Behavioral treatment of obesity in patients taking antipsychotic medications.

Abstract

Antipsychotic medications are associated with weight gain and metabolic dysregulation, yet little is known about the management of obesity among individuals with severe and persistent mental illness. Thus we sought to evaluate the potential utility of a behavioral weight control program for this population.
Outpatients receiving psychiatric care at a university medical center who had a body mass index (BMI; weight in kg/[height in m](2)) >or= 30 and were currently taking antipsychotic medication participated in a 12-week group behavioral weight control program. A medical chart review was conducted for each participant's body weight over the 10 months prior to beginning the program. A multiple baseline design was used to determine the impact of the intervention on BMI through 12-month posttreatment follow-up. We also assessed self-reported eating behavior, physical activity, and health-related quality of life. Data were collected from October 2000 to July 2003.
Among 35 patients who began the program, 29 (83%) completed treatment, with mean (+/- SD) weight loss of 5.04 (+/- 7.52) pounds (p = .001) and improvements in eating, activity, and quality of life. At 3-month posttreatment follow-up (N = 27; 77%), total mean weight loss was 7.14 (+/- 11.47) pounds (p = .003). Results of a longitudinal model based on general estimating equations indicated that, relative to the pretreatment period, BMI decreased significantly during treatment and remained stable through 12-month posttreatment follow-up.
Behavioral weight control is a promising approach to the treatment of obesity among outpatients taking antipsychotic medications, but longer and more robust interventions are needed.

Full text

Page 1

Behavioral Treatment of Obesity
J Clin Psychiatry 66:8, August 2005 10591058
oth first- and second-generation antipsychotic medi-
cations are associated with weight gain.1 Among the
second-generation drugs, clozapine has been associated
with the greatest weight gains, and ziprasidone the least.1,2
Consequently, there has been considerable focus on obe-
sity and metabolic dysregulation among patients taking
atypical antipsychotic medications.3
Little is known about the management of obesity among
individuals with severe and persistent mental illness taking
antipsychotic medications. Ball et al.4 evaluated the effec-
tiveness of the Weight Watchers 1-2-3 Program and super-
vised exercise sessions for patients with schizophrenia
who had olanzapine-related weight gain. Overall, partici-
pants in treatment (N = 11) decreased more in body weight
and body mass index (BMI; weight in kg/[height in m]2)
than a matched control group (N = 20), but the difference
between groups was not significant. However, there was
a sex-by-group-by-time interaction such that only men
(N = 7) experienced significant weight loss in treatment
(about 7 pounds in 10 weeks).
Menza and colleagues5 evaluated a behavioral weight
control program (Healthy Living) for individuals with
schizophrenia and schizoaffective disorder in day-treatment
programs who reported gaining weight while taking anti-
psychotic medications. Overweight and obese participants
