality disorders (48),
pared the Mitchell et al. (31) sample of 78 patients with a which are distinguished by a lack of introspection. A cau-
preoperative control group of 110 patients with regard to tious interpretation of the partly conflicting results could be
QoL, also using the SF-36 questionnaire. They showed that that, contrary to distress independent of or only reinforced
reduction in BMI between baseline and a ⬎13-year fol- by obesity, obesity-associated psychological distress such
low-up accounted for a significant proportion of variance in as low self-esteem, depression and anxiety, and social pho-
physical well-being. Patients with preoperative arthritis/ bia resulting in social isolation probably predicts more
joint pain had greater improvement in most aspects of QoL postsurgical weight loss. Patients suffering from serious and
than did other patients with measures close to those of chronic psychiatric diseases such as personality disorders
community normal values. Patients reporting a history of might have more difficulties in adapting to the major de-
depression preoperatively also improved in QoL; further- mands of controlled eating behavior imposed on them by
more, they demonstrated that BED after surgery negatively the operation. Apart from the question of whether psycho-
influenced mental well-being as measured by the mental logical distress results from obesity or vice versa, i.e.,
composite scores of the SF-36 (24). whether obesity results from a dysfunctional regulation of
negative affects or is more an independent variable, the
severity of the emotional problems and not the presence or
Discussion absence of emotional problems might be predictive of
The superiority of obesity surgery, as opposed to conser- weight loss. As Vallis and Ross (71) pointed out almost a
vative weight reduction measures, with regard to improving decade ago, distress based on serious psychiatric distur-
the medical (8 –12) and psychosocial outcome variables bance is probably an impediment to obesity surgery,
(14) in subjects with obesity grade III is undisputed. Weight whereas distress based on the experience of being morbidly
loss is characteristically on the order of one-third of preop- obese is probably a positive predictor of outcome. Even if
erative weight, which corresponds to 55% to 65% of excess prevalence of psychiatric disorders except personality dis-
weight (70). EWL is greater after gastric bypass than after orders is decisively less after surgery (14,44,48,63), there
simple gastroplasty, and more malabsorptive operations, seems to be a significant association between at least serious
such as biliopancreatic bypass, occasionally result in losses presurgery psychiatric symptoms and various mental health
of 80% of excess weight. In view of an inadequate weight parameters (20,24,31,34,48,53,56) or medical complica-

OBESITY RESEARCH Vol. 12 No. 10 October 2004 1565

Psychosocial Predictors of Obesity Surgery, Herpertz et al.

