Journal of Consulting and Clinical Psychology
2014, Vol. 82, No. 6, 1207–1211
© 2014 American Psychological Association
0022-006X/14/$12.00 http://dx.doi.org/10.1037/a0037484
BRIEF REPORT
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Weight Suppression and Body Mass Index Interact to Predict Long-Term
Weight Outcomes in Adolescent-Onset Anorexia Nervosa
Ashley A. Witt and Staci A. Berkowitz
Christopher Gillberg
Drexel University
University of Gothenburg
Michael R. Lowe
Maria Råstam
Drexel University
Lund University
Elisabet Wentz
University of Gothenburg
Research on anorexia nervosa (AN) has emphasized the importance of low absolute body weight, but
emerging research suggests the importance of also considering low body weight relative to an individual’s highest premorbid weight (weight suppression; WS). Objective: We investigated whether body
mass index and WS at lowest weight (BMI-LW and WS-LW) among adolescents with AN predicted BMI
at 6-, 10-, or 18-year follow-up, duration of AN, or total eating disorder duration, including time during
which criteria were met for bulimia nervosa or eating disorder not otherwise specified. Method:
Forty-seven cases of AN identified through community screening in Sweden were included. Weight and
height data were collected from medical records, school nurse charts, and study follow-up assessments.
Results: Higher WS-LW was associated with higher BMI at 6-year and 10-year follow-up, and this effect
was strongest among those with the lowest BMI-LW values. BMI-LW and WS-LW were positively
associated with BMI at 18-year follow-up, but there was no significant interaction. There was no
significant association between WS-LW and AN duration or eating disorder duration, although eating
disorder duration was longer among those with higher BMI-LW, controlling for WS-LW. Conclusions:
Absolute and relative weight status interact to predict weight outcomes in AN over the long term. Results
suggest that BMI and WS may be more relevant to the prediction of long-term weight outcomes than to
the persistence of other eating disorder symptoms.
Keywords: anorexia nervosa, weight suppression, body mass index, outcome
individuals with bulimia nervosa (BN; Butryn, Juarascio, & Lowe,
2011; Herzog et al., 2010). Furthermore, WS has been shown to
predict longer time to remission from BN (Lowe et al., 2011) and
persistence of bulimic symptoms over 10-year follow-up (Keel &
Heatherton, 2010).
Only two published studies have examined the predictive significance of WS in anorexia nervosa (AN; Berner, Shaw, Witt, &
Lowe, 2013; Wildes & Marcus, 2012). Although all individuals
with AN are at a low body weight, there is considerable variability
in highest past weight (Miyasaka et al., 2003) and thus considerable variability in WS. Although traditionally the field has focused
on the significance of objectively low body weight in AN, often
measured by body mass index (BMI), emerging evidence indicates
that WS may also have predictive utility.
Individuals with AN who are higher in WS have been shown to
score higher on measures of eating disorder psychopathology,
binge eating, purging, and depression, even when controlling for
BMI (Berner et al., 2013). Two studies have investigated WS as a
predictor of response to treatment in AN and have found that WS,
controlling for initial BMI, predicts total weight gain, a faster rate
Weight suppression (WS), the difference between an individual’s current weight and highest past weight at adult height, may be
an indicator of illness severity and a prognostic indicator among
individuals with eating disorders. Weight loss produces metabolic
changes that promote weight regain (e.g., MacLean, Bergouignan,
Cornier, & Jackman, 2011); in line with these findings, higher WS
predicts greater weight gain and frequency of binge eating among
This article was published Online First July 21, 2014.
Ashley A. Witt and Staci A. Berkowitz, Department of Psychology,
Drexel University; Christopher Gillberg, Gillberg Neuropsychiatry Center,
University of Gothenburg; Michael R. Lowe, Department of Psychology,
Drexel University; Maria Råstam, Department of Clinical Sciences, Lund
University; Elisabet Wentz, Gillberg Neuropsychiatry Center, University
of Gothenburg.
Correspondence concerning this article should be addressed to Michael
R. Lowe, Department of Psychology, Drexel University, Stratton Hall
Suite 119, 3141 Chestnut Street, Philadelphia, PA 19104. E-mail: ml42@
drexel.edu
1207
WITT ET AL.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1208
of weight gain, and likelihood of endorsing binge eating or purging
at the end of treatment (Berner et al., 2013; Wildes & Marcus,
2012). In addition, Berner et al. (2013) found that BMI and WS at
admission interacted to predict weight gain during residential
treatment and psychological symptoms at discharge, suggesting
that the biological and psychological significance of WS may
depend on the severity of the resulting low BMI. Conversely, the
significance of a low BMI may depend on the degree of difference
from previous highest weight. While prior research suggests that
the prediction of weight gain by WS is likely to involve biological
mechanisms, Berner et al.’s results further suggest that psychological response to weight gain in treatment may be influenced by
weight history.