in the yearlong study (N = 31) experienced significant re-
ductions in BMI and other medical risk factors, accompa-
nied by positive changes in nutrition and exercise. Patients
Behavioral Treatment of Obesity in Patients
Taking Antipsychotic Medications
Melissa A. Kalarchian, Ph.D.; Marsha D. Marcus, Ph.D.;
Michele D. Levine, Ph.D.; Gretchen L. Haas, Ph.D.;
Catherine G. Greeno, Ph.D.; Lisa A. Weissfeld, Ph.D.; and Li Qin, M.S.
Objective: Antipsychotic medications are as-
sociated with weight gain and metabolic dysregu-
lation, yet little is known about the management
of obesity among individuals with severe and
persistent mental illness. Thus we sought to
evaluate the potential utility of a behavioral
weight control program for this population.
Method: Outpatients receiving psychiatric
care at a university medical center who had a
body mass index (BMI; weight in kg/[height in
m]2) ≥ 30 and were currently taking antipsychotic
medication participated in a 12-week group be-
havioral weight control program. A medical chart
review was conducted for each participant’s body
weight over the 10 months prior to beginning the
program. A multiple baseline design was used to
determine the impact of the intervention on BMI
through 12-month posttreatment follow-up. We
also assessed self-reported eating behavior, phys-
ical activity, and health-related quality of life.
Data were collected from October 2000 to July
2003.
Results: Among 35 patients who began the
program, 29 (83%) completed treatment, with
mean (± SD) weight loss of 5.04 (± 7.52) pounds
(p = .001) and improvements in eating, activity,
and quality of life. At 3-month posttreatment
follow-up (N = 27; 77%), total mean weight loss
was 7.14 (± 11.47) pounds (p = .003). Results of
a longitudinal model based on general estimating
equations indicated that, relative to the pretreat-
ment period, BMI decreased significantly during
treatment and remained stable through 12-month
posttreatment follow-up.
Conclusion: Behavioral weight control is a
promising approach to the treatment of obesity
among outpatients taking antipsychotic medica-
tions, but longer and more robust interventions
are needed.
(J Clin Psychiatry 2005;66:1058–1063)
Received Nov. 1, 2004; accepted Jan. 10, 2005. From the Western
Psychiatric Institute and Clinic, University of Pittsburgh Medical Center,
Pittsburgh, Pa. (Drs. Kalarchian, Marcus, Levine, Haas, and Greeno);
Department of Biostatistics, Graduate School of Public Health,
University of Pittsburgh, Pittsburgh, Pa. (Dr. Weissfeld and Ms. Qin).
Research supported by the National Alliance for Research on
Schizophrenia and Depression (NARSAD) 2000 Young Investigator
Award, Essel Investigator and K23 DK62291 (principal investigator:
Melissa A. Kalarchian).
The authors report no other financial affiliations or other
relationships relevant to the subject of this article.
We gratefully acknowledge the contributions of nutritionist Lauren
Carlson, M.S., R.D., to the project.
Corresponding author and reprints: Melissa A. Kalarchian, Ph.D.,
Western Psychiatric Institute and Clinic, University of Pittsburgh Medical
Center, 3811 O’Hara Street, Pittsburgh, PA 15213
(e-mail: kalarchianma@msx.upmc.edu).
B