tions (56) at follow-up. According to Larsen (48) and Valley attempts to binge eat, as Powers et al. (55) have hypothe-
and Grace (56), personality disorders or a prior inpatient sized, has yet to be investigated in further studies.
psychiatric history had a predictive value for the postoper- This systematic review has several limitations: Compar-
ative psychiatric course. The study of Larsen (48) not only ison of the individual studies entails several problems based
showed that personality disorders were associated with in- on the study designs, methods of operation, investigation
sufficient weight loss but also demonstrated that patients instruments, and observation periods. Even if according to
with negative psychosocial reaction despite a sufficient our exclusion criteria, studies except one (34) with a drop-
weight loss were characterized mainly by a high frequency out rate of ⬎50% were not included in our study, the
of preoperative psychiatric help-seeking. One explanation average drop-out rate was 20%. Nevertheless, there was a
may be that these patients suffer from two different dis- wide range from 0% to 50%, so that for many studies the
eases, strictly speaking, that psychopathology might be un- question arises as to how representative they are.
related to body weight or obesity. Another explanation may Gastric bypass is superior to purely restrictive types of
be that although patients with personality disorders do not surgery for weight reduction; however, because few studies
manage to restrict their eating behavior after the operation are, at present, available that have investigated gastric by-
and, therefore, do not lose weight adequately, the group pass and restrictive surgery in subgroups of obese patients,
with major mental disorders complies with the required it is yet too early to discuss a possible specificity of predic-
eating behavior after surgery and loses weight but develops tors for different surgical procedures. Finally, it must be
an increase in psychiatric symptoms or a decrease in general understood that patients in the vast majority of studies
well-being instead (34). Also with regard to eating patterns survived a number of selection biases and cannot be re-
and eating pathology, it seems to be less the specificity, e.g., garded as representative of all morbidly obese patients
the question as to the extent of the consumption of sweets seeking obesity surgery. They have actively pursued weight
(18), but rather the extent of inadequate or disturbed eating reduction, were selected for referral to a surgeon by their
general practitioner, and, finally, were frequently inter-
behavior that influences the postoperative weight loss. Al-
viewed by a psychiatrist or psychologist before acceptance
though in the majority of the studies, sweets eaters were not
for surgery. Therefore, most results presented here are based
regarded to be a risk group for the postoperative course of
not on random but on highly selected samples, making it
weight (29,43,53), patients with a generally hypercaloric
difficult to draw general conclusions for clinical practice.
presurgery intake of foods such as sweets, chips, soft drinks
Because of these selection biases, we can assume that the
etc., which may be regarded as inadequate compliance,
outcome results from which possible predictors have been
might have an increased risk for a negative course of weight
derived, for example with regard to psychiatric comorbidity,
postsurgery (28,42,52). One important predictor may also
rather represent false positive than negative results. Never-
be the motivation on which the eating behavior is based.
theless, it seems that apart from serious psychiatric disor-
Eating as a way and means of reducing stress may be ders, including personality disorders, the predictive value of
regarded as a negative predictor for the postoperative course a psychiatric diagnosis according to DSM or ICD criteria
of weight as opposed to a controlled enjoyable eating be- with regard to weight loss postoperatively has been overes-
havior (44). However, an excessive energy intake in both timated in the past. Body weight is regulated by more-or-
quantity and quality may not only be an expression of less unknown complex cybernetic mechanisms that are af-
inadequate compliance but may also be determined biolog- fected by both psychological and biological-genetic factors
ically. For example, intake of large quantities of food may (75,76) Psychiatric comorbidity is only one aspect in this
be genetically determined. Recently, mutations in the mela- multidimensional cybernetic system. Furthermore, psychi-
nocortin receptor-4 gene have been implicated in BE (72) atric comorbidity may be a completely independent disor-
and a poor prognosis (73). Although these findings are der. On the other hand, it may have a decisive influence on
viewed controversially (26,74), they nevertheless point to body weight in the sense that eating has regulatory functions
the possibility that the aberrant eating behavior associated for unbearable emotions. Considering possible predictors
with a poor prognosis after bariatric surgery might, in part, for weight loss postsurgery, the sole diagnosis of a psychi-
have a genetic basis. atric disorder may be insufficient. What is additionally
With regard to BED being a negative predictor on the required is the assessment of the effect of a given psychi-
postoperative course of weight, the discrepancy between atric diagnosis on quantity and quality of the patient’s eating
prospective and retrospective studies may be resolved in behavior. Psychiatric comorbidity may also serve as an
that the redevelopment of BED postsurgery rather than BED important tool to estimate mental and physical well-being,
before surgery is of predictive value. Furthermore, restric- which are both essential aspects of QoL. There is an in-
tive surgery procedures are ineffective if pureed or hyper- creasing demand for obesity surgery to be performed only
caloric liquid food (e.g., chocolate or soft drinks) are con- by specialists in specialized centers. It seems appropriate
sumed. Whether vomiting is the consequence of failed that these centers cooperate with mental health profession-

1566 OBESITY RESEARCH Vol. 12 No. 10 October 2004

Psychosocial Predictors of Obesity Surgery, Herpertz et al.

