CE ACTIVITY
A 4-Year Prospective Study of Eating Disorder NOS
Compared with Full Eating Disorder Syndromes
W. Stewart Agras, MD1*
Scott Crow, MD2
James E. Mitchell, MD3
Katherine A. Halmi, MD4
Susan Bryson, MA, MS1
ABSTRACT
Objective: To examine the course of
Eating Disorder NOS (EDNOS) compared
with anorexia nervosa (AN), bulimia nervosa (BN), and binge eating disorder
(BED).
subtypes of EDNOS, in time to relapse following first remission. Only 18% of the
EDNOS group had never had or did not
develop another ED diagnosis during the
study; however, this group did not differ
from the remaining EDNOS group.
Method: Prospective study of 385 participants meeting DSM-IV criteria for AN,
BN, BED, and EDNOS at three sites.
Recruitment was from the community
and specialty clinics. Participants were
followed at 6-month intervals during a 4year period using the Eating Disorder Examination as the primary assessment.
Discussion: EDNOS appears to be a
way station between full ED syndromes
and recovery, and to a lesser extent from
recovery or EDNOS status to a full ED.
Implications for DSM-V are examined.
C 2009 by Wiley Periodicals, Inc.
V
Results: EDNOS remitted significantly
more quickly that AN or BN but not BED.
There were no differences between
EDNOS and full ED syndromes, or the
Introduction
Eating Disorder not Otherwise Specified (EDNOS)
the most common eating disorder1–4 is a residual
category composed of eating disorders that do not
meet full criteria for anorexia nervosa (AN) or bulimia nervosa (BN) and includes binge eating disorder (BED) as a provisional diagnosis.5 It has been
noted that EDNOS is similar to BN in the nature,
duration and severity of psychopathology1–4 and in
the use of health care services,6 yet it has been relatively neglected in the research literature. For
example, few studies have prospectively examined
Accepted 17 April 2009
Supported in part by grants from the McKnight Foundation to
Cornell University, the University of Minnesota, and Stanford
University and by P30 DK 50456 and K02 MH 65919.
*Correspondence to: W. Stewart Agras, Department of Psychiatry,
Stanford University School of Medicine, Stanford, California.
E-mail: [email protected]
1
Department of Psychiatry, Stanford University School of
Medicine, Stanford, California
2
Department of Psychiatry, University of Minnesota School of
Medicine, Minneapolis, Minnesota
3
Neuropsychiatric Research Institute and Department of Clinical
Neuroscience, University of North Dakota School of Medicine and
Health Services, Fargo, North Dakota
4
Department of Psychiatry, Weill Medical College of Cornell University, White Plains, New York
Published online 19 June 2009 in Wiley InterScience
(www.interscience.wiley.com). DOI: 10.1002/eat.20708
C 2009 Wiley Periodicals, Inc.
V
International Journal of Eating Disorders 42:6 565–570 2009
Keywords: EDNOS; anorexia nervosa;
bulimia nervosa; binge eating disorder;
prospective study; remission; relapse
(Int J Eat Disord 2009; 42:565–570)
the course of EDNOS compared with other eating
disorders such as AN and BN. Studies comparing
the course of different disorders may be helpful in
distinguishing between different entities.
One comparative study with 90 participants
found that 78% of an EDNOS group were recovered
at 5-year follow-up with 20% having another eating
disorder diagnosis. Comparable figures for AN were
59% and 14%, and for BN 74% and 18%.7 A larger
study found that a higher percentage of EDNOS
remitted compared with BN at 24 months followup.8 Another study followed 33 women with EDNOS
for an average of 41 months.9 During this time 46%
met criteria for a full eating disorder syndrome for
at least some period of time. At the end of follow-up
12% of the EDNOS group met full criteria for either
AN or BN and only 18% had recovered, suggesting
that the EDNOS group was moving both toward
recovery and toward a full ED syndrome.
Although these studies suggest that EDNOS
recovers more quickly than AN or BN, two other
studies10,11 found no differences in terms of remission between EDNOS and BN.