Page 2

Kalarchian et al.
10601059 J Clin Psychiatry 66:8, August 2005
receiving usual care in the same programs (N = 20) gained
weight over the course of the study.
In light of the limited information on the impact of
weight management in outpatients with severe and persis-
tent mental illness, we sought to evaluate the potential
utility of a behavioral weight control program for patients
taking antipsychotic medications. We chose to adapt the
Stoplight Diet,6 a program based on sound behavioral
principles that has been efficacious in producing weight
loss in children and adolescents, as well as their parents.7,8
We anticipated that this conceptually simple program
would be particularly useful in the treatment of individu-
als with schizophrenia and schizoaffective disorder. We
hypothesized that the modest caloric restriction and em-
phasis on regular physical activity would promote weight
loss and improvements in eating behavior, physical activ-
ity, and health-related quality of life.
METHOD
Design
The study employed a multiple baseline design. Spe-
cifically, we conducted a medical chart review for each
participant’s body weight over the 10-month period prior
to starting the treatment group. We documented weights
for 3 months (range, 2–4 months), 6 months (range, 5–7
months), and 9 months (range, 8–10 months) prior to the
start of treatment. This extended baseline allowed us to
evaluate whether the initiation of treatment was associ-
ated with a change in BMI trajectory.
The treatment program consisted of 12 weekly be-
havioral group sessions, followed by 2 biweekly booster
sessions and 2 monthly booster sessions, for a total of 16
sessions over 6 months. Research assessments were con-
ducted at pretreatment, posttreatment (after session 12),
and 3-, 6-, and 12-month posttreatment follow-up. The
study protocol was approved by the University of Pitts-
burgh Institutional Review Board.
Participants
Outpatients receiving care at Western Psychiatric Insti-
tute and Clinic, University of Pittsburgh Medical Center,
with a BMI of 30 or greater and currently taking antipsy-
chotic medication were eligible for the study. All partici-
pants were diagnosed with schizophrenia or related psy-
choses. Patients were recruited from a medication clinic
or referred by their doctor or mental health provider. Indi-
viduals who were mentally retarded, were participating in
another weight control program, or had evidence of se-
vere global cognitive deficits (Folstein Mini-Mental State
Examination9 score < 25) were excluded.
Sixty-one patients attended study information ses-
sions. After the principal investigator explained the study,
45 individuals signed written informed consent. Nine
patients dropped out of the study prior to attending the
first treatment session, most of whom reported that they
had enrolled in the study to collect the payment for com-
pleting the baseline assessment, but had not intended to
participate in the intervention. One additional patient was
excluded from the study because she was unsteady on her
feet and could not be weighed on our portable digital
scale. Thus, the final sample included 35 individuals who
consented and began treatment. Most patients (25; 71.4%)
were taking clozapine, and the rest were taking other anti-
psychotic medications.
Treatment
Treatment consisted of 12 weekly group sessions, fol-
lowed by 2 biweekly booster sessions, then 2 monthly
booster sessions. Sessions lasted 60 minutes and con-
sisted of a weight check, individualized review of self-
monitoring records, and a group presentation of the be-
havioral weight control program. A doctoral-level clinical
psychologist and a registered dietitian conducted the in-
tervention groups. Each session ended with a group walk
led by a study research assistant. All participants obtained
medical clearance from their physician to participate in
the research study intervention.
An eating plan was developed for use with this study
based on the Stoplight Diet.6 Food is classified according
to the colors of a stoplight: Green (Go! Include as many
as you can.), Yellow (Caution! Eat in limited amounts.),
and Red (Stop! Think before you eat.). Table 1 shows
examples of Red, Yellow, and Green foods. Participants
were instructed to increase their intake of fruits and veg-
etables (Green and Yellow foods), while decreasing the
number of foods high in fat and sugar (Red foods).
Participants also were directed to decrease sedentary ac-
tivities, especially time spent watching television. Partici-
pants were provided with pedometers and instructed to
gradually increase their walking toward a goal of 10,000
steps per day. Psychoeducation focused on the role of eat-
ing and activity in weight management, and the impor-
tance of medication compliance.
Standard behavioral techniques were used to help
participants implement and maintain the recommended
changes in eating and activity. Participants self-monitored
food type, amount, and “color” (Red, Yellow, or Green),
as well as pedometer steps and body weight. Self-
monitoring records were reviewed at each treatment ses-
sion in order to identify eating and activity patterns. Indi-
vidual feedback was provided with goals for behavior
change. Participants were awarded points for behavior
change that could be exchanged for gift certificates to
local stores. Additional behavioral strategies included the
use of stimulus control techniques to minimize cues for
eating and increase cues for activity, social assertiveness
training for situations involving eating and exercise, and
relapse prevention strategies to promote long-term main-
tenance of behavior change.