The present study investigated the long-term predictive effects
of WS and BMI, independently and in interaction. We examined
long-term weight outcomes as well as duration of illness (for AN
and other eating disorders). Data were analyzed from a prospective
study of individuals with adolescent-onset AN in Sweden who
were examined at intervals for 18 years after onset of AN. We
hypothesized that WS would be positively associated with weight
at follow-up, as well as with duration of illness. No a priori
predictions were made regarding main effects of BMI or interaction effects.
Method
Sample and Design
Data were used from a prospective study of 51 individuals with
adolescent-onset AN (48 females, three males). Half of this sample
(22 females, two males) consisted of all individuals born in 1970
in Göteborg, Sweden, who met criteria for AN by the age of 18
years and were living in the city in 1985 (except one individual
who declined to participate; for details, see Råstam, Gillberg, &
Garton, 1989). The other half of the sample included 26 females
and one male who had attended the same schools as the other
participants and who were born in years adjacent to 1970. Medical,
psychological, and neuropsychological characteristics and outcomes in the full sample have been reported in detail elsewhere
(e.g., Wentz, Gillberg, Anckarsater, Gillberg, & Råstam, 2009).1
Because of the small number of males in the sample and the
potential effects of sex and pubertal development on the variables
of interest, the present study included only female participants.
Participant diagnosis included physical examination and review
of growth charts kept by school nurses, with height and weight
data documented at regular intervals from first grade onward.
Participants also completed a structured clinical interview to assess
Diagnostic and Statistical Manual of Mental Disorders (3rd ed.,
rev.; DSM–III–R) criteria for AN; all participants met DSM–III–R
(and, on later review, DSM–IV) criteria for AN.
All 48 girls with AN participated in follow-ups 6, 10, and 18
years after AN onset (defined as first symptoms of restrictive
eating resulting in weight loss); mean age was 21, 24, and 32 years,
respectively, at follow-up points. At each time point, psychological
and physical measures were collected in person by research staff.
Three participants at 10-year follow-up and five at 18-year
follow-up completed assessments by phone. One person declined
to participate at 18-year follow-up, but some information was
obtained from a family member.
The current study used weight and height data from school
growth charts and medical records, as well as follow-up data on
BMI and duration of illness. The study was approved by the Drexel
University and Göteborg University ethical review committees.
Weight Variables During First Episode of Illness
The highest weight recorded in either school growth charts or
medical records, prior to any indications of weight loss associated
with AN, was taken as the participant’s highest premorbid weight
and was used along with height at that time to calculate highest
premorbid BMI (kilograms per meters squared). The lowest weight
recorded during the first episode of AN, in either school growth
charts or medical records, was used along with height at that time
to calculate BMI at lowest weight (BMI-LW). First episode of AN
included the period of time between onset of AN and the first time
criteria for AN were no longer met, either because the individual
met criteria for another eating disorder diagnosis or because criteria were not met for any eating disorder. BMI-LW was typically
reached within about one year following AN onset.
Weight suppression at lowest weight during the first episode of
AN (WS-LW) was calculated for each participant. In adults, WS is
typically calculated as the difference between current weight and
previous highest weight at adult height. Because some participants
had not reached adult height at the time of their premorbid highest
weight or the time of enrollment in the study, WS was calculated
in BMI units to account for changes in height. We used lowest
BMI rather than BMI at diagnosis to capture the greatest degree of
WS reached by each participant and to use a standard assessment
of low weight across participants. Accordingly, WS-LW was calculated as BMI at highest premorbid weight minus BMI-LW. The
data necessary to calculate WS-LW were available for 47 participants.
Long-Term Outcome
Height and weight were measured at each follow-up point.
These values were used to calculate BMI at 6-, 10-, and 18-year
follow-up. For the participants who completed follow-up assessments by phone, self-reported height and weight were used. Selfreported weight just prior to pregnancy was used for two individuals who were pregnant at the 10-year follow-up.