The present study expands previous work using a
larger sample of EDNOS compared with full
syndrome AN, BN, and BED with assessments
every 6 months over a 4-year period. Although BED
is at present classified within the EDNOS category
studies suggest that it should be regarded as a sepa565
AGRAS ET AL.
rate entity. For example, family studies suggest that
BED is transmitted separately from AN and BN,
that it increases the risk for obesity, and that it is a
chronic and stable syndrome.12–14 Based on the literature we hypothesized that the EDNOS group
would demonstrate faster rates of recovery and
relapse over time as compared with full syndrome
eating disorders including BED. Because EDNOS
appears to be composed of a heterogeneous group
of disorders, we comprised an EDNOS group for
this study consisting of partial AN, BN, and BED
entities that comprise the majority of the EDNOS
group.15 This allowed us to consider the partial syndromes together (as EDNOS) and separately.
TABLE 1.
Diagnostic criteria for the partial syndromes
Partial anorexia nervosa
Meets all SCID criteria for AN for the past 6 months except weight
!90% of ideal body weight together with either (a) amenorrhea, or
(b) a cognitive pattern typical of AN (body image disturbance and
intense fear of fat)
Met full criteria for AN in the past 12 months
Partial bulimia nervosa
Meets all SCID criteria for BN except the shape and weight criterion
Meets all SCID criteria for BN except frequency of binge eating and
purging is below the SCID threshold (but at least once/month on
average for the last 6 months
Partial binge eating disorder
Meets all SCID criteria for BED except for frequency of binge eating
below threshold, but at least 1 day per month on average for the
last 6 months
Method
the course of the study. Inter-rater reliabilities were purging 0.99; objective binge eating 0.99; eating concerns
0.98; restrained eating 0.99; shape concerns 0.97; and
weight concerns 0.90.
Participants
Statistical Analyses
This study was conducted at three sites: Cornell University, Stanford University, and the University of Minnesota. Details of recruitment to this study were described
in a previous publication.16 The Institutional Review
Board at each site approved the study and all subjects
gave their written informed consent to participate.
The primary data for the analysis were derived from
the semiannual EDE interviews. If a data point was missing it was acquired from the M-FED preceding the missing data point. If the M-FED was not available, the data
point was interpolated according to the following rules. If
the two points on either side of the missing data point
were the same value, that value was entered. However, if
the data points differed, a random choice was made
between the two options. Thirty-two data points were
estimated in this way. The stability of diagnosis used
transitional probabilities of Markov chains to determine
the probability of a specific ED diagnosis given the
diagnosis that occurred 6 months earlier. These eight
transitional matrices were averaged to determine the
mean 6-month transition between diagnoses. Betweengroup statistical comparisons used ANOVA or chi-square
as appropriate taking into account the effect of site.
Survival curves were used to compare the time to first
episode of no ED diagnosis and time to first relapse after
no diagnosis for the different eating disorder types.
A total of 385 individuals were entered into the study
and followed for 4 years. Participants met DSM-IV criteria for AN (N 5 45), BN (purging type) (N 5 87), BED (N
5 104), or for EDNOS (N 5 149) as noted above and
defined in Table 1 as partial AN (PAN), partial BN (PBN),
or partial BED (PBED).
Assessments
Participants were followed for 4 years with in-person
interviews every 6 months and shorter telephone interviews 3 months after each in-person interview. At the
in-person assessments participants were weighed (in
gowns if AN) and their height was measured. Interviews
consisted of the Eating Disorder Examination (EDE)17 to
assess core eating disorder psychopathology and an interim telephone interview with the McKnight Follow-up
for Eating Disorders (M-FED). The M-FED was based on
the LIFE interview18 modified to assess eating disorder
psychopathology with questions and definitions similar
to those of the EDE. Hence this interview documented
eating disorder symptoms and psychopathology in an
abbreviated form; treatment for eating disorder or other
conditions; and general psychopathology. The Structured
Clinical Interview for DSM-IV for Axis I and II disorders19
was administered at entry to the study. Assessors were
trained in the administration of the EDE and M-FED at
combined site meetings and were supervised on site. A
random sample of EDEs at each of the four sites was audited by independent assessors periodically throughout
566
Remission was defined in this study as the first point
after entry to the trial that the participant had no ED
diagnosis during the preceding 6 months as determined
by the EDE using DSM-1V criteria. Relapse was defined
as the first point after a remission when symptoms
for any ED diagnosis including EDNOS (as defined in
Table 1) were observed as determined by the EDE.