Page 3

Behavioral Treatment of Obesity
J Clin Psychiatry 66:8, August 2005 10611060
Assessments
Height was measured once at the start of the study.
Body weight was measured in street clothes without shoes
on a portable digital scale at each assessment. Research
assistants read self-report questionnaires to participants
individually at each assessment to ensure completion
and comprehension. Subjects were paid a total of $150
for completing all 5 assessments. Assessments were con-
ducted between October 2000 and July 2003.
Self-report questionnaires included the following: the
Eating Behavior Inventory (EBI),10 a self-report measure
of 26 specific behavioral strategies conducive to weight
loss with acceptable internal consistency and satisfactory
test-retest reliability; the Paffenbarger (Harvard Alumni
Activity Survey),11 a brief recall interview used to assess
physical activity, including walking, in the last week; and
the Medical Outcomes Study (MOS) Short-Form Health
Survey (SF-36),12 a 36-item self-report questionnaire used
to evaluate health-related quality of life. The MOS in-
cludes scales to assess physical functioning, role limita-
tions due to physical health, role limitations due to emo-
tional problems, energy/fatigue, emotional well-being,
social functioning, pain, and general health perceptions. It
has well-established reliability and validity, and has been
shown to be appropriate in diverse populations, including
mental health samples.13
Analytic Plan
Descriptive statistics were used to summarize demo-
graphic characteristics of study participants. t Tests (for
continuous variables) and χ2 analyses (for categorical
variables) were utilized to detect differences between
those who lost weight and gained weight during treat-
ment, as well as to compare those who dropped out to
those who completed treatment.
To determine whether treatment was associated with
improvements in body weight, eating, activity, and
health-related quality of life, we used paired t tests to
examine changes pretreatment to posttreatment. Linear
regression models, controlling for baseline BMI, were
used to probe whether changes in these factors were
related to decreases in BMI during treatment. Each
model was fit with change in BMI (pretreatment BMI–
posttreatment BMI) as the outcome variable. Each model
then included baseline BMI and the change in the covar-
iate of interest from pretreatment to posttreatment as pre-
dictor variables.
To test the hypothesis that the intervention would have
a positive impact on BMI trajectory, we fit a longitudinal
model using generalized estimating equations. We in-
cluded terms for time (each time point from the medical
chart review, weekly weights in treatment, booster ses-
sions, and follow-up assessments), period (pretreatment,
treatment, boosters, and posttreatment), and interaction
of time with period. Because patients taking clozapine,
the majority of the sample, did not differ in BMI trajec-
tory from those taking other medications, we grouped all
patients together for all analyses.
To evaluate potential factors related to BMI change
during treatment, further models were fit using general-
ized estimating equations. For these analyses, time was
limited to the weekly treatment sessions (first treatment
session attended—last treatment session attended), and
baseline BMI was included as a covariate. Selected other
covariates (medication [clozapine or other], number of
years taking antipsychotic medications, sex, age, race)
were then included in each model, one at a time. To as-
sess changes over time, each model included an interac-
tion term of time with each covariate. All generalized es-
timating equation models were fit using an exchangeable
Table 1. Examples of Foods Classified as Red, Yellow, and Green in a Study of Behavioral
Treatment of Obesity in Patients Taking Antipsychotic Medicationsa
Type of FoodRed
Fruits Peaches canned with sugar
Cranberry sauce
Vegetables French fries
Baked beans
Dairy Chocolate milk
Regular yogurt
Protein Chicken nuggets
Bacon
Grains Sweetened cereal
Prepackaged rice mix
Bread stuffing
Combination foods Egg rolls
Pizza
Fats and sweetsCandy bars
Butter
BeveragesRegular soda
Coffee with creamer
aFood classifications are adapted from the Stoplight Diet. For more information, see Epstein and Squires.6
Abbreviation: NA = not applicable.
YellowGreen
Orange juice Bananas
Apples
Carrots
Broccoli
NA
Yams
Pinto beans
Skim milk
Nonfat yogurt
Steamed fish
Hard-boiled eggs
Unsweetened oatmeal
Air-popped popcorn
Hamburger buns
Chicken noodle soup
Chili
Vanilla wafers
Pudding
Low-calorie cocoa
Apple juice
NA
NA
NA
Fat-free salad dressing
Butter spray
Diet soda
Unsweetened tea