A study clinician assessed AN duration and eating disorder
duration as of 18-year follow-up. AN duration was defined as the
number of years an individual met full DSM–III–R criteria for AN
between AN onset and 18-year follow-up. Eating disorder duration was defined as the number of years an individual met full
DSM–III–R criteria for any eating disorder diagnosis (AN, BN, or
eating disorder not otherwise specified). Both the AN duration and
the eating disorder duration variables were calculated such that
partial years were taken into account (e.g., 1 year 3 months ⫽ 1.25
years). The coding of these variables included review of a compilation of data from multiple assessment modalities. A study
1
The original study included a comparison group consisting of 51
healthy individuals who were matched in age, sex, and school with participants in the AN group. Individuals in the comparison group were not
included in the current study because of the nature of the current research
question.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
WEIGHT SUPPRESSION IN ADOLESCENT ANOREXIA NERVOSA
clinician (M.R.) administered both the Structured Clinical Interview for DSM–III–R psychiatric disorders (SCID; Spitzer, Williams, Gibbon, & First, 1990) and the Morgan–Russell outcome
scale (a structured interview that assesses symptoms associated
with AN; Morgan & Hayward, 1988) to participants at follow-up
time points. Participants were interviewed about symptoms experienced between follow-up points, relapses, and hospital admissions. Information was also gathered from family members, doctors, and nurses, and hospital registries in Sweden were examined
until all individuals in the screened population had reached the age
of 19. The coding of these variables involved an extensive, multimodal assessment process, and the information gathered is
thought to provide a valid assessment of AN and eating disorder
duration. Data on AN duration and eating disorder duration have
been reported for this sample previously (Wentz et al., 2009).
Data Analytic Strategy
We conducted a series of linear regressions to examine the
associations of each outcome variable with WS-LW, BMI-LW,
and their interaction, controlling for age at lowest weight. For each
regression, the normality of the residuals was examined using
Shapiro–Wilk statistics, with values greater than or equal to .95
considered to indicate adequate normality given the sample size
(Royston, 1992). Natural log transformations were applied to BMI
at 10- and 18-year follow-up as well as AN duration to obtain
normal distributions of residuals. For each regression, age at
lowest weight, WS-LW, BMI-LW, and the interaction term were
entered simultaneously. Thus, results for each predictor variable,
including the interaction term, are from analyses that controlled for
all other predictors. When the interaction was nonsignificant,
regressions were rerun without the interaction term to examine
main effects of BMI-LW and WS-LW, controlling for age.
Results
Preliminary Analyses
Descriptive statistics for the sample are shown in Table 1.
Participants were typically in ninth or 10th grade at the initial
assessment by research staff, with a mean age of 16.2 years (SD ⫽
Table 1
Sample Characteristics
Variable
Minimum
Maximum
M
SD
Age of AN onset
Age at study screening
BMI at highest premorbid weight
Age at highest premorbid weight
BMI at lowest weight
Age at lowest weight
WS (in BMI units)
BMI at 6-year follow-up
BMI at 10-year follow-up
BMI at 18-year follow-up
AN duration (years)
ED duration (years)
10.0
11.1
15.5
9.8
9.5
10.5
2.5
13.2
15.4
13.6
0.9
0.9
17.2
19.1
30.9
16.6
18.7
18.7
14.5
31.2
38.0
37.7
14.7
18.9
14.3
16.2
21.1
13.9
15.0
15.3
6.0
21.2
22.0
22.1
3.6
7.6
1.6
1.4
3.3
1.5
2.3
1.7
2.6
3.1
3.9
4.4
2.6
4.4
Note. AN ⫽ anorexia nervosa; BMI ⫽ body mass index; ED ⫽ eating
disorder; WS ⫽ weight suppression. Ages are given in years.
1209
1.4). Pubertal status (coded using the Tanner stages of development scale ranging from 1 to 5, with higher numbers indicating
more advanced developmental stage; Marshall & Tanner, 1969)
was classified as follows: 8.3% as Stage 2, 10.4% as Stage 3,
25.0% as Stage 4, and 56.3% as Stage 5. During the initial
assessment, 25.0% of the sample reported binge eating or purging
behaviors (an additional 47.9% developed binge eating or purging
over the course of follow-up).
Results from regression models are shown in Table 2. Assessment of the assumptions of regression indicated that the assumptions of nonmulticollinearity, homoscedasticity, and normality of
error variance were met for each model.