Results
Selected baseline characteristics for the four
diagnoses are shown in Table 2. A more detailed
description can be found in a previous
publication.16 The majority of participants in all
International Journal of Eating Disorders 42:6 565–570 2009
EDNOS VERSUS AN, BN, AND BED
TABLE 2.
Baseline characteristics for the four eating disorder groups
Age (yr)
BMI
Employed (%)
Current depression (%)
Lifetime depression (%)
Personality disorder (%)
Substance use (%)
Outpatient treatment (%)
Inpatient treatment (%)
Objective bingesa
Purgesa
Self-induced vomiting
Laxative use
EDE restraint
EDE weight concerns
EDE shape concerns
EDE eating concerns
a
AN (N 5 45)
BN (N 5 87)
BED (N 5 104)
EDNOS (N 5 149)
27.3 (10.0)
16.9 (2.0)
53
31
56
49
24
73
75
0 (0.75)
0 (14.8)
0 (1)
0 (0)
3.0 (2.0)
3.2 (1.8)
3.4 (1.9)
2.7 (1.7)
30.1 (8.4)
24.9 (9.4)
62
25
69
47
29
79
34
21 (18)
29 (46)
26.5 (46)
0 (1)
3.3 (1.6)
3.6 (1.5)
3.8 (1.4)
2.7 (1.4)
38.7 (7.2)
37.4 (8.6)
71
12
67
24
28
68
5
14 (11)
0 (0)
0 (0)
0 (2)
2.2 (1.3)
3.6 (1.1)
3.9 (1.1)
2.1 (1.3)
31.5 (8.8)
27.0
79
16
63
32
27
84
24
2 (5)
0 (5.75)
0 (0)
0 (0)
2.4 (1.7)
2.9 (1.6)
3.2 (1.5)
1.8 (1.4)
Median and interquartile range.
groups had outpatient treatment for their eating
disorder whereas inpatient treatment was most
common for AN followed by BN.
relapse following first remission for the three subtypes of EDNOS. There were no significant differences among groups and there was a high rate of
Times to Remission and Relapse for EDNOS
Compared with Full Syndromes
FIGURE 1. (a) Survival curve showing time to first remission for EDNOS and the three full syndrome ED groups. (b)
Survival curve showing time to relapse following remission for EDNOS and the three full syndrome ED groups.
As shown in Figure 1 the time to first remission,
i.e. having no ED diagnosis, differs among diagnoses W(3)517.8, p 5 .000 with EDNOS and BED
showing similar rates over the course of the study
and having the shortest times to remission, and BN
having the longest time to first remission followed
by AN. At the 4-year follow-up 78% of the EDNOS
group were remitted compared with 82% of the
BED group, 47% of the BN group, and 57% of the
AN group. Time to relapse following the first remission is also shown in Figure 1. Although the survival analysis showed no significant difference
between groups, inspection of the curves suggests
that BED, BN, and EDNOS had similar curves but
that AN participants took longer to relapse after the
first remission. Of note is the high rate of relapse in
the first 6 months after remission for all groups
except AN.
Times to Remission and Relapse for
EDNOS Subtypes
We next compared the subtypes comprising the
EDNOS group, PAN, PBN, and PBED, to determine
differences among them in times to first remission
and relapse. As shown in Figure 2 there were no
significant differences among the subtypes in time
to first remission. At the 4-year follow-up 74% of
the PBED group was remitted, compared with 70%
of the PBN group, and 68% of the PAN group.