Page 4

Kalarchian et al.
10621061 J Clin Psychiatry 66:8, August 2005
variance structure, and terms were retained or dropped
based on the statistical significance of each term.
RESULTS
Sample Characteristics
Participants were 65.7% (23/35) white and 54.3%
(19/35) male. Mean (± SD) age was 42.0 (± 10.3) years,
and participants reported having taken antipsychotic
medication for 17.1 (± 10.4) years. Mean BMI at study
entry was 36.5 (± 5.2). Most participants had pursued
some education beyond high school (22/30; 73.3%). Spe-
cifically, 7 patients completed college, 15 reported some
college or associate’s degree, 7 reported completing high
school or general equivalency diploma, and 1 patient re-
ported some high school. Education information was not
collected for the first 5 study participants.
Twenty-nine of 35 participants (82.9%) completed
weekly treatment, with a mean of 10.6 (± 1.7) sessions
attended. Mean weight loss during treatment was 5.04
(± 7.5) pounds (paired t = –3.61, p = .001). Twenty pa-
tients lost weight during treatment (range, 1.4–24.8
pounds), and 9 individuals gained weight (range, 0.2–9.4
pounds). No differences were identified between patients
who lost weight as compared to those who gained weight
during the 12 weekly sessions. Body weight continued to
decline during the period of booster sessions, with a mean
weight loss of 7.14 (± 11.47) pounds among 27 partici-
pants (77.1%) completing the 3-month posttreatment as-
sessment (paired t = –3.23, p = .003), corresponding to a
loss of 3.2% of initial body weight.
We compared individuals who completed weekly treat-
ment to those who dropped out with respect to demo-
graphic factors. Eighty percent (20/25) of those who com-
pleted treatment had pursued some education beyond high
school as compared to 40% of dropouts (2/5), a margin-
ally significant result (χ2= 3.4, p = .065). No other differ-
ences were identified between dropouts and completers.
Impact of Treatment on BMI Trajectory
With respect to BMI trajectory, the final model re-
tained a term for time, 3 terms for period (treatment,
boosters, and posttreatment, with pretreatment as the ref-
erence), and an interaction term between time and treat-
ment period. Findings from the model indicated that rela-
tive to the pretreatment period, BMI decreased during the
weekly treatment period (βtime*period= –0.04408, p = .003),
and remained stable throughout the posttreatment periods
(Figure 1). Estimates derived from a similar model for
body weight indicate that 9 months prior to the start
of treatment, mean body weight was 235.74 pounds; 6
months prior, it was 235.84; and 3 months prior, it was
235.94. During treatment, estimated mean body weight
dropped from 236.99 at the start to 232.22 at the end of
the 12-week intervention. At 3-, 6-, and 12-month post-
treatment follow-up, mean body weights were 230.55,
231.45, and 231.65 pounds, respectively.
After controlling for initial BMI, age (β = –0.023,
p = .006) and number of years taking antipsychotic medi-
cations (β = –0.022, p = .028) were associated with BMI
during weekly treatment, indicating that older patients and
those who had been taking medication for a greater num-
ber of years were significantly more obese. The interac-
tion of time with age (β = –0.0028, p = .058) and time
with number of years taking antipsychotics (β = –0.0035,
p = .052) approached significance, indicating a trend for
older patients and those who had taken antipsychotics for
a greater number of years to exhibit a more favorable BMI
trajectory during treatment. None of the remaining covar-
iates (medication, education, sex, and race), or the interac-
tions of these factors with time, were significant.
Changes in Eating, Activity,
and Health-Related Quality of Life
Participants increased their mean (± SD) total EBI
score by 13.42% (± 15.44) during weekly treatment, rep-
resenting a significant increase in behaviors associated
with behavioral weight management (paired t = –4.66,
p = .0001). Examination of the individual questionnaire
items indicated that 5 of 26 eating behaviors improved
significantly. The most strongly adopted behaviors were
recording the type and quantity of food eaten (paired
t = –7.34, p < .0001) and keeping a weight graph (paired
t = –4.60, p < .0001). Other behaviors that increased were
eating foods that one believes will aid one in losing weight
(paired t = –2.22, p = .036), keeping 1 or 2 raw vegetables
available for snacks (paired t = –2.50, p = .02), and stor-
ing food in containers where it is not readily visible or in
a closed container (paired t = –2.07, p = .05). Improve-
ments in 2 eating behaviors were significantly associated
with reductions in BMI during treatment after controlling
for BMI at study entry: keeping 1 or 2 raw vegetables
available for snacks (β = –0.3775, p = .045) and carefully
watching the quantity of food eaten (β = –0.3424, p =
.003). Participants’ self-reports of mean number of city
blocks walked per day went from 9.71 (± 10.13) at pre-
treatment to 16.75 (± 20.19) at posttreatment, although the
increase was not significant (paired t = –1.60, p = .124)
and not associated with change in BMI during weekly
treatment (β = –0.0138, p = .256).
With respect to health-related quality of life during
weekly treatment, 3 of 8 subscales of the SF-36 improved
significantly: physical functioning (paired t = –3.55, p =
.002), general health (paired t = –2.07, p = .049), and pain
(paired t = –2.56, p = .017). Patients also reported mar-
ginally significant increases in energy (paired t = –1.97,
p = .06) and social functioning (paired t = –1.77, p = .09).
Role limitations due to emotional problems, role limita-
tions due to physical health, and emotional well-being
did not change over the course of treatment. Only increase