als who are familiar with obesity in general and obesity multidisciplinary management of severe obesity. Am J Surg.
surgery-related problems in particular. 1995;169:361–7.
Future research should focus less on psychiatric comor- 16. Hsu LK, Benotti PN, Dwyer J, et al. Nonsurgical factors that
bidity and general psychological constructs such as person- influence the outcome of bariatric surgery: a review. Psycho-
som Med. 1998;60:338 – 46.
ality disorders as possible predictors of postsurgery weight.
17. Powers PS, Rosemurgy A, Boyd F, Perez A. Outcome of
Instead, specific aspects of the eating behavior with special gastric restriction procedures: weight, psychiatric diagnoses,
consideration of both motivation systems and the relation- and satisfaction. Obes Surg. 1997;7:471–7.
ship of energy intake and consumption should be evaluated. 18. Sugerman HJ, Kellum JM, Engle KM, et al. Gastric bypass
for treating severe obesity. Am J Clin Nutr. 1992;55:560 – 6S.
19. Pessina A, Andreoli M, Vassallo C. Adaptability and com-
Acknowledgments pliance of the obese patient to restrictive gastric surgery in
This work was supported by The Deutsche Forschungs- short term. Obes Surg. 2001;11:459 – 63.
gemeinschaft (German Research Foundation) Grants HE 20. Delin CR, Watts JM, Bassett DL. An exploration of the
2665/2-1 and HE 2665/2-2. outcomes of gastric bypass surgery for morbid obesity: patient
characteristics and indices of success. Obes Surg. 1995;5:
References 159 –70.
1. World Health Organization. Obesity Epidemic Puts Mil- 21. Black DW, Yates WR, Reich RH, Bell S, Goldstein RB.
lions at Risk from Related Diseases. Geneva, Switzerland: DSM-III personality disorder in bariatric clinic patients. Ann
Clin Psych. 1989;1:33–7.
World Health Organization; 1997.
22. Black DW, Goldstein RB, Mason EE. Prevalence of mental
2. National Institutes of Health. Clinical guidelines on the
disorder in 88 morbidly obese bariatric clinic patients. Am J
identification, evaluation, and treatment of overweight and
Psychiatry. 1992;149:227–34.
obesity in adults: the evidence report. Obes Res. 1998;6(Suppl
23. American Psychiatric Association. Diagnostic and Statisti-
2):51–209S.
cal Manual of Mental Disorders: DSM-III. 3rd ed. Washing-
3. Seidell JC, Deerenberg I. Obesity in Europe: prevalence and
ton, DC: American Psychiatric Association; 1980.
consequences for use of medical care. Pharmacoeconomics.
24. de Zwaan M, Lancaster KL, Mitchell JE, et al. Health
1994;5:38 – 44.
related quality of life in morbidly obese patients: effect of
4. Ayyad C, Andersen T. Long-term efficacy of dietary treat-
gastric bypass surgery. Obes Surg. 2002;12:773– 80.
ment of obesity: a systematic review of studies published
25. Pekkarinen T, Koskela K, Huikuri K, Mustajoki P. Long-
between 1931 and 1999. Obes Rev. 2000;1:113–9.
term results of gastroplasty for morbid obesity: binge-eating as
5. NHLBI Obesity Education Initiative Expert Panel. Obes
a predictor of poor outcome. Obes Surg. 1994;4:248 –55.
Res. 1998;6:96 –9.
26. Herpertz S, Siffert W, Hebebrand J. Lack of association of
6. Wadden TA. The treatment of obesity: an overview. In: melanocortin-4 receptor gene mutations and binge eating dis-
Stunkard AJ, Wadden TA, eds. Obesity: Theory and Therapy. order. N Engl J Med 2003;349:6.
2nd ed. New York: Raven Press; 1993, pp. 197–217. 27. Herpertz S, Saller B. Psychosomatic aspects of obesity.
7. Monteforte MJ, Turkelson CM. Bariatric surgery for morbid Psychother Psychosom Med Psychol. 2001;51:336 – 49.
obesity. Obes Surg. 2000;10:391– 401. 28. Yale CE, Weiler SJ. Weight control after vertical banded
8. Greenway FL. Surgery for obesity. Endocr Metab Clin North gastroplasty for morbid obesity. Am J Surg. 1991;162:13– 8.
Am. 1996;25:1005–27. 29. Lindroos AK, Lissner L, Sjöstrom L. Weight change in
9. Pories WJ, MacDonald KG. The surgical treatment of mor- relation to intake of sugar and sweet foods before and after
bid obesity. Curr Opin Gen Surg. 1993; 195–205. weight reducing gastric surgery. Int J Obes. 1996;20:634 – 43.
10. Herbst CA, Hughes TA, Gwynne JT, Buckwalter JA. Gas- 30. Busetto L, Valente P, Pisent C, et al. Eating pattern in the
tric bariatric operation in insulin-treated adults. Surgery. 1984; first year following adjustable silicone gastric banding
95:209 –14. (ASGB) for morbid obesity. Int J Obes Relat Metab Disord.
11. Pories WJ, Swanson MS, MacDonald KG, et al. Who 1996;20:539 – 46.
would have thought it? An operation proves to be the most 31. Mitchell JE, Lancaster KL, Burgard MA, et al. Long-term
effective therapy for adult-onset diabetes mellitus. Ann Surg. follow-up of patients’ status after gastric bypass. Obes Surg.
1995;222:339 –50. 2001;11:464 – 8.
12. Gleysteen JJ. Results of surgery: long-term effects on hyper- 32. Lang T, Hauser R, Buddeberg C, Klagehofer R. Impact of
lipidemia. Am J Clin Nutr. 1992;55:591–3S. gastric banding on eating behavior and weight. Obes Surg.
13. Mason EE, Renquist KE, Jiang D. Perioperative risks and 2002;12:100 –7.
safety of surgery for severe obesity. Am J Clin Nutr. 1992;55: 33. Dixon JB, Dixon ME, O‘Brien PE. Quality of life after
573– 6. Lap-Band placement: influence of time, weight loss, and co-
14. Herpertz S, Kielmann R, Wolf AM, Langkafel M, Senf W, morbidities. Obes Res. 2001;9:713–21.
Hebebrand J. Does obesity surgery improve psychosocial 34. Powers PS, Boyd F, Blair CR, Stevens B, Rosemurgy A.
functioning? Int J Obes. 2003;27:1300 –14. Psychiatric issues in bariatric surgery. Obes Surg. 1992;2:
15. Benotti PN, Forse RA. The role of gastric surgery in the 315–25.