Figure 2 also displays the survival curve for time to
International Journal of Eating Disorders 42:6 565–570 2009
567
AGRAS ET AL.
FIGURE 2. (a) Survival curve showing time to first remission for the three subcomponents of EDNOS. (b) Survival
curve showing time to relapse following remission for the
three subcomponents of EDNOS.
FIGURE 3.
diagnoses.
Average
6-month
transitions
between
is noteworthy that there are no transitions more
than 5% between the AN, BN, and BED groups, and
only a 7% average transition between EDNOS and
BED and none between EDNOS and the other categories. There are, however, some transitions below
the 5% level. AN transitions directly to BN in 2% of
cases and to no ED diagnosis in 4%. BED transitions to BN in 1% of cases.
Retrospective Study of the EDNOS Group
relapse for all three groups in the first 6 months following remission.
Stability of Diagnosis
Figure 3 shows the Markov probabilities for the
average 6-month diagnostic transitions less than
5% between the four diagnoses on a person basis.
The four diagnoses (AN, BN, BED, and EDNOS)
have similar levels of stability in terms of moving
either between diagnostic categories or to no ED
diagnosis. The most common pathway to remission
is via the EDNOS category with relatively few cases
going straight to remission. Similarly the most
frequent pathway to relapse is to the EDNOS
diagnosis. Hence, EDNOS appears to be largely
composed of individuals most frequently on the
path to remission and less frequently to relapse
because the overall group is improving over time. It
568
A review of past ED diagnoses for the EDNOS
group determined by the SCID reveals that 78% of
the EDNOS group had a past full ED diagnosis.
Hence, only a minority of the group (N 5 37) was
‘‘pure’’ EDNOS at entry to the study. During the
duration of the study 10 (27%) participants of this
group developed either AN nor BN, 5 (14%) continued as EDNOS, and 22 (59%) recovered without
developing another ED diagnosis. Hence, of the 149
individuals entered to the study with a diagnosis of
EDNOS only 27 (18%) finished the study with no
other ED diagnosis. This group did not differ in
time to remission or relapse from the EDNOS
group who had met criteria for a full ED syndrome.
There were also no significant differences between
the two groups on any baseline measure except ED
symptoms.
Discussion
The primary aim of this study was to compare the
course and stability of EDNOS compared with AN,
International Journal of Eating Disorders 42:6 565–570 2009
EDNOS VERSUS AN, BN, AND BED
BN, and BED. As a recent review of the literature
pointed out ‘‘virtually nothing is known about the
persistence of EDNOS.’’20 Yet, as many authors
have noted EDNOS comprises the largest ED group,
has similar psychopathology to other eating disorders particularly BN, and is accompanied by considerable disability.1,3–5 In this study we constructed an EDNOS group from the partial syndromes of AN, BN, and BED, hence the
subcomponents of EDNOS as defined in Table 1
could be examined individually.
The first question posed concerned the course of
EDNOS. Hence we examined time to first remission
from entry to the study. There was a significant difference in time to first remission among the four
syndromes with EDNOS and BED showing the
shortest and almost identical times, and BN showing the longest time followed by AN. Nearly 80% of
the EDNOS group had no ED at the 4-year followup with a similar figure for BED. Next, time from
first remission to relapse to any ED diagnosis was
examined. No significant differences were found
among the four groups in this analysis. Overall,
these findings are similar to those of previous
studies, i.e. EDNOS remits more quickly and has
a higher rate of recovery over time than AN or
BN.7,8 However, in the present study no differences in course were found between EDNOS and
BED. As found in other studies the majority of
cases from all the ED syndromes no longer have
their original diagnosis at the end of the 4-year
follow-up.11,21,22
We next examined the time to remission and to
first relapse following remission for the three
subcomponents (PAN, PBN, PBED) of EDNOS.
There were no statistically significant differences
among these subcomponents for either time to
remission or time to first relapse, suggesting identical courses.