Long-Term Weight Outcomes
There was a significant main effect of WS-LW and a significant
interaction between WS-LW and BMI-LW in the prediction of
BMI at 6-year follow-up, such that higher WS-LW was associated
with higher BMI at 6-year follow-up, and this effect was strongest
among those with the lowest BMI-LW (see Figure 1). Among
those with high BMI-LW relative to the rest of the sample, the
relation between WS-LW and BMI at 6-year follow-up was positive but weak. Those with low BMI-LW and low WS-LW relative
to the rest of the sample had the lowest BMIs at 6-year follow-up.
For BMI at 10-year follow-up, there were significant main effects
of BMI-LW and WS-LW, as well as a significant interaction (with
a similar pattern to the one found for BMI at 6-year follow-up). For
BMI at 18-year follow-up, there was a significant main effect of
BMI-LW and a trend toward a significant effect of WS-LW, but
there was no significant interaction.
Eating Disorder Duration
Neither BMI-LW, WS-LW, nor the interaction term significantly predicted AN duration. BMI-LW significantly predicted
total eating disorder duration, such that participants with higher
BMI-LW had a longer duration of illness, controlling for WS and
age at lowest weight.
Eating Disorder Treatment
We conducted exploratory analyses to determine whether variability in treatment received might account for the findings described above (i.e., if those higher in WS were more likely to
receive treatment, perhaps because of caregiver alarm at the discrepancy from highest past weight, and thereby more likely to
experience weight gain). As of 10-year follow-up, 21% of participants included in the present analyses had never received treatment for their eating disorder; 17% had never received treatment
as of 18-year follow-up. However, an independent samples t test
indicated that participants who received treatment (N ⫽ 39) did not
differ from participants who did not receive treatment (N ⫽ 8) on
WS-LW, t(45) ⫽ 0.01, p ⫽ .99, d ⫽ 0.01. Participants who
received no treatment, fewer than eight sessions of treatment, or
eight or more sessions were also compared using a univariate
analysis of variance, and no group differences were found on WS,
F(2, 44) ⫽ 0.52, p ⫽ .60, ␩p2 ⫽ .023. Receipt of treatment was
subsequently included as a dichotomous predictor variable in
regression models; results were unchanged.
WITT ET AL.
1210
Table 2
Independent and Interactive Predictive Effects of Weight Suppression (WS) and Body Mass Index
(BMI) at Lowest Weight
WS ⫻ BMI
WS
Variable
B
SE
p
B
SE
p
B
SE
p
R2
BMI 6-yr follow-up
BMI 10-yr follow-up
BMI 18-yr follow-up
AN duration (years)
ED duration (years)
0.29
0.63
0.76
⫺0.34
0.69
0.20
0.27
0.31
0.19
0.32
.159
.025
.018
.099
.038
0.91
0.75
0.46
⫺0.18
⫺0.20
0.20
0.27
0.25
0.15
0.26
⬍.001
.004
.064
.242
.450
⫺0.26
⫺0.36
⫺0.21
0.16
⫺0.09
0.11
0.15
0.18
0.10
0.18
.029
.033
.133
.619
.647
.34
.25
.15
.15
.13
Note. AN ⫽ anorexia nervosa; ED ⫽ eating disorder. For ease of interpretation, B and standard errors of B
values are reported in original units of the outcome variable rather than in log-transformed units. In cases of
nonsignificant interactions, the main effect data shown are from models that were rerun without the interaction
term. Age at lowest weight was controlled in all analyses.
Discussion
This study is the first to investigate the long-term predictive
significance of WS in AN, independently of and in interaction with
BMI. Although prior studies using this sample have found that
lowest BMI in absolute terms did not independently predict overall
functioning at follow-up (e.g., Wentz et al. 2009), we sought to
examine the significance of both absolute and relative weight
status with respect to illness duration and weight outcomes. To
account for the possibility that WS might be confounded with age
of onset, with those who developed AN at younger ages more
likely to have lower WS because they had achieved less premorbid
weight gain, we controlled for age in all analyses. WS-LW and
BMI-LW interacted to predict BMI at 6- and 10-year follow-up:
those with higher WS-LW had higher BMIs at follow-up, and this
effect was strongest among those with the lowest BMI-LW relative
to the rest of the sample. Those with both low WS-LW and low
BMI-LW tended to remain at very low BMIs at 6- and 10-year
follow-up (relative to the rest of the sample and in absolute terms).