OBESITY RESEARCH Vol. 12 No. 10 October 2004 1567

Psychosocial Predictors of Obesity Surgery, Herpertz et al.

35. Brolin RE. Critical analysis of results: weight loss and quality 52. Brolin RE, Clemow LP, Kasnetz KA, Fynan TM, Silva
of data. Am J Clin Nutr. 1992;55:577– 81S. EM, Greenfield DP. Outcome predictors after gastroplasty
36. Hsu LK, Betancourt S, Sullivan SP. Eating disturbances for morbid obesity. Nutr Int. 1986;2:322– 6.
before and after vertical banded gastroplasty: a pilot study. Int 53. Busetto L, Segato G, De Marchi F, et al. Outcome predictors
J Eat Disord. 1996;19:23–34. in morbidly obese recipients of an adjustable gastric band.
37. Hsu LK, Sullivan SP, Benotti PN. Eating disturbances and Obes Surg. 2002;12:83–92.
outcome of gastric bypass surgery: a pilot study. Int J Eat 54. Dixon JB, Dixon ME, O‘Brien PE. Pre-operative predictors
Disord. 1997;21:385–90. of weight loss at 1-year after Lap-Band Surgery. Obes Surg.
38. Rowston W, McCluskey S, Gazet JC, Lacey JH, Franks G, 2001;11:200 –7.
Lynch D. Eating behavior, physical symptoms and psycho- 55. Powers PS, Perez A, Boyd F, Rosemurgy A. Eating pathol-
logical factors associated with weight reduction following the ogy before and after bariatric surgery: a prospective study. Int
Scopinaro Operation as modified by Gazet. Obes Surg. 1992; J Eat Disord. 1999;25:293–300.
2:355– 60. 56. Valley V, Grace DM. Psychosocial risk factors in gastric
39. Olsson SA, Ryden O, Danielsson A, Nilsson-Ehle P. Weight surgery for obesity: identifying guidelines for screening. Int J
reduction after gastroplasty: the predictive value of surgical, Obes. 1987;11:105–13.
metabolic, and psychological variables. Int J Obes. 1984;8: 57. Davidson T, Rohde P, Wastell C. Psychological profile and
245–58. outcome in patients undergoing gastroplasty for morbid obe-
40. Ryden O, Olsson SA, Danielsson A, Nilsson-Ehle P. Weight sity. Obes Surg. 1991;1:177– 80.
loss after gastroplasty: psychological sequelae in relation to 58. Barrash J, Rodriguez EM, Scott DH, Mason EE, Sines JO.
clinical and metabolic observations. J Am Coll Nutr. 1989;8: The utility of MMPI subtypes for the prediction of weight loss
15–23. after bariatric surgery: Minnesota Multiphasic Personality In-
41. Sugerman HJ, Starkey JV, Birkenhauer R. A randomized ventory. Int J Obes. 1987;11:115–28.
prospective trial of gastric bypass versus vertical banded gas- 59. Hathaway SR, McKinley FC. Minnesota Multiphasic Per-
troplasty for morbid obesity and their effects on sweets versus sonality Inventory: Manual. New York: The Psychological
non-sweets eaters. Ann Surg. 1987;205:613–24. Corporation; 1967.
42. Cook CM, Edwards C. Success habits of long-term gastric 60. Lazare A, Klerman GL, Armor DJ. Oral, obsessive and
bypass patients. Obes Surg. 1999;9:80 –2. hysterical personality patterns. Arch Gen Psychiat. 1966;14:
43. Hudson SM, Dixon JB, O‘Brien PE. Sweet eating is not a 624 –30.
predictor of outcome after Lap-Band placement: can we fi- 61. Gough HG, Heilbrun AB. The Adjective Checklist Manual.