With the exception of EDNOS as shown in Figure
3, there are few transitions from one eating disorder to another with most transitions occurring
between AN and BN. These findings suggest that
EDNOS is mainly composed of individuals with an
ED diagnosis transitioning to no diagnosis or from
no diagnosis to an eating disorder diagnosis. It is
noteworthy that few individuals transition directly
on average over a 6-month time interval from a full
eating disorder diagnosis to no diagnosis. A retrospective study of the EDNOS group found that the
majority had a full ED in the past. This is consistent
with the prospective finding that patients with a
full ED transition to remission via EDNOS. In
addition only 18% of the EDNOS group without a
International Journal of Eating Disorders 42:6 565–570 2009
full ED diagnosis recovered without developing
another ED or persisted as EDNOS. However, this
‘‘pure’’ EDNOS group did not differ from the
remaining EDNOS group in terms of times to remission or relapse or on any baseline characteristic.
Despite the strengths of this study, namely its
prospective nature, frequent assessment by standard interview, and fairly large sample size, there are
some limitations that need to be taken into
account. The majority of participants received
some form of treatment during the follow-up
period specifically directed at their eating disorder.
Although the proportions of individuals treated as
outpatients were similar for the four groups, hospitalization was more frequent for AN than for the
other groups possibly differentially affecting the
course of this syndrome. It is probable that without
specific ED treatment the outcome of these disorders, particularly for AN, would have been worse.
In addition, the number of participants with AN is
relatively small. However, given the frequency of
some form of treatment for eating disorders found
in this and other studies6 it is unlikely that a prospective study would be able to follow an untreated
cohort over time. Finally, the EDNOS group was
composed of partial cases of AN, BN, and BED
omitting other syndromes that now populate
EDNOS such as night eating syndrome and purging
syndrome.4 However, the majority of EDNOS cases
appear to be subclinical variants of the three
main EDs.4
In conclusion, the results of this study suggest
that EDNOS is a way-station for those moving from
a full ED or from remission to another ED. The retrospective examination of the EDNOS group found
that the majority had a full ED diagnosis in the past
and that there are few true EDNOS cases. It is likely
that the composition of EDNOS will vary with the
age of the cohort. For example, in adolescence it
may be that a higher proportion of cases would be
on the path toward a full ED syndrome.
These results may have implications for DSM-V
and for the problematic diagnostic category of
EDNOS.2 Given that EDNOS is largely a transient
category it may make sense to use that diagnosis
only for cases that have never met criteria for a full
ED. For example, a patient with AN who transitions
to another ED diagnosis including EDNOS should
continue to be diagnosed as having AN. This would
remove the majority of diagnoses from the EDNOS
category to a full ED syndrome leaving only those
who have not yet met lifetime criteria for a full ED
diagnosis and those with as yet undefined syndromes as EDNOS cases.
569
AGRAS ET AL.
Earn CE credit for this article!
Visit: http://www.ce-credit.com for additional information. There may be a delay in the posting of the article, so
continue to check back and look for the section on Eating
Disorders. Additional information about the program is
available at www.aedweb.org
11.
12.
References
1. Fairburn CC, Cooper Z, Bohn K, O’Connor M, Doll HA, Palmer
RL. The severity and status of eating disorder NOS: Implications
for DSM-V. Behav Res Ther 2007;45:1705–1715.
2. Fairburn CG, Bohn K. Eating disorder NOS (EDNOS): An example
of the troublesome ‘‘not otherwise specified’’ (NOS) category in
DSM-IV. Behav Res Ther 2005;43:691–701.
3. Hay PJ, Mond J, Buttner P, Darby A. Eating disorder behaviors
are increasing: Findings from two sequential community
surveys in Australia. PLoS ONE 2008;6:e1541.
4. Rockert W, Kaplan AS, Olmsted MP. Eating disorder not otherwise specified: The view from a tertiary care treatment center.
Int J Eat Disord 2007;40:S99–S103.