There was no significant interaction between BMI-LW and
WS-LW in the prediction of BMI at 18-year follow-up. However,
higher BMI-LW predicted higher follow-up BMI, controlling for
30
Ave BMI (15.1)
28
BMI (kg/m2) at 6-year follow-up
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
BMI
+1 SD BMI (17.4)
26
-1 SD BMI (12.8)
24
22
20
18
16
14
12
-2
-1
0
1
2
3
Weight suppression (mean-centered)
4
5
Figure 1. Interaction of body mass index (BMI) and weight suppression
at lowest weight in the prediction of BMI at 6-year follow-up. Weight
suppression was calculated in BMI units. A similar pattern was found for
the outcome variable of BMI at 10-year follow-up. Ave ⫽ average.
WS-LW, and there was a trend toward a positive association
between WS-LW and BMI at 18-year follow-up, controlling for
BMI-LW.
These results suggest that when BMI at lowest weight is held
constant, individuals with AN with a history of higher premorbid
weight (and therefore higher WS) are likely to experience the
greatest weight regain in the long term. It is unclear whether this
is because of a predisposition toward higher weight or whether WS
produces biological or behavioral changes that promote greater
weight gain. However, the interaction effects at 6- and 10-year
follow-up suggest that WS is most important as a predictor of
weight gain when an especially low weight has been reached.
The present findings for weight outcomes are largely consistent
with those of two previous studies (Berner et al., 2013; Wildes &
Marcus, 2012) and suggest that the previously observed short-term
predictive effects of WS and BMI persist over the long term.
Interestingly, the poor weight outcomes among individuals with
low BMI-LW and low WS-LW parallel prior findings of poor
outcome on psychological symptoms (Berner et al., 2013). Berner
et al. (2013) suggested that those with low WS and low BMIs may
experience increased distress in response to weight gain because
they quickly reach or surpass their highest lifetime weight. It is
possible that increased resistance to weight gain due to elevated
shape and weight concerns might contribute to the poor long-term
weight outcomes among this group.
Contrary to hypotheses, WS-LW did not predict AN duration or
total eating disorder duration, independently or in interaction with
BMI-LW. However, controlling for WS-LW, we found that those
with higher BMI-LW met criteria for an eating disorder for a
longer time period. The mechanisms of this effect should be
investigated in future research. Taken together, our findings suggest that WS may be more relevant in the prediction of long-term
weight outcomes than in the prediction of long-term persistence of
other eating disorder symptoms. These results were unexpected
given previous findings that WS and BMI interact to predict
psychological symptom change during treatment (Berner et al.,
2013). Given the relatively low power in the current study, these
results should be replicated in larger samples. In addition, the
duration of illness variables did not include subclinical eating
disorder syndromes, which may have reduced the sensitivity of
these measures.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
WEIGHT SUPPRESSION IN ADOLESCENT ANOREXIA NERVOSA
Strengths of this study include the naturalistic, longitudinal
design; lengthy follow-up; comprehensive assessment procedures;
and use of measured weights rather than self-reported past weights.
In addition, because this sample was identified by a comprehensive
community screening procedure, selection bias is unlikely to have
affected the results. Limitations include the small sample size,
which resulted in limited statistical power for analyses of interaction effects, and the possibility that true highest and/or lowest
weights occurred in between recorded observations of body weight
in some instances. Although we calculated WS-LW and BMI-LW
on the basis of lowest weight during the first episode of AN to
examine the longitudinal significance of these early weight
changes, it is possible that some participants may have experienced
even lower weights during episodes of relapse, which may have
prognostic significance for the outcome variables studied. In addition, although weight in adolescents is commonly reported using
BMI percentiles, our weight-related variables are reported in BMI
units (controlling for age in all analyses) because of a floor effect
for BMI percentile at lowest weight; this may complicate the
comparison of the current findings with other reports of adolescent
weight. Finally, because a significant proportion of participants
never received treatment, it is possible that these results may not be
generalizable to treatment-seeking samples.