nally bury the myth? Obes Surg. 2002;12:789 –94. CA: Consulting Psychologists Press, Inc.; 1983.
44. Gentry K, Halverson JD, Heisler S. Psychologic assessment 62. American Psychiatric Association. Diagnostic and Statisti-
of morbidly obese patients undergoing gastric bypass: a com- cal Manual of Mental Disorders: Revision, DSM-III-R. 3rd ed.
parison of preoperative and post-operative adjustment. Sur- Washington, DC: American Psychiatric Association; 1987.
gery. 1984;95:215–20. 63. Gertler R, Ramsey-Stewart G. Pre-operative psychiatric as-
45. Hafner RJ, Watts JM, Rogers J. Quality of life after gastric sessment of patients presenting for gastric bariatric surgery
bypass for morbid obesity. Int J Obes. 1991;15:555– 60. (surgical control of morbid obesity). Aust N Z J Surg. 1986;
46. Schrader G, Stefanovic S, Gibbs A, Elmslie R, Higgins B, 56:157– 61.
Slavotinek A. Do psychosocial factors predict weight loss 64. Sugerman HJ, Londrey Gl, Kellum JM, et al. Weight loss
following gastric surgery for obesity? Aust N Z J Psychiatry. with vertical banded gastroplasty and Roux-Y gastric bypass
1990;24:496 –9. for morbid obesity with selective versus random assignment.
47. van Gemert WG, Severeijns RM, Greve JW, Groenman N, Am J Surg. 1989;157:93–102.
Soeters PB. Psychological functioning of morbidly obese 65. Karlsson J, Sjöström L, Sullivan M. Swedish obese subjects
patients after surgical treatment. Int J Obes Relat Metab (SOS): an intervention study of obesity: two-year follow-up of
Disord. 1998;22:393– 8. health-related quality of life (HRQL) and eating behavior after
48. Larsen F. Psychosocial function before and after gastric gastric surgery for severe obesity. Int J Obes Relat Metab
banding surgery for morbid obesity: a prospective psychiatric Disord. 1998;22:113–26.
study. Acta Psychiatr Scand Suppl. 1990;359:1–57. 66. Stunkard AJ, Messick S. The three-factor eating question-
49. Dubovsky SL, Haddenhorst A, Murphy J, Liechty RD, naire to measure dietary restraint, disinhibition and hunger.
Coyle DA. A preliminary study of the relationship between J Psychosom Res. 1985;29:71– 83.
preoperative depression and weight loss following surgery for 67. American Psychiatric Association. Diagnostic and Statisti-
morbid obesity. Int J Psychiatry Med. 1985;15:185–96. cal Manual of Mental Disorders: DSM-IV. 4th ed. Washing-
50. Kopec-Schrader EM, Gertler R, Ramsey-Stewart G, Beu- ton, DC: American Psychiatric Association; 1994.
mont PJ. Psychosocial outcome and long-term weight loss 68. Herpertz S, Kielmann R, Siffert W, Stang A, Jöckel K-H,
after gastric restrictive surgery for morbid obesity. Obes Surg. Senf W. Predictor variables of the course of weight: a multi-
1994;4:336 –9. center, prospective, controlled trial. Int J Obes. 2002;26:S74.
51. Larsen F, Torgersen S. Personality changes after gastric 69. Adami GF, Gandolfo P, Bauer B, Scopinaro N. Binge
banding surgery for morbid obesity: a prospective study. eating in massively obese patients undergoing bariatric sur-
J Psychosom Res. 1989;33:323–34. gery. Int J Eat Disord. 1995;17:45–50.