Page 5

Behavioral Treatment of Obesity
J Clin Psychiatry 66:8, August 2005 1063 1062
in energy (β = –0.0353, p = .01) was significantly associ-
ated with BMI decrease during weekly treatment.
DISCUSSION
The association between antipsychotic medication and
weight gain is well established, yet little is known about
the management of obesity among patients taking anti-
psychotic medications. Although behavioral approaches
have proven effective in producing weight loss in the gen-
eral adult population of 1 to 2 pounds per week for up to
6 months,14 patients with severe and persistent mental ill-
ness are routinely excluded from controlled weight man-
agement treatment trials. This investigation is among the
first preliminary studies to document that behavioral
weight control is a feasible and promising approach to the
treatment of obesity in this population. Results indicate
that participants in a behavioral weight control program
adapted for outpatients taking antipsychotic medications
experienced significant reductions in body weight and
BMI, accompanied by improvements in eating, activity,
and health-related quality of life.
Relative to patients’ BMI trajectory prior to treatment,
participants’ BMI decreased significantly during weekly
treatment and remained stable at follow-up. However,
small increases in BMI were documented during the pre-
treatment and posttreatment intervals (see Figure 1), and
it is not unlikely that over time these trends would lead to
significant increases in BMI. That is, patients’ BMI may
have been on a modest upward trajectory prior to starting
the behavioral treatment program, and may have resumed
this trajectory after contact with the treatment program
ended. However, the observation that BMI continued to
decrease after the end of weekly treatment during the pe-
riod of booster sessions, corresponding to a maximum
weight loss of 7.14 pounds and 3.2% of initial body
weight at 3 months posttreatment, suggests that lengthen-
ing the duration of contact with the program could serve
to enhance weight loss and maintenance.
Although there is growing consensus that a sustained
decrease of 7% to 10% of initial body weight is a reason-
able goal associated with modest health benefits, current
clinical guidelines caution that a standard approach to
weight loss might work differently in diverse patient
populations.14 Findings from the present study are similar
to a quasi-experimental study with yearlong follow-up5 in
which 87% of severely mentally ill patients completed the
intensive 12-week portion of a behavioral weight loss
program, with weight loss of 6.6 pounds and improve-
ments in health parameters such as hemoglobin A1c and
blood pressure. Both the current study and the Menza et
al. study5 achieved somewhat more promising results than
a small study of patients taking olanzapine,4 in which only
men lost a significant amount of weight. Collectively,
these studies do suggest that behavioral weight control
can have positive effects not only on weight, but also on
medical, behavioral, and psychological parameters.
Clinical guidelines emphasize considering patient
characteristics in obesity treatment programs by adapting
the setting and staffing for the program, considering how
the obesity treatment program integrates into other as-
pects of patient health care and self-care, and expecting
and allowing for program modifications based on patient
responses and preferences.13 Our primary recruitment site
was a medication clinic for patients with severe and per-
sistent mental illness. Behavioral treatment groups were
held at the same location and scheduled to coordinate
with the days and times when patients taking clozapine
report for mandatory blood work. We worked with a nutri-
tionist and the participants in the initial treatment group
to simplify the eating plan. Modifications to treatment
format included streamlined self-monitoring forms and a
point system to reinforce changes in eating, physical ac-
tivity, and body weight. Furthermore, when working with
Figure 1. BMI Trajectory for GEE Analysis of Participants in Behavioral Treatment Program for
Patients Taking Antipsychotic Medications
Abbreviations: BMI = body mass index, GEE = generalized estimating equation.
Pretreatment TreatmentBoostersPosttreatment
–12–9–6 –30369 121518
36.6
36.4
36.2
36.0
35.8
35.6
35.4
Time (months)
36.31
36.34
36.36
36.49
35.73
35.47
35.61
35.67
BMI (kg/m2)