The present findings have several clinical and research implications. Future studies should investigate whether including WS in
the determination of treatment goal weights improves long-term
outcome in AN. In addition, the fact that individuals with the
combination of lower BMIs and low WS have shown poor outcome in two studies suggests that intensified treatment or different
treatment strategies may be required for these individuals. Further
research among larger samples is needed to clarify whether WS
and BMI predict the maintenance of eating disorder symptoms
over time among individuals with AN. The greater propensity
toward weight gain found in the present study among individuals
with high WS (particularly among those with low BMI-LW) might
be associated with maintenance of eating disorder symptoms over
the long term because of distress associated with weight gain, as is
thought to be the case in BN (Keel & Heatherton, 2010). In
addition, future longitudinal studies should examine whether WS
and BMI are associated with the emergence and maintenance of
binge eating among individuals with AN. Finally, although the
mechanisms of the effects of WS in adults are not fully understood,
changes in metabolic efficiency and hormonal functioning, in
combination with psychological variables, are thought to play a
role. Further research should investigate whether similar mechanisms of WS operate in adult-onset versus adolescent-onset eating
disorders, as early onset AN arrests normal development and often
has lasting effects on growth.
1211
References
Berner, L. A., Shaw, J. A., Witt, A. A., & Lowe, M. R. (2013). The relation
of weight suppression and body mass index to symptomatology and
treatment response in anorexia nervosa. Journal of Abnormal Psychology, 122, 694 –708. doi:10.1037/a0033930
Butryn, M. L., Juarascio, A., & Lowe, M. R. (2011). The relation of weight
suppression and BMI to bulimic symptoms. International Journal of
Eating Disorders, 44, 612– 617. doi:10.1002/eat.20881
Herzog, D. B., Thomas, J., Kass, A. E., Eddy, K. T., Franko, D. L., &
Lowe, M. R. (2010). Weight suppression predicts weight change over 5
years in bulimia nervosa. Psychiatry Research, 177, 330 –334. doi:
10.1016/j.psychres.2010.03.002
Keel, P. K., & Heatherton, T. F. (2010). Weight suppression predicts
maintenance and onset of bulimic syndromes at 10-year follow-up.
Journal of Abnormal Psychology, 119, 268 –275. doi:10.1037/a0019190
Lowe, M. R., Berner, L. A., Swanson, S. A., Clark, V. L., Eddy, K. T.,
Franko, D. L., . . . Herzog, D. B. (2011). Weight suppression predicts
time to remission from bulimia nervosa. Journal of Consulting and
Clinical Psychology, 79, 772–776. doi:10.1037/a0025714
MacLean, P. S., Bergouignan, A., Cornier, M.-A., & Jackman, M. R.
(2011). Biology’s response to dieting: The impetus for weight regain.
American Journal of Physiology: Regulatory, Integrative and Comparative Physiology, 301, R581–R600. doi:10.1152/ajpregu.00755.2010
Marshall, W. A., & Tanner, J. M. (1969). Variations in pattern of pubertal
changes in girls. Archives of Disease in Childhood, 44, 291–303. doi:
10.1136/adc.44.235.291
Miyasaka, N., Yoshiuchi, K., Yamanaka, G., Sasaki, T., Kumano, H., &
Kuboki, T. (2003). Relations among premorbid weight, referral weight,
and psychological test scores for patients with anorexia nervosa. Psychological Reports, 92, 67–74. doi:10.2466/pr0.2003.92.1.67
Morgan, H. G., & Hayward, A. (1988). Clinical assessment of anorexia
nervosa: The Morgan–Russell outcome assessment schedule. The British
Journal of Psychiatry, 152, 367–371. doi:10.1192/bjp.152.3.367
Råstam, M., Gillberg, C., & Garton, M. (1989). Anorexia nervosa in a
Swedish urban region: A population-based study. The British Journal of
Psychiatry, 155, 642– 646. doi:10.1192/bjp.155.5.642
Royston, P. (1992). Approximating the Shapiro-Wilk W-test for nonnormality. Statistics and Computing, 2, 117–119. doi:10.1007/
BF01891203
Spitzer, R. L., Williams, J. B. W., Gibbon, M., & First, M. B. (1990).
Structured Clinical Interview for DSM–III–R (Patient ed.). Washington,
DC: American Psychiatric Association.
Wentz, E., Gillberg, I., Anckarsater, H., Gillberg, C., & Råstam, M. (2009).
Adolescent-onset anorexia nervosa: 18-year outcome. The British Journal of Psychiatry, 194, 168 –174. doi:10.1192/bjp.bp.107.048686
Wildes, J. E., & Marcus, M. D. (2012). Weight suppression as a predictor
of weight gain and response to intensive behavioral treatment in patients
with anorexia nervosa. Behaviour Research and Therapy, 50, 266 –274.
doi:10.1016/j.brat.2012.02.006
Received September 16, 2013
Revision received April 8, 2014
Accepted June 16, 2014 䡲