1568 OBESITY RESEARCH Vol. 12 No. 10 October 2004

Psychosocial Predictors of Obesity Surgery, Herpertz et al.

70. Kral JG. Surgical treatment of obesity. In: Bray GA, Bou- O‘Rahilly S. Clinical spectrum of obesity and mutations in
chard C, James WPT, eds. Handbook of Obesity. New York: the melanocortin 4 receptor gene. N Engl J Med. 2003;348:
Marcel Dekker; 1998, pp. 977–993. 1085–95.
71. Vallis MT, Ross MA. The role of psychological factors in 75. Blundell JE, Hill AJ. Serotonin, eating disorders and
bariatric surgery for morbid obesity: identification of psycho- the satiety cascade. In: Cassano GB, Akiskal HS, eds.
logical predictors of success. Obes Surg. 1993;3:346 –59. Serotonin-System-Related Syndromes (SRS): Psychopatho-
72. Branson R, Potoczna N, Kral JG, Lentes K-U, Hoehe MR, logical and Therapeutic Links. Royal Society of Medicine;
Horber FF. Binge eating as a major phenotype of melano- 1991.
cortin 4 receptor gene mutations. N Engl J Med. 2003;348: 76. Blundell JE, Hill AJ. The psychobiological control of appe-
1096 –103. tite. In: Ferrari E, Brambilla F, Solerte SB, eds. Primary and
73. Kral J. Selection of patients for anti-obesity surgery. Int J Secondary Eating Disorders: A Psychoneuroendocrine and
Obes. 2001;25:107–12. Metabolic Approach: Advances in the Biosciences 90. Perga-
74. Farooqi IS, Keogh JM, Yeo GS, Lank EJ, Cheetham T, mon Press; 1993.

OBESITY RESEARCH Vol. 12 No. 10 October 2004 1569