5. American Psychiatric Association. Diagnostic and Statistical
Manual of Mental Disorders, 4th ed, Text Revision. Washington
DC, American Psychiatric Association, 2000.
6. Striegel-Moore RH, Debar L, Wilson GT, Dickerson J, Rosselli F,
Petrin N, et al. Health services use in eating disorders. Psychol
Med 2008;38:1465–1474.
7. Ben-Tovim D, Kay W, Gilchrist P, Freeman R, Kalucy R,
Esterman A. Outcome in patients with eating disorders: A
5-year study. Lancet 2001;357:1254–1257.
8. Grilo CM, Sanislow CA, Shea MT, Skodol AE, Stout RL, Pagano
ME, et al. The natural course of bulimia nervosa and eating disorder not otherwise specified is not influenced by personality
disorders. Int J Eat Disord 2003;34:319–330.
9. Herzog DB, Hopkins JD, Burns CD. A follow-up study of 33
subdiagnostic eating disordered women. Int J Eat Disord 1993;
14:261–267.
10. Grilo CM, Pagano ME, Skodol AE, Sanislow CA, McGlashan TH,
Gunderson JG, et al. Natural course of bulimia nervosa and
eating disorder not otherwise specified: 5-year prospective
570
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
study of remissions, relapses, and the effects of personality
disorder psychopathology. J Clin Psychiatry 2007;68:738–
746.
Fichter MM, Quadflieg N, Hedlund S. Long-term course of
binge eating disorder and bulimia nervosa: Relevance for
nosology and diagnostic criteria. Int J Eat Disord 2008;41:
577–586.
Lilenfeld LR, Ringham R, Kalarchian MA, Marcus MD. A family
history study of binge eating disorder. Compr Psychiatry
2008;49:247–254.
Pope HG, Lalonde JK, Pindyck LJ, Walsh BT, Bulik CM, Crow SJ,
et al. Binge eating disorder: A stable syndrome. Am J Psychiatry
2006;163:2181–2183.
Hudson JI, Lalonde JK, Berry JM, Pindyck LJ, Bulik CM, Crow SJ,
et al. Binge eating as a distinct familial phenotype in obese
individuals. Arch Gen Psychiatry 2006;63:313–319.
Mitchell JE, Crosby RD, Wonderlich SA, Hill L, le Grange D,
Powers P, et al. Latent profile analysis of a cohort of patients
with eating disorder not otherwise specified. Int J Eat Disord
2007;40(Suppl):S104–S106.
Crow SJ, Agras WS, Halmi K, Mitchell JE, Kraemer HC. Full
syndromal versus subthreshold anorexia nervosa, bulimia
nervosa, and binge eating disorder: A multicenter study. Int J
Eat Disord 2002;32:309–318.
Cooper Z, Cooper PJ, Fairburn CG. The validity of the eating disorder examination and its subscales. Br J Psychiatry 1989;154:
807–812.
Keller MB, Lavori PW, Friedman B. The life: A comprehensive
method for assessing outcome in prospective longitudinal studies. Arch Gen Psychiatry 1987;44:540–548.
First MB, Spitzer RL, Gibbon M, Williams JBW. Structured
Clinical Interview for DSM-IV Axis I Disorders: Patient Edition
(SCID-I/P Version 2.0). New York: Biometrics Research Department, New York State Psychiatric Institute, 1995.
Berkman ND, Lohr KN, Bulik CM. Outcomes of eating disorders:
A systematic review of the literature. Int J Eat Disord 2007;
40:293–309.
Fairburn CG, Cooper Z, Doll HA, Norman P, O’Connor M. The
natural course of bulimia nervosa and binge eating disorder in
young women. Arch Gen Psychiatry 2000;57:659–665.
Herzog DB, Dorer DJ, Keel PK, Selwyn SE, Sherrie E, Ekeblad ER,
et al. Recovery and relapse in anorexia and bulimia nervosa.
J Am Acad Child Adolesc Psychiatry 1999;38:829–837.
International Journal of Eating Disorders 42:6 565–570 2009